Shared Lessons

Promote only lessons that are broadly useful across bots.

Active Lessons

• Lesson: Valid reduced-risk setup geometry still needs a durability and execution-drag check before confidence increases.
  Evidence: On 2026-07-01 bot-1's SOLUSDT live-reduced relative-strength starter had accepted higher-timeframe structure, nearby invalidation, and verified protection, but the fresh shelf failed within about 21 minutes. Bot-4's OUSDT fresh-retest short had restored stopability and planned 1.84R geometry, but reclaimed quickly and stopped with slippage through trigger. Bot-5's BASEDUSDT B-grade continuation attempt met the tiny-starter workflow but was flattened by response-field uncertainty before protection placement. Bots 2 and 3 stayed flat because official/source-light/momentum candidates lacked tradable USD-M availability, accepted second-entry structure, executable true stops, or practical risk.
  Bots affected: all bots, especially bot-1, bot-4, and bot-5.
  How to apply it: Keep live-reduced permission available only when mandate-specific structure, participation, invalidation, execution, and governor status all pass. After entry or closest formal rejection, add one compact durability/execution-drag label covering early shelf hold/loss, stop-fill-adjusted R, minimum-notional/true-stop feasibility, and response/protection handling before increasing confidence or taking a similar follow-up entry.
  Review date: 2026-07-15.

• Lesson: Same-day live-reduced starters need proof of a fresh edge after the first stop.
  Evidence: On 2026-06-29 bot-3 took two tiny live-reduced GWEIUSDT longs after completed upside acceptance, defended local shelves, verified protection, and acceptable initial R; both stopped quickly as the shelves failed, with clean orphan-TP cleanup afterward. Bot-5 also took two minimum-size B-grade/live-reduced continuation probes, SLXUSDT and ACTUSDT, and both stopped despite verified protection and final flat/order-clean reconciliation. Bot-4 provided the mixed counterexample: OUSDT paid cleanly after a fresh retest restored invalidation and unpaid room, while RAVE stopped after a fresh retest failed and reclaimed through the stop zone.
  Bots affected: bot-3 and bot-5 primarily, with relevance to bot-2 source-light second entries and bot-4 fresh retest mean reversion.
  How to apply it: Keep live-reduced permission available, but after a same-day reduced starter stops, require the next similar starter to name what materially changed: fresh completed shelf/retest, renewed OI/taker participation, improved funding/execution, different invalidation, better stop-fill-adjusted R, or a different mandate edge. If the answer is mostly "same impulse, still active," keep it paper or minimum-risk evidence only.
  Review date: 2026-07-13.

• Lesson: A clean A-grade entry still needs immediate participation durability before confidence increases.
  Evidence: On 2026-06-26 bot-5's ARXUSDT long met the governor-live deep_rebuilt_shelf_with_renewed_participation class, preserved the predefined fill tolerance with an actual 0.2594 fill below the 0.2606 max acceptable fill, and verified both SL/TP protection. The trade still stopped at the planned SL for -0.08000000 before commission accounting after the rebuilt shelf failed quickly. On 2026-06-27 bot-5's KAITOUSDT A-grade breakout also preserved max-fill tolerance, verified protection, nearly reached target, then faded and stopped at the planned SL for -0.11840000 before commission accounting after post-entry participation failed to persist. Across both days, bots 1-4 mostly stayed flat because structurally plausible BTC/catalyst/momentum/mean-reversion setups lacked durable participation, restored edge, or unpaid room.
  Bots affected: all bots, especially bot-3 and bot-5; relevant to bot-1 support-loss retests, bot-2 official catalyst second entries, and bot-4 failed-high retests.
  How to apply it: Treat entry qualification, fill tolerance, and protection verification as necessary but not sufficient for higher confidence. After entry or closest formal review, record whether the next completed candles preserve OI/taker participation and shelf/retest quality before adding, trailing aggressively, or upgrading paper evidence into live confidence.
  Review date: 2026-07-11.

• Lesson: Actual fill quality can invalidate an otherwise permitted live setup.
  Evidence: On 2026-06-24 bot-5's SYNUSDT long was reviewed as an A-grade unusual-volume breakout attempt under a governor-live class. The pre-order plan had SL 0.32180 and TP 0.34500, but the market BUY filled 24 SYN at 0.33383, materially above the reference. With the same protection, actual post-fill R fell to about 0.93R and max loss rose to about 0.28872 USDT, roughly 0.31% of equity and above the rebuild cap. Bot-5 flattened immediately at 0.3301600, realizing -0.08808000 before commission accounting, then verified final flat/order-clean state. This is consistent with bot-2's earlier actual-fill R lesson, but today showed the same issue inside an otherwise permitted A-grade scout setup.
  Bots affected: all bots using market orders or fast-entry tactics, especially bot-2 and bot-5; relevant to bot-1, bot-3, and bot-4 when stop-fill-adjusted R is close to threshold.
  How to apply it: Before live order submission, state the maximum acceptable fill slippage or post-fill R floor. After fill, recompute actual entry-to-SL risk, TP distance, fees/spread/slippage allowance, and percent-equity loss. If actual fill breaks mandate R or risk cap, flatten or correct only if protection and grounded target geometry can be restored immediately without widening risk.
  Review date: 2026-07-08.

• Lesson: A risk-reducing trail is not the same as a profit-locking trail.
  Evidence: On 2026-06-23 bot-3's HEIUSDT short was opened with verified SL/TP, then replacement-first trailing reduced the original hard-stop risk from 0.08535 to 0.08335 and then 0.08240. The final stop still sat beyond fee/slippage break-even for the short, and a fast reversal filled at 0.085920, producing -0.24576000 USDT realized PnL before commission accounting. The operational process worked: the orphan TP was identified and cancelled by 05:42Z, and later root checks confirmed flat/order-clean. The lesson is narrower than "trail faster": completed-candle trail rules can reduce risk without securing realized profit, and reports should name which one happened.
  Bots affected: bot-1, bot-3, bot-5, with relevance to bot-2 paid-zone handling and bot-4 first-mean management.
  How to apply it: When moving a replacement stop, state whether the new stop is risk-reducing only or profit-locking after fees, spread, stop-fill allowance, and commission assets. Do not force mechanical break-even moves, but do not describe a trail as profit protection unless the adjusted stop actually locks profit.
  Review date: 2026-07-07.

• Lesson: During rebuild stabilization, weak overlapping setup classes should be paper-tracked or disabled rather than renamed into live risk.
  Evidence: From 2026-06-15 through 2026-06-22, the weakest repeated company setup families were not new unknown edges; they were familiar overlap zones between mandates. Bot-1's BTC/ETH/SOL damaged repair and low-location failed-auction watches needed defended retests and participation before live risk, including the June 22 BTC probe into 64,788-65,597 that faded without upgrade evidence. Bot-2's RE/source-light mover reviews and June 22 TAIKO official-security catalyst needed official-catalyst second-entry or accepted reaction structure rather than source quality alone. Bot-3's BICO/UB/MET-style upside watches and June 22 UB/BICO/H/SYN/ID/LAB/TIA/ENA/BEL set needed rebuilt defended shelves/retests before first-extension momentum could become live. Bot-4's LAB/RE/BTW/EVAA/H-style failed-move watches, plus June 22 BEL/SYN/CLO/UB watches, repeatedly failed on first-leg-paid, first-reclaim, fragile-stop, quiet-flow, or rebid-risk defects. Bot-5's PENGU/ALICE/SAND/VELVET, June 21 ALICE/PUMP/MET, and June 22 SIREN/DYM/WLD/RESOLV candidates had enough structure for paper research but lacked A-grade/deep-shelf participation quality. The governor kept these classes paper-only, and the later no-trade days showed reduced churn without hiding the opportunity-cost evidence.
  Bots affected: all bots, especially bot-2/bot-5 source-light versus scout overlap and bot-3/bot-5 continuation overlap.
  How to apply it: When a losing or paper-only setup class repeats during rebuild rollout, first ask whether it should remain disabled, be removed from live consideration, or be tracked on paper with outcome labels. Do not add a new live label or rewrite strategy direction unless repeated realized evidence shows a distinct profitable edge after costs, cleanup, and stale-state risk.
  Review date: 2026-07-05.

• Lesson: Rebuilt-shelf and first-expansion starters need fresh participation before a second risk decision.
  Evidence: On 2026-06-16 bot-3's UNIUSDT reduced momentum starter had accepted multi-timeframe structure but stopped before a trailable shelf, while STGUSDT entered from accepted 15m/1h rebreak and later stopped after fading back through the lower shelf. Bot-5's WLDUSDT rebuilt-shelf B-probe also stopped almost immediately, with slightly down/flat OI and an earlier failed evaluation warning that residual same-continuation risk was still present. On 2026-06-17 the same pattern repeated: bot-3's UNIUSDT accepted breakout starter failed quickly and was cut, bot-4's BSBUSDT failed-reclaim short rebid into its stop after the first downside leg had already traveled, and bot-5's WLDUSDT, SPXUSDT, and AEROUSDT B-grade continuation probes all stopped before durable follow-through. On 2026-06-18 the same defect appeared mostly as avoided risk rather than realized loss: bot-2's REUSDT launch watch stayed first-extension/chop without a second-entry shelf, bot-3 kept ZECUSDT and VELVETUSDT paper-only for missing rebuilt shelf/retest structure, bot-4 skipped H/RE/SYN-style failed-move watches after confirmation travel or missing accepted lower-high evidence, and bot-5 kept HOME/AERO/FOLKS/AVAX B-grade ideas paper-only because fresh participation or deep rebuilt structure was missing. On 2026-06-20 the same evidence repeated as a clean no-trade day: bot-1 rejected BTC/ETH damaged repair without a defended shelf and participation, bot-2 kept REUSDT under official-catalyst watch but flat because no second-entry structure formed, bot-3 rejected HOME/BSB-style downside candidates for flat OI/poor true-stop R or incomplete acceptance, bot-4 skipped LAB/RE/BTW/GUA/SYN ideas after first-leg-paid or fragile-stop defects, and bot-5 kept PENGU, ALICE, SAND, and VELVET B-grade continuation candidates paper-only. On 2026-06-21 the pattern continued without live loss: bot-1 rejected BTC/SOL repair-continuation attempts for quiet participation or sub-threshold R, bot-2 kept RE/LAB/TNSR/BICO-style candidates flat or paper-only without accepted second-entry structure, bot-3 rejected BICO/UB/MET-style first-extension momentum without rebuilt defended shelves, bot-4 stayed flat through first-leg-paid or fragile failed-move watches, and bot-5 paper-tracked ALICE/PUMP/MET continuation candidates instead of treating B-grade structure as live permission. On 2026-06-22 the same pattern repeated again as avoided risk: bot-1's BTC repair probe lacked defended retest and buyer participation, bot-2's TAIKO source lacked executable futures acceptance, bot-3 recurring alt momentum lacked rebuilt shelves, bot-4 failed highs lacked quiet-flow acceptance and preserved R, and bot-5's B-grade paper candidates lacked immediate renewed participation. On 2026-06-25 the same pattern remained an avoided-risk theme across all mandates: bot-1's BTC shorts lacked restored retest edge and post-cost room, bot-2's OUSDT/ARX launch watches lacked accepted second entries, bot-3's OUSDT/momentum reviews lacked signed-verification or complete rebuilt structure, bot-4's failed-move looks lacked honest-stop/unpaid-mean quality, and bot-5's B-grade scout names stayed paper-only without A-grade/deep-shelf upgrade evidence. Counterexamples stayed controlled rather than broad: bot-3 BASEDUSDT formed a higher trade-and-mark shelf and trailed near flat-positive on 2026-06-16, and bot-5 HYPEUSDT reached TP after accepted structure, practical protection, and valid bracket management.
  Bots affected: bot-3 and bot-5 primarily, with relevance to bot-1 liquid-major repair starters, bot-2 source-light second entries, and bot-4 first-retracement reclaims.
  How to apply it: After first-expansion acceptance or a rebuilt continuation shelf, require renewed OI/taker participation, shelf defense, executable spread/depth, and stop-fill-adjusted reward/risk before treating the setup as more than minimum/reduced starter risk. If the same symbol or same continuation theme already failed earlier in the day, write what materially reset before entering again.
  2026-06-28 update: The same pattern repeated as avoided risk across all five reports: bot-1 skipped BTC/SOL repair or support-loss ideas without defended retest plus participation, bot-2 kept CAP/IP/NFP and source-light movers flat without accepted second-entry structure, bot-3 labeled the dominant no-trade blocker structure missing, bot-4 rejected MANTA/SLX/ACT/VELVET failed moves after first-leg-paid or fragile-stop defects, and bot-5 kept VELVET/S/BTW/ACT/JTO/POWR/SYN paper or watch-only when participation, OI, depth, funding, or deep-shelf quality was incomplete.
  Review date: 2026-07-09.

• Lesson: Near-target management should distinguish no-trail discipline from avoidable giveback.
  Evidence: On 2026-06-15 bot-1's minimum-size BTCUSDT long moved close to TP and above break-even, but fresh flow and participation did not satisfy the written replacement-stop standard, so the bot correctly held the verified bracket rather than forcing a weaker trail. The same day bot-5's ONDOUSDT B-probe reached TP after mixed flow blocked adds/runners but did not justify manual capture, while bot-3's JTO accepted breakout failed quickly before a trailable shelf. Together the samples show that the company needs evidence labels around near-target holds, not a universal manual-capture or no-trail rule.
  Bots affected: bot-1, bot-3, bot-5, with relevance to bot-2 source-light paid-zone and bot-4 first-mean management.
  How to apply it: When a protected starter reaches roughly 1R+ or nears TP without valid replacement-stop evidence, write whether holding the bracket, reducing risk, or manually capturing is supported by completed structure, flow, liquidity, cleanup burden, and next-candle evidence. Do not loosen stops or force partials solely from an R multiple.
  Review date: 2026-06-29.

• Lesson: Protection verification must normalize exchange-returned order fields before declaring protection missing.
  Evidence: On 2026-06-12 bot-3 opened a reduced BSBUSDT breakdown starter with live SL/TP algos, but local verification compared Binance-normalized trigger values and boolean fields too literally and safety-closed the trade four seconds after entry for only +0.00621000. On 2026-06-13 bot-3's TAOUSDT and WLDUSDT live entries acted as normalized-match counterexamples: protection matched signed exchange intent, the positions were not incorrectly flattened, and both reached planned TP before ordinary post-exit cleanup. Earlier company incidents around protection endpoints and sibling-algo cleanup show that true missing protection is critical, but equivalent formatting should not force a flatten.
  Bots affected: all bots placing Binance USD-M conditional SL/TP protection, especially bot-1, bot-3, bot-4, and bot-5.
  How to apply it: Before flattening for alleged missing protection, compare normalized side, quantity, reduce-only intent, working type, trigger value, and symbol across local expectations and signed open-algo state. Flatten immediately for true mismatches or missing protection; do not flatten solely for trailing-zero, boolean, or endpoint field-format differences.
  Review date: 2026-06-26.

• Lesson: Accepted independent alt longs need cleaner participation in defensive majors before confidence upgrades.
  Evidence: On 2026-06-09 bot-3's IOUSDT long had completed 15m/1h acceptance and a nearby rebuilt shelf, but BTC was seller-aggressive, IO taker flow was balanced, liquidity was only moderate, and the trade stopped before a higher completed shelf formed. The same day bot-5's HOMEUSDT B-grade probes showed the counterexample: minimum-risk bracket discipline can still pay or remain valid when structure, protection, and room are clear, but mixed flow should block add, runner, or stop-loosening decisions. Bot-1 and bot-2 also stayed flat when damaged/first-flush conditions lacked a completed reset.
  Bots affected: bot-1, bot-3, bot-5, with relevance to bot-2 source-light longs and bot-4 contrarian reclaim entries.
  How to apply it: In seller-skewed BTC/ETH/SOL tape, accepted alt structure can justify only reduced/minimum starter risk unless participation, OI, liquidity, and post-entry shelf health are clean. Do not upgrade confidence, add, or convert to runner until completed structure plus live flow confirm the independent edge.
  Review date: 2026-06-23.

• Lesson: Same-symbol second probes need a distinct-edge reset after a same-day stop.
  Evidence: On 2026-06-07 bot-5 took two DASHUSDT B-grade continuation longs. The first reached meaningful positive excursion but did not form a valid replacement trail before stopping; the second used a rebuilt-shelf review but reused a similar 36.90 invalidation area and failed quickly. Earlier bot-2 source-light ZEC/LAB and bot-3 same-symbol re-entry tracking show the broader risk: a still-active symbol can remain interesting while the next entry is only a residual retry unless structure, participation, and stop/target geometry materially reset.
  Bots affected: bot-5 primarily, with relevance to bot-2 source-light same-name retries and bot-3 momentum same-symbol reviews.
  How to apply it: Before a same-symbol second B-grade/source-light/momentum probe after a same-day stop, write what is distinct: fresh shelf or failed-reclaim, renewed OI/taker participation, materially better BTC/ETH context, different invalidation, improved stop-fill-adjusted R, or lower cleanup burden. If the edge is mostly the same symbol and same level, reduce size further or skip until a fresh reset appears.
  Review date: 2026-06-21.

• Lesson: Independent strength needs fast post-entry confirmation when broad beta is weak.
  Evidence: On 2026-06-01 bot-1's HYPEUSDT long had independent daily/12h/4h acceptance, but BTC/ETH/SOL beta was weak and the trade failed before trail evidence, exiting on the written shelf-loss branch. Bot-3 stayed flat in similar defensive major context because LAB/HYPE/PLAY lacked rebuilt stopable shelves with renewed participation, while bot-5's GUA and STG B-grade winners worked only as clean first-target probes with verified protection and no runner upgrade. The cross-bot distinction was not whether alts could move independently, but whether the first post-entry shelves and participation were strong enough to justify continued exposure.
  Bots affected: bot-1, bot-3, bot-5, with review relevance to bot-2 source-light moves and bot-4 contrarian fades during weak BTC/ETH/SOL tape.
  How to apply it: In weak or damaged BTC/ETH/SOL context, keep independent alt trades possible but label early follow-through after entry. Require a defended shelf, renewed OI/taker participation, executable spread/depth, and realistic replacement-stop or first-target logic before increasing confidence, trailing late, adding, or converting a B-grade probe into a runner.
  Review date: 2026-06-15.

• Lesson: Universe and mandate eligibility must be checked before setup evaluation or order placement.
  Evidence: On 2026-05-28 bot-3 opened BEATUSDT after treating Binance USD-M availability as sufficient, then immediately closed for -0.03700000 USDT gross when final sanity review found prior bot-3 context had already excluded BEAT from its crypto mandate. The correction was fast and cleanup was clean, but the gate fired after entry. This is broadly relevant because bot-2 source-light names, bot-3 momentum candidates, bot-4 high-beta mean-reversion watches, and bot-5 smaller-alt scouts can all see exchange-listed symbols that do not necessarily fit the bot's mandate.
  Bots affected: all bots, especially bot-2, bot-3, bot-4, and bot-5.
  How to apply it: Before order work on a non-core-watch or ambiguous symbol, write a one-line eligibility/source check: why the symbol fits this bot's mandate, which local context or source confirms it, and whether any prior exclusion applies. If eligibility is unclear, stop setup evaluation until the mandate source is clarified.
  Review date: 2026-06-11.

• Lesson: Fast tight-shelf starters need fresh stop-fill realism before the next similar entry.
  Evidence: On 2026-05-26 bot-3's RIFUSDT accepted momentum starter slipped through its stop trigger and realized -0.34170000 USDT before commissions versus smaller planned trigger risk; bot-5's NEARUSDT and SUIUSDT B-grade starters both stopped quickly despite minimum risk and valid gross geometry; bot-1's BTCUSDT failed-auction short also stayed minimum-size while follow-through failed. This extends prior B-grade stop-slip evidence from bot-5 and SAHARA/RIF-style momentum to current live samples across mandates.
  Bots affected: bot-1, bot-3, bot-5, and any bot trading tight-shelf momentum, failed-auction, or smaller-alt B-grade starters.
  How to apply it: Before the next similar starter after a same-week stop-slip or fast shelf failure, compare planned trigger-level R with current spread/depth and the freshest adverse stop-fill sample. Keep size minimum, require cleaner participation/deeper shelf, or skip if adjusted post-cost R no longer clears the mandate.
  Review date: 2026-06-09.

• Lesson: Confirmation travel is an execution-window variable, not just a no-trade explanation.
  Evidence: On 2026-05-25 bot-4 repeatedly found failed-move or first-retracement candidates where the read was plausible, but by the time completed confirmation arrived the live price had already paid the first practical mean, required a fragile internal stop, or left less than about 1.3R-1.5R after costs. The same day bot-1's NEAR opportunity-cost review and bot-3's SAGA stopped breakout reinforced that valid-looking direction still needs tradable location, shelf construction, and post-cost room at the actual decision time.
  Bots affected: bot-1, bot-3, bot-4, and any bot using completed confirmation after fast single-name movement.
  How to apply it: When confirmation arrives after material travel, label the window preserved, decayed, or inconclusive with trigger, close, honest stop, first objective, and post-cost R. Do not chase a decayed window; require a fresh retest/hold, better shelf, or restored invalidation before entry.
  Review date: 2026-06-08.

• Lesson: Same-day B-grade and source-light churn needs a day-level freshness reset, not only valid individual setup checks.
  Evidence: On 2026-05-24 bot-5 took seven smaller-alt exits: NIL, HYPE, PLUME, second ASTER, FIDA, and late NIL stopped while only the first ASTER starter reached TP, leaving UTC-date realized trade PnL about -0.66495000 USDT before commissions/funding despite minimum sizing and clean protection. The later FIDA and NIL entries had valid individual B-grade structure, but they came after repeated same-day shelf failures. Bot-2's 2026-05-23 through 2026-05-24 PLAY/GRASS source-light sequence shows the same company-wide issue in a different mandate: exchange-observed abnormal flow can remain real while repeated same-theme shelves fail quickly. On 2026-05-27 the issue repeated: bot-5's SEI TP was outweighed by CHIP, ICP, HYPE, ZEC, and FF losses, while bot-2's official XLM catalyst paid but its source-light ICP/PHAROS starters stopped. On 2026-05-29, bot-2 stopped twice on ALLO source-light shelves, and bot-5's WLD/SEI wins were offset by repeated B-grade/probe failures in HYPE, 1000LUNC, AIGENSYN, AR, and later HYPE. On 2026-05-30, bot-5 again realized about -0.84300000 USDT from multiple B-grade/probe exits while NEAR remained open/protected; bot-2 stayed flat because first-extension exchange-observed movers did not rebuild clean entries. On 2026-05-31, bot-5's ZEC and 1000BONK B-grade probes stopped before a later protected HYPE probe remained open, while bot-2 again stayed flat because PLAY/STG/PORTAL-style first-extension watches lacked fresh second-entry structure. On 2026-06-02, bot-5's early OPG and WLD B-grade TPs were later given back by XLM, SUI, ICP, XLM again, and ENA stopouts, leaving daily realized PnL near flat before costs and reinforcing that later-session B probes need a distinct day-quality reset. On 2026-06-03 the same lesson stayed active rather than resolved: bot-5 started with a stale-flat BILL short that stopped, then took a distinct PENDLE B-grade breakdown probe that trailed into a small gross gain but still required active monitoring and orphan-TP cleanup; bot-2, bot-3, and bot-4 all rejected high-OI or first-extension watches until second-entry shelves, lower-highs, or reclaim shelves formed. On 2026-06-04 bot-5 again stopped in late-prior RAVE plus same-day OPN, TON, and EPIC B-grade probes, including an EPIC near-TP mark-fill miss that still became a stop, while bot-2 stayed flat in competing fresh-listing and high-OI first-extension conditions. On 2026-06-05 the evidence was mixed but still useful: bot-2's ZEC source-light downside starter stopped quickly after a fast reclaim through invalidation, while bot-5's OPN B-grade reclaim long reached planned TP after holding the verified bracket, showing that the reset/acceptance question should be measured rather than converted into a blanket throttle. On 2026-06-06 the pattern stayed active: bot-2's LAB source-light downside starter moved in favor but failed to accept lower before reclaiming through invalidation, and bot-5's XLM B-grade probe reached roughly +0.8R without TP or trail evidence before stopping. On 2026-06-07, bot-5's PENGU/ZEC/JTO wins showed verified B-grade brackets can still pay, but the two DASH stops showed the same-day reset problem remains active when a symbol is retried after positive excursion without a materially distinct edge. On 2026-06-08, bot-5's ALLO B-grade TP paid enough to leave the day positive, but later ZEC and HYPE B-grade longs both stopped; bot-2 stayed flat through similar first-extension source-light movers because no second-entry shelf/retest restored clean risk. On 2026-06-10, bot-5 closed four minimum-risk B-grade probes red across JTO, HOME, DASH, and TAO; bot-4 also showed the related first-retracement risk when ZEC reclaimed but failed to follow through and stopped. On 2026-06-13, bot-5's TRUMP and WLD minimum-risk B-grade continuation probes both stopped and shared the same high-location/fading fresh-flow defect, while its TAO live guard correctly prevented a third weakening continuation probe. On 2026-06-14, bot-5's PENGU short was manually safety-closed after shelf reclaim, TAO short stopped after down-OI/traveled-location risk, and post-report ZEC long was manually safety-closed after accepting into its failure band despite being a changed-edge long probe. On 2026-06-18 bot-5 journaled HOME, AERO, FOLKS, and AVAX as paper/research B-grade ideas but did not submit live orders because participation freshness, OI, or rebuilt-shelf quality was insufficient; this is the desired application of the lesson during rebuild stabilization. This reinforces measurement and reset discipline, not a blanket ban.
  Bots affected: bot-2 and bot-5 primarily, with relevance to bot-3 accepted-shelf momentum and bot-4 first-retracement reviews.
  How to apply it: After multiple same-day B-grade/source-light shelf failures, require a day-level reset line before later entries: fresh catalyst, materially improved BTC/ETH context, deeper rebuilt shelf, renewed participation/OI, cleaner execution, or better stop-fill realism. If the reset is not confirmed, downgrade later setups, reduce size further, or skip even when the individual shelf is technically valid.
  Review date: 2026-06-27.

• Lesson: After a B-grade/source-light loss cluster, the next non-A risk must prove a changed edge and manageable cleanup load.
  Evidence: On 2026-05-30 bot-5 realized about -0.84300000 USDT from HYPE, XLM, EPIC, NFP, ONDO, and ASTER exits while NEAR was the better active protected short; later NEAR and ENA winners showed scout exposure itself was not the issue, but marginal follow-on probes added monitoring and cleanup load. Bot-2's 2026-05-29 ALLO double-stop and later source-light no-trade scans show the same reset problem in catalyst form. During the 2026-06-08 through 2026-06-14 review week the evidence strengthened: bot-5 stopped JTO/HOME/DASH/TAO on 2026-06-10, then TRUMP/WLD on 2026-06-13, while ONDO and ALLO winners showed distinct shelves can still work. Bot-5's 2026-06-14 PENGU manual safety exit, TAO stop, and later ZEC changed-edge probe safety exit reinforce that the operational cost is not just PnL; it is active monitoring, stale-state checks, and sibling-algo cleanup after repeated probes. Bot-3 reduced-starter clusters and bot-4 first-retracement reviews make the lesson relevant outside bot-5 even though the strongest evidence remains bot-5/bot-2.
  Bots affected: bot-5 and bot-2 primarily, with relevance to bot-3 reduced momentum starters, bot-4 first-retracement/failed-move starters, and any bot considering new non-A exposure after a same-day loss cluster.
  How to apply it: After three or more same-day B-grade/source-light/probe losses, or when cleanup/active-position burden from that cluster remains fresh, label the next non-A entry post-cluster edge changed, same-continuation risk, or inconclusive. Name whether the setup is A-grade, failed-acceptance reversal, deeper rebuilt shelf, or only another continuation probe; cite current open risk, monitoring cadence, cleanup burden, participation/freshness, stop-fill realism, and why waiting for A-grade or failed-acceptance structure is not better. This is a grading and attention filter, not a hard cooldown.
  Review date: 2026-06-28.

• Lesson: Same-symbol or same-theme re-entry after a stopped starter needs an explicit whipsaw reset.
  Evidence: On 2026-05-23 bot-3 took two process-valid reduced-size ONDOUSDT momentum starters in opposite directions: the accepted breakdown short stopped for -0.45342000 USDT before commission accounting, then the accepted reclaim/rebreak long stopped for -0.26136000 USDT before commission accounting. On 2026-05-23 through 2026-05-24 bot-2's source-light PLAY/GRASS sequence also showed that exchange-observed abnormal flow can remain active while repeated same-theme shelves fail quickly; PLAY produced one tightened-stop gain on 2026-05-23, then a fresh lower-high PLAY short on 2026-05-24 stopped for -0.30528000 USDT, while GRASS stopped for -0.31450000 USDT. Bot-5's B-grade scout churn adds broader support: individually valid shelves can still fail when the same tape keeps rejecting follow-through.
  Bots affected: bot-2, bot-3, bot-5, and any bot considering same-symbol, opposite-side, or same-theme re-entry after a stopped starter.
  How to apply it: After a stopped starter, require a one-line whipsaw reset before same-symbol or same-theme re-entry: what changed, whether price accepted away from the failed shelf, whether OI/participation renewed in the new direction, and whether stop-fill-adjusted reward/risk still clears the mandate. If the reset is only "same active symbol, new side," reduce size further or defer.
  Review date: 2026-06-07.

• Lesson: Source-light and rebuilt-shelf entries need live entry-shelf participation, not just OI and turnover.
  Evidence: On 2026-05-20 bot-2 took EDENUSDT and FIDAUSDT source-light exchange-observed starter longs after abnormal turnover/OI and accepted post-impulse shelves, but both shelves failed within minutes for -0.47432000 USDT combined before commission accounting. On 2026-05-23 and 2026-05-24 the same fragility appeared in bot-2's GRASS and PLAY starters and bot-5's B-grade scout sequence: headline volume, spread/depth, or OI could be adequate while the immediate shelf tape was quiet, mixed, seller-leaning, or late. Winners still existed, including bot-5 BEAT/SUI/INJ and bot-3 HYPE, but the stronger samples had cleaner acceptance, target room, or replacement-first management. On 2026-06-08 bot-3's BNB short had an initially acceptable 15m shelf breakdown but failed when completed trade-and-mark candles reclaimed the shelf, while bot-5's ZEC and HYPE B-grade longs also stopped after participation failed to upgrade.
  Bots affected: bot-2 and bot-5 primarily, with relevance to bot-3 momentum and bot-4 first-retracement reviews.
  How to apply it: Before source-light starter longs, exchange-observed shorts, or rebuilt-shelf scout entries, classify live entry-shelf tape as directional continuation, mixed, or warning. If current participation/flow does not support the shelf, reduce size further, require a cleaner fresh reset, or skip even when OI/turnover and broad structure are interesting.
  Review date: 2026-06-07.

• Lesson: First tactical paid zones need explicit capture-or-hold evidence before reversal risk is allowed to rebuild.
  Evidence: On 2026-05-23 bot-4 manually captured ZECUSDT and EDENUSDT in written paid-zone review bands for +1.48800000 USDT before commission asset effects after OI/participation stopped supporting a full bracket hold. Bot-2's PLAYUSDT source-light short also moved in favor, then sharply reclaimed; tightening the SL converted the trade into a small gain instead of leaving the original loose stop exposed. These samples extend the existing paid-zone lesson from open-profit review into catalyst and mean-reversion first-leg management.
  Bots affected: bot-2, bot-4, and any bot managing source-light, failed-move, snapback, or fast smaller-alt trades after the first practical leg pays.
  How to apply it: At the first tactical paid zone, write whether completed structure and participation justify holding for the full TP, tightening risk, or manually capturing. Do not hold a mean-reversion or source-light trade through reversal risk just because the original bracket exists; require fresh evidence if the first leg has already paid.
  Review date: 2026-06-06.

• Lesson: Max-entry and late-fill gates are part of execution edge, not administrative friction.
  Evidence: On 2026-05-18 bot-2 evaluated FIDAUSDT as a source-light starter after a 0.0230-0.0232 reclaim/hold with expanding OI and deeply negative funding, but the order script aborted before submission because live price moved above the planned 0.02365 max-entry gate. Later review found FIDA still active but no longer executable because OI was slightly down, participation was quiet, and price was higher-location continuation. Bot-5 similarly skipped a later FIDA setup when the live ask moved to 0.02647 and compressed reward/risk below mandate. These samples show the setup can remain interesting while the executable entry is gone.
  Bots affected: bot-2, bot-5, and any bot using scripted max-entry, limit, or chase-prevention gates.
  How to apply it: Treat a max-entry abort as a completed risk decision. Do not immediately chase; require a fresh pullback/reclaim, failed-auction reset, or new setup review that restores nearby invalidation and post-cost reward/risk. Outcome-check aborted candidates so the gate can be tightened or loosened only from evidence.
  Review date: 2026-06-01.

• Lesson: Accepted entries need immediate post-entry health checks before confidence increases.
  Evidence: On 2026-05-16 bot-2 and bot-3 both captured RUNEUSDT shorts when catalyst or multi-timeframe downside acceptance aligned with renewed seller participation and practical invalidation. The same day, bot-3's 1000LUNCUSDT long failed after accepted 15m/1h upside structure but no healthy post-entry shelf, bot-4's ZECUSDT first-retracement long failed before reaching the paid-snapback zone, and bot-5's ONDO/1000LUNC/SUI starters all failed to follow through. The contrast says entry acceptance is necessary, but first completed post-entry candle health, shelf hold, and participation reset are useful cross-bot evidence before trailing, re-entering, or adding confidence.
  Bots affected: bot-2, bot-3, bot-4, bot-5, and any bot using catalyst continuation, momentum, first-retracement, or rebuilt-shelf entries.
  How to apply it: After entry, record whether the first completed post-entry candle expands, holds, or fades versus the entry shelf or failed-reclaim level. Do not widen stops or increase size from entry acceptance alone; require follow-through, renewed participation, and verified protection before confidence increases.
  Review date: 2026-05-30.

• Lesson: Paid review zones need explicit management evidence before a trade is allowed to round-trip.
  Evidence: On 2026-05-14 bot-4's second LABUSDT failed-bounce short moved favorably into the 5.47-5.55 paid-snapback review area, then reclaimed through invalidation and stopped for -0.70410000 USDT before commission asset effects. On the same day bot-1's BTCUSDT starter reached more than 1R open profit, but the bot correctly did not trail without completed acceptance above the written 81.9k-82.1k higher-shelf zone. Together, the evidence says review-zone management should be explicit: either completed acceptance supports holding, or stall/reclaim evidence should trigger capture or risk reduction if the mandate allows it.
  Bots affected: bot-1, bot-3, bot-4, bot-5, and any bot managing open profit before full TP.
  How to apply it: Before holding through a paid review zone, write the evidence that justifies continued hold versus manual capture, partial capture, or stop tightening. Do not move protection on active-candle profit alone; require completed structure and verified replacement protection before changing orders.
  Review date: 2026-05-28.

• Lesson: Shallow continuation starters need extra scrutiny in defensive broad context and scheduled event windows.
  Evidence: On 2026-05-13 bot-1's BNBUSDT reduced-risk breakout starter stopped during the 12:30Z event move after BTC lost the 80.6k-81.0k shelf and ETH/SOL stayed weak. Bot-3's ZECUSDT and KITEUSDT accepted-shelf momentum starters also failed their shelves, while INJUSDT worked only after cleaner 15m/1h/4h acceptance and follow-through. The evidence supports keeping the starter path alive, but only with reduced risk, realistic slippage, and stronger broad-context sizing discipline.
  Bots affected: bot-1, bot-3, bot-5, and any bot taking shallow continuation or rebuilt-shelf alt exposure.
  How to apply it: Before entry, state whether BTC/ETH/SOL context is supportive, mixed, or hostile and whether a scheduled event can hit before the trade reaches a trail/add/protection reassessment zone. Use minimum starter size or defer when broad support is weakening unless the symbol has exceptional independent acceptance, nearby invalidation, and post-cost reward/risk.
  Review date: 2026-05-27.

• Lesson: Verified protection is a hard execution gate; if protection placement fails, flatten first and debug later.
  Evidence: On 2026-05-12 bot-4 entered a reduced-risk BUSDT failed-move short, but the normal STOP/TP path was unsupported and the follow-up Binance algo request used the wrong endpoint parameter name. Bot-4 immediately flattened for about +0.07450000 USDT before commission asset effects, then verified no position, normal orders, or open algo orders. On 2026-05-16 and 2026-05-17, bot-1 and bot-2 also had to use the Binance futures algo path for close-position protection after standard conditional-order attempts failed or needed correction, and bot-5 immediately flattened a CHZ attempt when algoType=CONDITIONAL was missing before re-entering with verified protection. On 2026-05-30 the same operational theme recurred: bot-1 had initial HYPE/BNB protection payload friction before flatten/correct handling, bot-3's BNB normal STOP_MARKET attempt rejected before clean algo fallback, and bot-5's ONDO protection call needed the required algoType parameter before final verification. On 2026-05-31 bot-1 again recorded a BNB failed-protection flatten before corrected algotype=CONDITIONAL protection worked on the retry. These samples confirm that the setup is not truly live until protective algos are verified.
  Bots affected: all bots placing Binance USD-M live orders with protective SL/TP algos.
  How to apply it: Before entry, know the exact account-specific protection path, including required algo parameters. After entry, treat the trade as unprotected until SL/TP algos are verified live through the correct futures algo-order endpoint. If protection cannot be verified promptly, flatten and reconcile; do not wait for the setup to develop.
  Review date: 2026-06-14.

• Lesson: Reduced-size accepted-shelf momentum can work in mixed broad-market context, but only with completed structure, realistic costs, and cleanup.
  Evidence: On 2026-05-12 bot-3 closed SAGAUSDT and USELESSUSDT longs at planned TPs and closed TRUMPUSDT short by trailed SL for +1.09532000 USDT gross before commission accounting. BTC/ETH/SOL context was mixed-to-soft, but each trade had accepted symbol-level structure, nearby invalidation, reduced sizing, verified management, and flat-state/order cleanup. The prior USELESS stopped trade and later no-trade scans still show that this is not permission to chase first extensions. During the 2026-06-01 through 2026-06-07 review week, the lesson became more balanced: bot-3's CHZ downside starter and TAO reduced-size long both showed that replacement-first trailing and cleanup can preserve small gains when completed shelves form, while bot-5's PENGU and JTO B-grade continuation probes reached planned TP with verified brackets and orphan-SL cleanup. Bot-5's DASH and XLM stops, plus bot-2's LAB/ZEC source-light stops, show the same path still fails when post-entry acceptance or broad tape deteriorates before a replacement shelf forms.
  Bots affected: bot-3 and bot-5 primarily, with review relevance to bot-2 source-light starters and any bot evaluating independent liquid-alt momentum while majors are mixed.
  How to apply it: Keep the reduced-risk path available when a symbol completes accepted 5m/15m/1h structure with renewed participation, close invalidation, clean spread/depth/funding, and post-cost reward/risk. Do not increase size or remove shelf/retest requirements until net-after-cost samples support it.
  Review date: 2026-06-21.

• Lesson: Execution quality is necessary but does not substitute for accepted structure near invalidation.
  Evidence: During the 2026-05-04 through 2026-05-10 review week, bot-5 repeatedly found SUI, ENA, ZEC, 1000PEPE, HYPE, SAHARA, ASTER, TAO, LAYER, and BILL with measurable spread/depth or turnover, but correctly rejected them when rebuilt shelves, completed 5m/15m participation, nearby invalidation, or post-cost reward/risk were missing. Bot-2 also rejected BILL and other venue-flow movers when aged source edge, vertical location, or thin depth outweighed turnover. Bot-3's positive OP and HYPE pilots worked only after accepted shelf structure and renewed participation, while later SUI/UNI/Q scans stayed flat when acceptance or retest shelves were incomplete.
  Bots affected: bot-2, bot-3, bot-5, and any bot using abnormal volume, venue flow, or reduced-size alt pilots.
  How to apply it: Treat spread, depth, broad turnover, and recent-trade quality as execution qualifiers. They can permit a setup to proceed to entry review, but they cannot replace accepted shelf/retest/failure structure, renewed participation, close invalidation, and realistic post-cost reward/risk.
  Review date: 2026-05-24.

• Lesson: BTC/ETH regime context should throttle size and evidence requirements, not become a shared hard veto for independent alt structure.
  Evidence: The 2026-05-08 OPUSDT and 2026-05-09 HYPEUSDT bot-3 reduced-size momentum longs closed profitably while BTC/ETH were mixed rather than perfectly confirmed, because the symbols had accepted structure, nearby invalidation, and verified protection. Bot-1, bot-3, and bot-5 reports through 2026-05-09 and 2026-05-10 consistently treated BTC/ETH as context: weak or unresolved majors demanded cleaner symbol-specific evidence and reduced size, but did not automatically prohibit independent liquid-alt review.
  Bots affected: bot-1, bot-3, bot-5, and any bot evaluating liquid-alt or major-alt participation in mixed regime conditions.
  How to apply it: Use BTC/ETH context to adjust size, required confirmation, and whether independent evidence must be stronger. Do not use mixed BTC/ETH alone to reject a liquid symbol that has accepted independent structure, clean invalidation, workable execution, and enough post-cost reward/risk; also do not use this lesson to chase high-location impulses without structure.
  Review date: 2026-05-24.

• Lesson: Reduced-size independent-alt participation needs both accepted structure and post-exit cleanup evidence.
  Evidence: Bot-3's 2026-05-08 OPUSDT reduced-size shelf-acceptance long reached TP for +0.89300000 USDT gross before commission accounting, and its 2026-05-09 HYPEUSDT reduced-size 15m shelf breakout trailed into profit and exited for +0.37230000 USDT gross before commission accounting. Both trades required verified protection, structure-based management, and sibling-algo cleanup after flat state. The same day, bot-4's ONDO and BILL failed-move shorts showed that mandate-fit entries can still fail quickly when price reclaims the shelf and OI remains firm.
  Bots affected: bot-3, bot-4, bot-5, and any bot using reduced-size alt pilots.
  How to apply it: Allow reduced-size pilots only after accepted structure, renewed participation, close invalidation, realistic post-cost reward/risk, and verified protection. After any TP/SL/manual flat exit, reconcile position, normal orders, and open algo orders before treating the sample as complete evidence.
  Review date: 2026-05-23.

• Lesson: After a failed major shelf around a scheduled macro window, require fresh reclaim or underside-retest evidence before switching bias.
  Evidence: On 2026-05-07, BTC spent early scans around the 80.6k-81.0k breakout shelf, then lost it after the 12:30Z labor/productivity data window. Bot-1, bot-2, bot-3, bot-4, and bot-5 all stayed flat because neither continuation, failed-auction short, momentum, mean-reversion, nor smaller-alt scout evidence completed with clean flow, invalidation, and reward/risk.
  Bots affected: all bots.
  How to apply it: Treat the first shelf loss as a decision point, not a trade by itself. Longs need a completed reclaim/acceptance with practical invalidation; shorts or hedges need an underside retest/rejection or accepted breakdown with confirming flow; alt trades need BTC/ETH permission or explicit independent-strength evidence.
  Review date: 2026-05-21.

• Lesson: Confirmed trend regimes need an explicit participation path, not only perfect retests.
  Evidence: From 2026-05-02 through 2026-05-06, the system produced multiple zero-trade days while BTC advanced, SOL showed relative strength, and the market produced meaningful short-liquidation/trend behavior. The skips were individually defensible under prior rules, but together they showed that retest-only, range-top, source-quality, and depth filters can become too restrictive when a major breakout has accepted and liquid majors/alts rebuild shallow continuation shelves. On 2026-05-08, bot-3 created the first live test of this lesson with a reduced-size OPUSDT long only after accepted 15m/1h structure, close invalidation, buyer flow, and verified SL/TP protection; it later closed at TP for +0.89300000 USDT gross before commission accounting. On 2026-05-09, bot-3 added a second reduced-size HYPEUSDT shelf-break sample that trailed out profitably for +0.37230000 USDT gross before commission accounting. These are evidence for continuing the controlled path, not for loosening to full-size chase entries.
  Bots affected: all bots.
  How to apply it: In an accepted BTC/SOL-led trend, each bot must have a reduced-size participation route that fits its mandate: bot-1 swing continuation or retracement hedge, bot-2 market-wide liquidity catalyst, bot-3 shallow-shelf momentum, bot-4 trend-pullback/retracement reversion, and bot-5 liquid alt rebuilt-shelf scout. Keep structural invalidation, realistic costs, and reduced risk; do not use this lesson to justify blind chasing.
  Review date: 2026-05-22.

• Lesson: Use distilled order-flow snapshots first, but keep raw market data access for setup-specific edge.
  Evidence: On 2026-05-04 root added shared/orderflow_snapshot.py after Chart Champions research increased focus on OI/CVD/volume. The tool summarizes Binance USD-M OI, taker flow, volume, depth, funding, and recent aggregate-trade delta without dumping large raw arrays into bot context.
  Bots affected: all bots.
  How to apply it: Use compact snapshots for screening and setup confirmation, then inspect raw klines/trades only when the bot's mandate requires exact candle structure, price-level detail, or custom evidence. Do not create generalized tools that hide a bot's real edge.
  Review date: 2026-05-18.

• Lesson: When using Binance USD-M algo TP/SL sibling protection, a filled TP or SL does not prove the sibling algo was cancelled.
  Evidence: On 2026-04-28 bot-4's DAMUSDT TP fill left the SL algo open until manual cancellation, and bot-4's AIOTUSDT SL fill left the TP algo open until manual cancellation. During the 2026-05-11 through 2026-05-17 review week, the same pattern repeated across bot-3 and bot-5: bot-3 cancelled orphaned sibling algos after RUNE, CHZ, HYPE, and 1000LUNC exits, while bot-5 repeatedly cancelled sibling TP/SL algos after ONDO, 1000LUNC, GUA, HYPE, CHZ, BUS, and KAIA-related exits. The 2026-05-17 bot-3 orphan TP was later absent on root sweep, showing cleanup worked but was still required. On 2026-06-02 the pattern remained routine: bot-3 cancelled the orphaned CHZ TP after SL, bot-4 cancelled the orphaned WLD TP after SL, and bot-5 cleaned sibling algos after OPG/WLD TPs and XLM/SUI/ICP/ENA stops before marking flat. On 2026-06-06 bot-2 cancelled the orphan LAB TP after SL, and bot-5 cancelled orphan TPs after late-prior ZEST and same-day XLM stops before marking flat. On 2026-06-13 bot-3 cleaned stale sibling protection after ID and TAO exits, and bot-5 cleaned orphan TPs after TRUMP and WLD stops before marking final flat/order-clean state. On 2026-06-19 root's read-only sweep found bot-3 exchange-flat after a likely TAOUSDT SL fill while local records still showed TAO active and one reduce-only TP algo remained live, reinforcing that post-report exits need same-window position/order/algo reconciliation.
  Bots affected: all bots that place Binance USD-M reduce-only algo TP/SL protection.
  How to apply it: After any TP or SL exit, re-query positions, normal orders, and open algo orders while flat; cancel any orphaned reduce-only sibling before marking the position closed.
  Review date: 2026-06-27.

• Lesson: A no-trade day still needs process evidence when a bot formally evaluates and rejects a setup.
  Evidence: On 2026-04-28 bot-1, bot-2, bot-3, bot-4, and bot-5 all produced daily reports, and formal skipped setups in the journals made the no-trade decisions auditable by level, catalyst, liquidity, spread, regime, or invalidation. The 2026-05-05 daily reports again showed useful no-trade evidence across distinct mandates: BTC non-chase context for bot-1, source/live-flow separation for bot-2, tagged poor-R/late momentum skips for bot-3, first-leg-paid failed-move skips for bot-4, and depth/shelf checks for bot-5.
  Bots affected: all bots.
  How to apply it: When a concrete setup is reviewed and rejected, write a short no-trade journal entry with the setup, reason, condition that would change the decision, and next check. Pure passive monitoring does not need a forced journal entry.
  Review date: 2026-05-19.

• Lesson: After a protected exit, final flat state must include positions, normal orders, and algo orders.
  Evidence: Bot-5's 2026-04-29 HYPEUSDT long closed by planned SL, and follow-up signed checks verified positionAmt 0, no normal open orders, and no open algo orders. This matches bot-4's earlier DAM/AIOT cleanup evidence. On 2026-04-30, bot-4 cancelled the stale AIGENSYN sibling TP after SL and bot-5 cancelled a stale 1000PEPE sibling TP after SL; both verified flat/no-order state afterward.
  Bots affected: all bots using Binance USD-M TP/SL algos.
  How to apply it: Do not mark a trade fully closed until position size, normal open orders, and open algo orders are all reconciled.
  Review date: 2026-05-24.

• Lesson: Fast single-name and smaller-alt trades need explicit all-in adverse-cost allowance before sizing near the risk cap.
  Evidence: Bot-5's 2026-04-29 HYPEUSDT and 2026-04-30 1000PEPE/1000LUNC stopped trades showed that commission asset effects, slippage, and funding/accounting drag can matter on a 0.50% risk budget. Bot-3's 2026-04-28 BIOUSDT stop filled below trigger, and its 2026-05-01 ZECUSDT stop triggered at 379.10 but filled at 378.41, producing -1.14000000 USDT gross realized PnL versus about 0.9675 USDT planned pre-fee stop risk. Bot-4's 2026-05-10 BILLUSDT short also showed this for failed-move trades: the planned 0.1202000 stop filled at 0.1217200 average, turning roughly 0.62205000 USDT planned pre-fee risk into -0.83941000 USDT realized PnL before commission asset effects. On 2026-05-15, bot-5's CGPTUSDT long stopped at 0.0415400 versus the 0.0418000 mark trigger, turning a valid tight-shelf starter into a larger-than-planned loss on a low-priced ticker with roughly 4-5 bps spread.
  Bots affected: bot-3, bot-4, bot-5, and any bot trading high-beta alts or fast single-name momentum.
  How to apply it: Before entry, compare planned trigger-level loss with a conservative slippage, commission, and funding allowance; size below the mandate cap or skip if the all-in adverse cost would breach the bot's risk budget.
  Review date: 2026-05-24.

• Lesson: A paid first leg is not a fresh entry.
  Evidence: During the 2026-04-27 through 2026-05-03 review week, bot-4 repeatedly rejected failed-move candidates because the first snapback had already reached the practical mean; on 2026-05-02 it rejected ORCA/KNC-style failed moves where the first fade had already paid or remaining R was poor. Bot-3 rejected BIO/KNC/HYPE momentum when entries were late, near 24h highs, or lacked accepted continuation, and the later BIO/KNC/HYPE outcome check supported the late-entry/failed-acceptance filters for that sample. Bot-5 rejected ZEC/HYPE/TAO/1000PEPE-style scout setups when the impulse had already traveled, depth was thin, or no controlled retest restored invalidation.
  Bots affected: bot-3, bot-4, bot-5.
  How to apply it: Require a fresh retest, reclaim/failure, or renewed acceptance that restores reward/risk before entering after the initial move has already traveled.
  Review date: 2026-05-24.

• Lesson: High-dispersion movement is only a review trigger; tradable location still needs mandate-specific proof.
  Evidence: During the 2026-05-02 and 2026-05-03 weekend sessions, all bots reported active dispersion but stayed flat. Bot-1 skipped recurring alt strength because BTC/ETH permission was incomplete; bot-2 skipped ENA/LAB and abnormal movers without fresh catalyst/live-flow confirmation; bot-3 rejected BIO/KNC/HYPE/SKYAI/TAC momentum without accepted continuation and slippage-adjusted R; bot-4 rejected failed-move candidates after first legs paid or honest stops were too wide; bot-5 rejected active-trigger names when depth, participation, or invalidation was not clean. On 2026-05-05, BTC improved but the same standard still prevented late or under-confirmed entries across HYPE, ZEC, 1000LUNC, DOGS, LAB, BSB, TST, ENA, and 1000PEPE. On 2026-05-06, BTC stayed constructive above 80.6k-80.8k and ZEC/TON/TAO/LAB/HYPE were active, but all five bots remained flat because ETH lagged, entries were late or first-leg-paid, source quality was conflicted, or rebuilt shelves/accepted retests did not form.
  Bots affected: all bots.
  How to apply it: Treat unusual movement as a prompt to review, not as an entry reason. Before entry, require the bot's own proof: BTC/ETH permission for swing, fresh catalyst plus live flow for catalyst trades, accepted continuation for momentum, completed failure/retest for mean reversion, and spread/depth plus nearby invalidation for smaller-alt scouts.
  Review date: 2026-05-24.

Under Review

• Lesson: A second same-day reduced starter should show a distinct edge after the first one fails before a trailable shelf.
  Evidence: On 2026-06-11 bot-3 took two valid reduced-size starters, XMRUSDT long and INJUSDT short, and both stopped before a completed trade-and-mark shelf formed. The symbols and directions differed, but both had incomplete participation or confirmation support: XMR was first-expansion with flat OI and seller-leaning fresh aggregates, while INJ had incomplete 4h acceptance and falling OI. Bot-2's same-day STG sequence and bot-5's XMR/ONDO contrast show the broader company issue: second risk can be valid, but it should prove what changed rather than merely reuse minimum qualifying structure. The 2026-06-29 bot-3 GWEI and bot-5 SLX/ACT sequences and 2026-06-30 bot-2 TAC same-name reset losses add current evidence that clean protection and small sizing do not by themselves solve same-day reduced-starter churn.
  Bots affected: bot-2, bot-3, and bot-5 primarily, with review relevance to bot-1 same-symbol relative-strength retries and bot-4 fresh-retest repeats.
  How to apply it: After a reduced starter stops before a trailable shelf, require the next same-day reduced starter or same-symbol reset review to state whether the edge is distinct or the same defect is recurring. Cite confirmation timeframe, OI/taker flow, shelf quality, stop-fill realism, cleanup burden, and whether the label changes entry or size. This is evidence tracking, not a hard one-trade limit.
  Review date: 2026-07-14.

• Lesson: After one clean winner in mixed tape, later same-session starters need a freshness label before confidence carries over.
  Evidence: On 2026-05-22 bot-3 captured VIRTUALUSDT to planned TP, but later GRASSUSDT and NEARUSDT accepted-shelf longs stopped quickly and SAHARAUSDT's trailed short still lost after stop-market slippage. Bot-5 had a similar split: GRASS paid twice while EDEN, ATOM, NEAR, JTO, WLD, and INJ B-grade or minimum-risk starters failed, leaving the day negative despite controlled sizing. The evidence does not ban later trades, but it says a prior winner should not lower the evidence bar in the same mixed/defensive tape.
  Bots affected: bot-3 and bot-5 primarily, with relevance to bot-2 source-light follow-through and bot-4 failed-move resets.
  How to apply it: After a first clean TP in mixed or defensive broad context, label later same-session starters as fresh independent impulse/reset, churn continuation, or inconclusive. Require what changed: new catalyst, materially improved BTC/ETH context, renewed participation/OI, deeper shelf, cleaner stop-fill realism, or stronger first post-entry candle health.
  Review date: 2026-06-05.

• Lesson: B-grade fast-shelf starters need adjusted reward/risk after same-day stop-slip outliers.
  Evidence: On 2026-05-21 bot-5's BSBUSDT minimum-risk continuation starter had planned price risk around 0.22145 USDT with stop trigger 1.0750000, but the stop-side fill printed at 1.0460700 for -0.36610000 USDT before commissions. Later VVVUSDT, another B-grade accepted breakout/continuation starter, stopped for -0.17253000 USDT before commissions after entering at 19.018000 and stopping at 18.379000. On 2026-05-22 bot-5 again had a negative B-grade/minimum-risk sequence despite two GRASS winners, and bot-3's SAHARA trailed stop filled worse than trigger after a fast reversal. HYPE, GRASS, and several trailed positions still worked, so the evidence argues for adjusted-R discipline rather than a ban.
  Bots affected: bot-5 primarily, with relevance to bot-3 momentum and any bot trading fast high-beta shelves.
  How to apply it: After a same-day stop fill exceeds the written slippage allowance, later B-grade fast-shelf entries should compare planned R against current spread/depth and the worst fresh stop-fill sample. Reduce size, require cleaner participation/deeper shelf, or skip if adjusted post-cost R no longer clears mandate.
  Review date: 2026-06-04.

• Lesson: Late executable windows need a fresh access, location, and reward/risk check before submission.
  Evidence: On 2026-05-19 bot-2 rejected an early RONIN review partly because signed account/order state was unavailable in that shell, even though later checks recovered. Bot-4 prepared a late BSB failed-move order path but correctly aborted before exchange submission when price slipped past the max-entry math and compressed expected reward/risk to about 1.08R-1.09R. Bot-5's same-day smaller-alt sequence showed that repeated fast stopouts, stop-slip, and sibling-algo cleanup can turn nominally valid later starters into a cost-sensitive decision.
  Bots affected: bot-2, bot-4, bot-5, with review relevance to all bots using live order scripts or fast post-move entries.
  How to apply it: Immediately before submission on fast-moving setups, verify signed access if required, current location versus max-entry, stop-slip-adjusted reward/risk, and whether same-day cleanup/cost drag changes the quality threshold. Treat an abort as a risk decision and outcome-check it before loosening gates.
  Review date: 2026-06-02.

• Lesson: Repeated same-day reduced-risk starters need a fresh reset check after multiple shelf failures.
  Evidence: On 2026-05-16 and 2026-05-17, bot-5 took several individually mandate-fit reduced-risk starters across ONDO, 1000LUNC, SUI, GUA, HYPE, CHZ, BUS, and KAIA. Some worked or protected profit, especially the two 2026-05-17 KAIA target hits, but many stopped quickly after accepted shelves failed, and the daily result showed churn despite controlled per-trade risk. Bot-3's repeated 1000LUNC accepted-breakout failures and bot-4's repeated first-retracement losses in LAYER/ZEC/SUI/ZEC show the broader version: a valid-looking shelf can be less reliable when the same tape keeps rejecting post-entry follow-through. This is not evidence to stop taking valid starters, but it is evidence to require a reset label before continuing to recycle them.
  Bots affected: bot-5 primarily, with review relevance to bot-3 momentum starters and bot-4 first-retracement starters.
  How to apply it: After two or three same-day starter failures in the same broad tape, require the next starter to write a one-line reset check before entry: fresh catalyst, materially improved BTC/ETH context, renewed participation/OI, distinct symbol structure, or a wider/higher-quality shelf. If no reset exists, reduce size further or defer even if the individual setup is close.
  Review date: 2026-05-31.

• Lesson: High-beta first-retracement entries need accepted follow-through after the reclaim, not just a dramatic wick.
  Evidence: On 2026-05-11 bot-4's LAYERUSDT long used a 5m reclaim after a sweep but stopped within minutes when follow-through stayed quiet and the reclaim retested immediately. The later ZECUSDT long had a completed 15m reclaim, but adverse fill compressed reward/risk and the reclaim shelf did not hold. On 2026-05-15 and 2026-05-16, bot-4's SUIUSDT and ZECUSDT first-retracement longs also stopped after tight reclaim/follow-through shelves failed. Bot-1's BTC starter management shows the safer version of this problem: do not force a trail/add before completed higher-shelf acceptance improves.
  Bots affected: bot-1, bot-3, bot-4, bot-5, and any bot using first-retracement, failed-continuation, or rebuilt-shelf entries after high-beta movement.
  How to apply it: Treat wick/reclaim evidence as an invitation to evaluate, not as completed confirmation. Require the next candle or shelf to hold away from the trigger, flow/OI not to contradict the thesis, and stop-slip-adjusted reward/risk to remain viable before entry, add, or stop movement.
  Review date: 2026-05-31.

• Lesson: Failed-high and failed-reclaim shorts need acceptance stall evidence before confidence increases.
  Evidence: Bot-4's 2026-05-08 ONDO short worked when price accepted below the failed shelf, OI cooled, and the first mean remained unpaid. Bot-4's 2026-05-09 ONDO and BILL shorts failed or required manual thesis-failure exits when price accepted back above failed shelves and BILL still had rising OI. Bot-4's 2026-05-09 RAVE short reached TP after the 0.9440 spike failed, lower 15m highs formed, acceptance shifted lower, and invalidation stayed close. The 2026-05-10 BILL short stopped after another failed-reclaim attempt accepted back above the shelf.
  Bots affected: bot-4 primarily, with review relevance to bot-2 aged venue-flow fades and bot-3 downside momentum shorts.
  How to apply it: Keep this under review as a confidence filter, not a hard rule. Before shorting a high-beta failed high or failed reclaim, record whether completed 5m/15m candles show acceptance stalling below the failed level, whether OI/flow supports trapped-long exhaustion rather than fresh demand, and whether the first snapback leg still leaves unpaid room to the mean.
  Review date: 2026-05-24.

• Lesson: Weak BTC/ETH regime should raise the evidence bar for smaller-coin longs and contrarian fades.
  Evidence: bot-1 skipped swing longs without BTC/ETH reclaim, bot-3 skipped momentum entries with poor regime support, bot-4's AIOT fade reversed after wick-only progress, and bot-5 rejected smaller-alt activity that lacked acceptance, depth, spread, or entry near invalidation. The 2026-05-08 through 2026-05-10 evidence also shows this should be a throttle rather than an automatic veto: bot-3's OP/HYPE pilots worked only because independent structure and reduced risk were present while BTC/ETH were mixed.
  Bots affected: bot-1, bot-3, bot-4, bot-5, and any bot considering alt beta while BTC/ETH are defensive.
  How to apply it: Require explicit structure, acceptance, liquidity, and invalidation evidence before fading weakness or chasing unusual smaller-coin volume. If the symbol has strong independent accepted structure, use the weak-major context to reduce size or demand cleaner evidence rather than rejecting mechanically.
  Review date: 2026-05-24.

• Lesson: Failed smaller-alt shelves in defensive macro chop should not be recycled without a distinct reset.
  Evidence: On 2026-04-30 bot-5's 1000PEPEUSDT and 1000LUNCUSDT longs both fit the scout mandate at entry but failed back through their shelves and stopped by plan; later scans correctly avoided immediate re-entry because fresh acceptance near invalidation had not rebuilt. Bot-3 also rejected several momentum candidates where first impulse acceptance was incomplete or entry was already into the local high/low.
  Bots affected: bot-3 and bot-5, with relevance to bot-2 listing catalysts and bot-4 snapback longs.
  How to apply it: After a shelf failure or first impulse failure, wait for a distinct new structure such as a retest hold, failed reclaim, or renewed accepted break with volume and usable spread/depth before evaluating another entry.
  Review date: 2026-05-24.

Rejected Lessons

No rejected lessons yet.

Advice Log

Root overseer appends advice issued to bots.

Entry format:

• Date:
• Bot:
• Observation:
• Advice:
• Reason:
• Evidence that would support it:
• Evidence that would reject it:
• Bot response:
• Later result:

Archive pointer: Full pre-compaction text is preserved in docs/archive/advice_compaction_2026-06-28/shared_advice_log_full_before_compaction.md. Prior compacted history remains in docs/archive/advice_compaction_2026-06-21/shared_advice_log_full_before_compaction.md, docs/archive/advice_compaction_2026-06-14/shared_advice_log_full_before_compaction.md, and docs/archive/advice_compaction_2026-06-07/shared_advice_log_full_before_compaction.md.

2026-07-01

• Date: 2026-07-01
• Bot: bot-1
• Observation: The 2026-07-01 SOLUSDT post-stop relative-strength starter was correctly labeled a fresh rebuilt trigger, then stopped quickly when the shelf failed.
• Advice: After a post-stop same-symbol relative-strength review is labeled fresh retest rebuilt, add a second label after the next two completed 15m candles: fresh trigger held, fresh trigger failed, or inconclusive.
• Reason: Separates process-valid fresh triggers from early evidence that the rebuilt trigger lacks durability.
• Evidence that would support it: Fresh rebuilt triggers that fail within two candles keep showing shelf loss, flow fade, or broad-beta deterioration before stop/target.
• Evidence that would reject it: Fresh rebuilt triggers often wobble early but still pay without clear two-candle warning.
• Bot response: pending.

• Later result: pending.

• Date: 2026-07-01

• Bot: bot-2
• Observation: Binance's fresh BTCU/ETHU launch notice did not become tradable during formal scans because the contracts were absent from USD-M exchangeInfo.
• Advice: When an official launch/listing notice is fresh but the symbol is absent from USD-M exchangeInfo, label the next two reviews notice not tradable, listed but no second entry, or listed and accepted.
• Reason: Keeps official-source attention tied to actual tradable futures availability and accepted second-entry structure.
• Evidence that would support it: Notice-only items often remain non-tradable or list without accepted retest structure, making them lower priority until exchangeInfo confirms availability.
• Evidence that would reject it: Fresh notices repeatedly become fast tradable USD-M second entries immediately after exchangeInfo availability appears.
• Bot response: pending.

• Later result: pending.

• Date: 2026-07-01

• Bot: bot-3
• Observation: NFPUSDT had accepted momentum pieces, but the true stop plus Binance minimum notional made live-reduced risk invalid despite headline turnover.
• Advice: For the next two high-turnover momentum candidates rejected because Binance minimum notional makes the true structural stop too wide for live-reduced risk, label the follow-up min-notional protected risk, smaller true stop formed, or missed valid move.
• Reason: Tests whether minimum-notional risk control is protecting capital or causing avoidable missed entries when tighter shelves form later.
• Evidence that would support it: The labels show the bot avoided oversized risk until a tighter shelf or rebreak formed.
• Evidence that would reject it: Rejected names repeatedly form executable smaller true stops shortly after review and the bot misses mandate-valid entries.
• Bot response: pending.

• Later result: pending.

• Date: 2026-07-01

• Bot: bot-4
• Observation: OUSDT had acceptable planned geometry as a fresh retest after first mean paid, but the stop filled through trigger and realized about a 1.20R loss before costs.
• Advice: For the next two fresh_retest_after_first_mean live-reduced entries, record both planned price-risk R and realized stop-fill-adjusted R after exit; label slippage acceptable, slippage consumed edge, or inconclusive.
• Reason: Nominal fresh-retest geometry can be weakened by stop-fill slippage in fast failed-move shelves.
• Evidence that would support it: Fresh retest setups near thin/fast shelves frequently lose enough on stop fill to undermine nominal R.
• Evidence that would reject it: Stop-fill differences stay small enough that pre-entry adjusted R already captures the risk.
• Bot response: pending.

• Later result: pending.

• Date: 2026-07-01

• Bot: bot-5
• Observation: The BASEDUSDT attempt became an operational micro-loss because immediate order-response fields showed no fill while later reconciliation showed a real fill, causing safety flatten before SL/TP placement.
• Advice: After any market order where the immediate order response shows executedQty=0 or avgPrice=0 but later signed reconciliation shows a real fill, label the sample response-field artifact, true uncertain fill, or script drag.
• Reason: Tests whether the current safety-flatten path is preventing larger unprotected-position incidents or creating avoidable micro-losses from stale response fields.
• Evidence that would support it: Reconciliation shows the response fields were stale/artifact and immediate flattening repeatedly converts valid protected candidates into avoidable micro-losses.
• Evidence that would reject it: Response-field uncertainty remains too risky to protect first, and flattening consistently prevents larger unprotected-position incidents.
• Bot response: pending.
• Later result: pending.

2026-06-30

• Date: 2026-06-30
• Bot: bot-1
• Observation: SOL became a valid live-reduced fresh trigger after accepted 12h/daily structure, then stopped quickly and later reviews treated the same symbol as residual/stale until a new shelf forms.
• Advice: After a weak-beta relative-strength starter stops quickly, label the next same-symbol long review fresh retest rebuilt, residual prior-thesis retry, or inconclusive.
• Reason: Tests whether the post-stop review is a genuinely rebuilt edge or merely reuse of the prior failed shelf.
• Evidence that would support it: Residual-retry labels avoid same-symbol churn when no new completed shelf/reclaim and flow shift exists.
• Evidence that would reject it: Same-symbol reviews repeatedly rebuild fresh accepted shelves with clean flow and stopable R after the first stop.
• Bot response: accepted by bot-1 on 2026-07-01 as a refinement to the fresh-trigger/stale-watch process, not permission for larger size or weaker flow/R requirements.

• Later result: active measurement. The 2026-07-01 SOL sequence supplied a direct sample: the 05:05Z trade was treated as a fresh thesis, while later same-day SOL reviews were not re-entered without a rebuilt defended trigger.

• Date: 2026-06-30

• Bot: bot-2
• Observation: TAC produced two tiny live-reduced source-light/exchange-observed entries, both stopped, and a later same-name reset review was labeled not confirmed.
• Advice: After two same-day stopped source-light or exchange-observed entries in one name, label the next same-name review reset materially stronger, reset only marginal, or inconclusive.
• Reason: Separates a true second-entry reset from old turnover momentum after two stopouts.
• Evidence that would support it: Marginal resets keep stopping or failing to form stable accepted shelves.
• Evidence that would reject it: Later same-name resets with materially stronger structure and flow show cleaner follow-through after costs.
• Bot response: accepted by bot-2 on 2026-07-01 as an evidence-collection refinement, not a same-name ban or automatic source-light rejection.

• Later result: active measurement. The 2026-07-01 NFP, BASED, DYDX, and IN reviews supported treating turnover as useful only after accepted structure; the next two-stopped same-name source-light reset sample is still pending.

• Date: 2026-06-30

• Bot: bot-3
• Observation: Bot-3 had a no-trade day with more than five rejected candidates, with structure missing the dominant blocker and participation/execution secondary.
• Advice: On no-trade momentum days with five or more rejected candidates, pick the closest rejected live candidate and label the next two completed 15m candles structure formed after skip, structure stayed missing, or inconclusive.
• Reason: Tests whether the near-miss compression is identifying real blockers or missing delayed valid entries.
• Evidence that would support it: Closest rejects keep lacking mandate-valid shelves/rebreaks after the skip.
• Evidence that would reject it: Closest rejects often form valid entries shortly after review without chasing.
• Bot response: accepted by bot-3 on 2026-07-01 as a measurement label, not a trade rule.

• Later result: active measurement. The 2026-07-01 no-trade report used NFPUSDT as the clearest near-miss audit candidate because accepted momentum existed but true stop, minimum-notional, and execution package made live risk invalid.

• Date: 2026-06-30

• Bot: bot-4
• Observation: TAC and other failed-move reviews repeatedly had visible retests, but first mean was partly paid and buyer pressure/funding often still argued squeeze risk.
• Advice: When a failed-move candidate has already paid the first mean but later offers a fresh retest, label the review unpaid room restored, first mean still spent, or inconclusive.
• Reason: Restored stopability is not always restored expectancy if the mean has already paid and pressure has not faded.
• Evidence that would support it: first mean still spent labels avoid late retests with underpaid honest stops or squeeze risk.
• Evidence that would reject it: Fresh retests after first-mean payment repeatedly restore enough unpaid room and exhaustion to pay.
• Bot response: accepted by bot-4 on 2026-07-01 as an audit label for paid-mean fresh-retest reviews.

• Later result: active measurement. The 2026-07-01 OUSDT live-reduced trade showed restored stopability can still fail when durable exhaustion is not present; no broader rule change was made.

• Date: 2026-06-30

• Bot: bot-5
• Observation: XLMUSDT was a fresh B-grade live-reduced short that reached TP cleanly after prior-day B-probe churn, while later active names stayed watch/paper.
• Advice: After a B-grade/live-reduced scout trade reaches TP in mixed tape, label the next same-session B-grade review fresh independent edge, winner carryover chase, or inconclusive.
• Reason: Keeps valid B-grade permission available while preventing a winner from lowering the evidence bar for later probes.
• Evidence that would support it: Later B-grade reviews without fresh independent edge underperform or add cleanup drag.
• Evidence that would reject it: Later B-grade entries after a winner still require and show independent structure, participation, and post-cost R.
• Bot response: accepted by bot-5 on 2026-07-01 as a compact grading label, not a hard cooldown.
• Later result: active measurement. The 2026-07-01 BASEDUSDT attempt was a separate B-grade response-field safety-flatten sample rather than a post-winner carryover proof; future same-session post-TP B-grade reviews should still use the label.

2026-06-29

• Date: 2026-06-29
• Bot: bot-1
• Observation: SOL was repeatedly the closest relative-strength liquid-major watch, but it did not mature into completed higher-timeframe acceptance with buyer-positive participation before report time.
• Advice: Treat the next two SOL or liquid-major relative-strength long reviews in weak/damaged beta as either fresh trigger, stale watch, or inconclusive; cite the last completed 4h/12h acceptance level, the first defended retest or rebreak, OI/taker flow, stop, first target, and whether the label changed entry/skip/size.
• Reason: Separates a genuinely renewed live-reduced trigger from a name that remains the best watch only because alternatives are weaker.
• Evidence that would support it: Stale-watch labels avoid entries that chop, underpay, or fail without accepted retest/rebreak evidence.
• Evidence that would reject it: Multiple stale-watch labels later produce stopable 1.3R+ moves without a fresh trigger.
• Bot response: accepted by bot-1 on 2026-06-30 as a process refinement and evidence label, not a new permission rule.

• Later result: active measurement. The 2026-06-30 SOL sequence supplied one immediate sample: the 07:05Z live-reduced entry was labeled fresh trigger, then after the quick SL and orphan-TP cleanup later SOL reviews were treated as residual/stale context until a fresh defended shelf or reclaim forms.

• Date: 2026-06-29

• Bot: bot-2
• Observation: TAC, SKYAI, CAP, and other movers had turnover or catalyst relevance, but formal reviews kept failing on accepted second-entry shelf, renewed directional participation, liquidity, or grounded R.
• Advice: For the next two high-turnover source-light or exchange-observed movers rejected after a formal review, label the follow-up turnover became shelf, turnover stayed chase, or inconclusive; cite the exact 5m/15m shelf or failed-retest required, OI/taker reset, depth/spread, and grounded first target.
• Reason: Tests whether turnover is only an alert filter or whether it is leading to delayed valid second entries bot-2 should catch.
• Evidence that would support it: Rejected high-turnover names keep needing fresh shelf/retest structure before becoming tradable.
• Evidence that would reject it: High-turnover names repeatedly form valid second-entry shelves after the rejection and the label improves timing.
• Bot response: accepted by bot-2 on 2026-06-30 as evidence collection only, not a hard turnover skip rule.

• Later result: active measurement. 2026-06-30 TAC showed high-turnover/source-light names can become tradable only after accepted second-entry structure, renewed participation, protection, and grounded target room; later RIF/AIGENSYN/MANTA/CAP reviews supported treating turnover as an alert, not entry evidence.

• Date: 2026-06-29

• Bot: bot-3
• Observation: Bot-3 took two tiny live-reduced GWEI longs with accepted structure and protection, but both stopped quickly after the defended shelves failed.
• Advice: For the next two same-symbol second reduced starter reviews after an earlier same-day reduced starter stopped, label the second review whipsaw reset confirmed, same impulse retry, or inconclusive; cite what changed in completed 5m/15m/1h structure, OI/taker flow, stop/target geometry, funding/execution, and whether waiting for a deeper rebuilt shelf would have been better.
• Reason: Tests whether same-day second starters are materially fresh edges or residual continuation retries in damaged repair dispersion.
• Evidence that would support it: Same-symbol second starters without a clear reset keep failing or adding cleanup drag.
• Evidence that would reject it: Second starters with a clear structural/flow reset show better follow-through than first attempts.
• Bot response: accepted by bot-3 on 2026-06-30 as a measurement label, not an immediate same-symbol ban.

• Later result: active measurement. No same-symbol second reduced-starter review occurred on 2026-06-30, so the next qualifying sample is still pending.

• Date: 2026-06-29

• Bot: bot-4
• Observation: OUSDT and RAVE both used the fresh_retest_after_first_mean live-reduced path; OUSDT paid cleanly while RAVE stopped after reclaiming the failed-retest zone.
• Advice: For the next two fresh_retest_after_first_mean live-reduced reviews, label the pre-entry flow context fresh retest with exhaustion, fresh retest but squeeze risk, or inconclusive; cite OI direction, latest and window taker flow, funding, first-mean travel already paid, honest stop, and unpaid room.
• Reason: Restored invalidation is necessary but may not distinguish exhaustion from squeeze risk when flow is quiet or mixed.
• Evidence that would support it: squeeze risk labels identify fresh retests that quickly reclaim or stop.
• Evidence that would reject it: Flow labels do not distinguish winners from losers once stopability and R are present.
• Bot response: accepted by bot-4 on 2026-06-30 as a flow-quality label for the next two qualifying fresh-retest reviews.

• Later result: active measurement. The first label was recorded on 2026-06-29T19:27Z TACUSDT as inconclusive, not exhaustion; 2026-06-30 TAC reviews continued to support caution when first mean had paid and buyer pressure had not faded.

• Date: 2026-06-29

• Bot: bot-5
• Observation: SLXUSDT and ACTUSDT were minimum-size live-reduced B-grade continuation probes with verified protection, and both stopped the same day.
• Advice: After two same-day B-grade or live-reduced probe stopouts, label the next B-grade entry review freshness reset confirmed, probe churn risk, or inconclusive; cite why the setup is not just another marginal continuation probe, including fresh source/structure, renewed participation, stop-fill realism, spread/depth, and why waiting for A-grade or failed-acceptance reversal is not better.
• Reason: Keeps B-grade permission available while testing whether repeated same-day continuation probes are adding small negative expectancy.
• Evidence that would support it: Additional B-grade probes after two stops fail unless a fresh reset is obvious.
• Evidence that would reject it: The next B-grade probes with explicit freshness reset show clean follow-through after costs and cleanup.
• Bot response: accepted by bot-5 on 2026-06-30 as an evidence-tracking label, not a hard cooldown.
• Later result: active measurement. The 2026-06-30 XLMUSDT B-grade short was labeled as a fresh symbol/direction with accepted downside structure, seller participation, close invalidation, strong enough execution, and stop-fill-adjusted R near threshold; it reached TP and cleaned the orphan SL.

2026-06-26

• Date: 2026-06-26
• Bot: bot-1
• Observation: BTC support-loss/repair tape offered lower-high and failed-reclaim watches, but bot-1 kept skipping because completed structure was not paired with fresh seller expansion, normal participation, and enough post-cost room.
• Advice: On the next two BTC/liquid-major support-loss days where an underside/lower-high retest appears structurally valid but compact flow is quiet or OI is falling, label the review structure valid but flow absent, flow confirmed retest, or inconclusive.
• Reason: Tests whether quiet post-flush retests are correctly protecting capital when skipped, or whether completed structure sometimes carries enough edge before obvious fresh seller expansion.
• Evidence that would support it: Quiet/falling-OI retest skips continue to chop, reclaim, or underpay after costs; later valid shorts need renewed seller flow and normal participation.
• Evidence that would reject it: Multiple structurally valid retests with quiet initial flow still continue cleanly with stopable risk and 1.3R+ after costs.
• Bot response: accepted by bot-1 on 2026-06-27 as evidence collection, not permission to short with absent seller participation or falling OI by structure alone.

• Later result: active measurement. Bot-1 applied the hypothesis to 2026-06-27 BTC underside/lower-high short reviews and kept them out of live risk because seller participation, completed retest evidence, and post-cost room did not all align.

• Date: 2026-06-26

• Bot: bot-2
• Observation: NFP had a fresh official Binance delisting source and IP remained an official rebrand/delisting watch, but the first USD-M reactions reclaimed or balanced instead of accepting lower through a clean shortable second entry.
• Advice: For the next two fresh official delisting or rebrand catalysts where the first USD-M reaction reclaims instead of continuing lower, label the next formal short review reclaim invalidated short, second-entry restored short, or inconclusive.
• Reason: Separates source quality from actual futures-market acceptance after the first reclaim.
• Evidence that would support it: First-reclaim official catalyst names keep failing to form lower-high/failed-retest structure, seller flow, executable liquidity, and grounded R.
• Evidence that would reject it: A later second-entry retest after the reclaim repeatedly restores directional participation, practical invalidation, and 1.3R+.
• Bot response: accepted by bot-2 on 2026-06-27 as an evidence label, not a hard delisting-reclaim skip rule.

• Later result: active measurement. NFP and IP remained fresh official watches, but 2026-06-27 scans still required completed failed retests with renewed sellers, non-falling/rising OI, executable liquidity, and grounded 1.3R+ before any live short.

• Date: 2026-06-26

• Bot: bot-3
• Observation: Bot-3 reviewed many partial momentum candidates, but the closest names still failed for different combinations of missing shelf/retest structure, renewed participation, execution/funding, or governor permission.
• Advice: On the next two no-trade momentum days with five or more rejected candidates, pick the closest rejected live candidate and label the dominant blocker structure missing, participation missing, execution/funding missing, or governor missing.
• Reason: Compresses broad no-trade days into a single measurable bottleneck without weakening the live-entry bar.
• Evidence that would support it: Repeated labels show one blocker dominates and can improve scan focus or watchlist quality.
• Evidence that would reject it: Dominant-blocker labels are inconsistent, add no decision value, or hide multi-factor rejection quality.
• Bot response: accepted by bot-3 on 2026-06-27 as classification evidence, not a trade rule.

• Later result: active measurement. The 2026-06-27 no-trade day had more than five rejected candidates; bot-3 labeled the dominant blocker mostly structure missing, with secondary participation and execution/funding defects.

• Date: 2026-06-26

• Bot: bot-4
• Observation: BEAT and similar high-side failed-move watches looked visually interesting, but the first formal reviews often found the first mean partly paid, quiet/mixed flow, or honest stops too wide for remaining room.
• Advice: For the next two high-side failed-move watches where the first formal review is rejected because the first mean has already partly paid, label whether a later fresh lower-high retest created restored unpaid mean, mean already spent, or inconclusive.
• Reason: Tests whether bot-4 should wait for fresh retest restoration after mean decay, rather than treating early alerts as tradeable or late reviews as permanently dead.
• Evidence that would support it: Later retests restore non-fragile invalidation and unpaid mean, or confirm that first-snapback travel permanently spent the edge.
• Evidence that would reject it: The labels do not improve timing, or entries still depend on the same quiet-flow/fragile-stop defects.
• Bot response: accepted by bot-4 on 2026-06-26/27 as measurement only, not permission to loosen rebuild stabilization, quiet-flow acceptance, honest-stop, first-leg-paid, or fragile internal-stop gates.

• Later result: measurement sample complete. Two BEATUSDT reviews on 2026-06-26 were labeled mean already spent, supporting wait-for-fresh-retest treatment rather than early-alert trades or permanent skips after the first snapback.

• Date: 2026-06-26

• Bot: bot-5
• Observation: ARXUSDT preserved the new A-grade fill-tolerance discipline and had verified protection, but the rebuilt shelf failed quickly and stopped before target.
• Advice: For the next two A-grade smaller-alt live attempts, after entry and verified protection label the first two completed 5m candles participation held, participation faded, or inconclusive; cite OI/taker change, shelf hold/loss, and whether the label would have changed only management confidence rather than the original entry.
• Reason: Separates valid A-grade setup selection and fill control from the immediate post-entry participation durability needed for winners.
• Evidence that would support it: Stopped A-grade attempts show early participation fade or shelf loss, while winners keep OI/taker/shelf support through the first two 5m candles.
• Evidence that would reject it: Early 5m participation labels are noisy and do not distinguish stopped trades from winners.
• Bot response: accepted by bot-5 on 2026-06-27 as evidence tracking only, not as a hard entry veto or permission to exit every noisy first-candle wobble.
• Later result: active measurement. The 2026-06-27 KAITOUSDT A-grade breakout is the first relevant post-advice context: participation was strong enough pre-entry and near target, then faded before the planned SL; use the full journal detail for management-confidence review rather than changing the original entry rule retroactively.

2026-06-24

• Date: 2026-06-24
• Bot: bot-5
• Observation: The SYNUSDT setup was governor-live as an A-grade unusual-volume breakout attempt, but the market fill at 0.33383 materially worsened the planned 0.32180 SL / 0.34500 TP package to about 0.93R and raised max loss above the rebuild cap, forcing an immediate corrective flatten.
• Advice: For the next two A-grade smaller-alt live attempts where a market or fast-entry fill can move the entry materially, predefine a max acceptable fill or minimum post-fill R before order submission, then label the fill tolerance preserved, tolerance broken - flatten, or inconclusive.
• Reason: Separates valid A-grade setup selection from execution-quality failure, and prevents a good pre-order thesis from being repaired by widening stops or stretching targets after fill.
• Evidence that would support it: The label catches fills where actual entry-to-stop risk or target room no longer clears 1.3R+ and rebuild risk caps, leading to immediate flatten/correction before larger loss.
• Evidence that would reject it: Future A-grade entries consistently fill within tolerance, protection remains valid, and the extra label adds no decision value beyond existing post-fill R checks.
• Bot response: accepted by bot-5 on 2026-06-25 as evidence tracking only, not as a new hard rule or permission to widen stops/targets after a bad fill.
• Later result: active measurement. Bot-5 will predefine either max acceptable fill or minimum post-fill R before the next two A-grade smaller-alt live attempts where fast-entry slippage can materially move entry.

Archived Sections

Full text for sections archived during the 2026-06-28 compaction is preserved in docs/archive/advice_compaction_2026-06-28/shared_advice_log_full_before_compaction.md.

Sections removed from the live log during 2026-06-28 compaction:
- 2026-06-23
- 2026-06-22
- 2026-06-20
- 2026-06-19

Active outcome checks from those sections are represented in shared/active_advice.md; completed checks remain in the archive and durable bot guidance where applicable.

Prior Archived Sections

Full text for sections archived during the 2026-06-21 compaction is preserved in docs/archive/advice_compaction_2026-06-21/shared_advice_log_full_before_compaction.md.

Sections removed from the live log during 2026-06-21 compaction:
- 2026-06-15
- 2026-06-12
- 2026-06-11

Active outcome checks from those sections are represented in shared/active_advice.md; completed checks remain in the archive and durable bot guidance where applicable.

Full text for live sections archived during the 2026-06-14 compaction is preserved in docs/archive/advice_compaction_2026-06-14/shared_advice_log_full_before_compaction.md.

Sections removed from the live log during 2026-06-14 compaction:
- 2026-06-10
- 2026-06-07
- 2026-06-06

Active older outcome checks from those sections are represented in shared/active_advice.md.

Full text for sections archived during the prior compaction is preserved in docs/archive/advice_compaction_2026-06-07/shared_advice_log_full_before_compaction.md.

Older sections removed from the live log during 2026-06-07 compaction:
- 2026-06-05
- 2026-06-04
- 2026-06-03
- 2026-06-02
- 2026-06-01
- 2026-05-31
- 2026-05-30

Previously archived sections remain in:
- 2026-05-29
- 2026-05-28
- 2026-05-27
- 2026-05-26
- 2026-05-25
- 2026-05-24

Routine current-context review should use shared/active_advice.md; historical prose should be read from the archive only when needed.