Add quantile confidence trading strategy#5
Open
lee101 wants to merge 105 commits into
Open
Conversation
…ld trade under the value
lee101
force-pushed
the
main
branch
2 times, most recently
from
198c06f to
4b5c43f
Compare
lee101
added a commit
that referenced
this pull request
Mar 25, 2026
s87 is now #3 overall at +3315% ann (honest=4.70x/180d), beating s70 (+3303%) s83 is #5 at +2527% ann (pool_val=3.75, wr=68.6%) s85 is #8 at +1510% ann All saved to crypto70_champions/. Long queue reordered: s71→s87→s70→s83... crypto15: s78 still champion (+497% ann), 14/77 seeds genuine, at s129. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
lee101
added a commit
that referenced
this pull request
Mar 25, 2026
s98 is a highly robust seed — consistent at both 5bps (+2964% ann) and 8bps (+2918% ann). Ranks #5 in production 5bps leaderboard: s71(+9381%) >> s70(+3487%) > s65(+3257%) > s87(+3074%) > s98(+2964%) Stopped stocks12 s200-350 sweep (0/18 positive, freeing GPU for crypto sweeps). Full 5bps leaderboard now complete for all saved champions. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
lee101
added a commit
that referenced
this pull request
Mar 25, 2026
Seed s240 from the s201-300 sweep is the new production champion: - 5bps: +17,642% ann, Sortino=7.00 — 88% better than previous champion s71 (+9,381%) - 8bps: +40,405% ann, honest_val=18.32x in 180 days - Pool_val=16.99, fully saved to champions dir - Isolated spike: s239=+262%, s241=+448% Also new today: - s507: +8,273% @5bps (new #3) - s520: +6,132% @5bps, Sortino=7.70 (new #8) - s523: +5,919% @5bps, Sortino=7.56 (new #9) - s404: +7,319% @5bps, Sort=6.28 (new #5) - s416: +4,099% @5bps, Sort=6.19 (new #13) - s609: +4,450% @5bps, Sort=5.04 (new #12) 7-9 parallel pool sweeps now covering s1-1000. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
lee101
added a commit
that referenced
this pull request
Mar 25, 2026
…%),s834(+1917%) More major finds: - s452: +8002% ann (VERY ROBUST! 5bps>>8bps) NEW #5 all-time - s747: +2905% ann (s701-800 strong find) - s543: +1636% ann (pool=3.55≈honest=3.53) - s834: +1917% ann (pool=2.04→honest=3.52→5bps=3.40) - s835: +912% ann (VERY ROBUST, 5bps>>8bps) - s832: +1163% ann - s450: +619% ann Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
lee101
added a commit
that referenced
this pull request
Mar 25, 2026
Leaderboard updated (169 evaluated at 5bps): #1 s275: +23,595% (Sortino=9.0, ultra-robust: 5bps > 8bps) #2 s240: +17,642% #3 s434: +10,359% #4 s71: +9,381% #5 s456: +8,802% (new) #6 s507: +8,273% #7 s452: +8,002% (new) Top-10 mean: +10,796% ann | 7/10 ultra-robust (5bps >= 8bps) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
lee101
added a commit
that referenced
this pull request
Apr 13, 2026
s87 is now #3 overall at +3315% ann (honest=4.70x/180d), beating s70 (+3303%) s83 is #5 at +2527% ann (pool_val=3.75, wr=68.6%) s85 is #8 at +1510% ann All saved to crypto70_champions/. Long queue reordered: s71→s87→s70→s83... crypto15: s78 still champion (+497% ann), 14/77 seeds genuine, at s129. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
lee101
added a commit
that referenced
this pull request
Apr 13, 2026
s98 is a highly robust seed — consistent at both 5bps (+2964% ann) and 8bps (+2918% ann). Ranks #5 in production 5bps leaderboard: s71(+9381%) >> s70(+3487%) > s65(+3257%) > s87(+3074%) > s98(+2964%) Stopped stocks12 s200-350 sweep (0/18 positive, freeing GPU for crypto sweeps). Full 5bps leaderboard now complete for all saved champions. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
lee101
added a commit
that referenced
this pull request
Apr 13, 2026
Seed s240 from the s201-300 sweep is the new production champion: - 5bps: +17,642% ann, Sortino=7.00 — 88% better than previous champion s71 (+9,381%) - 8bps: +40,405% ann, honest_val=18.32x in 180 days - Pool_val=16.99, fully saved to champions dir - Isolated spike: s239=+262%, s241=+448% Also new today: - s507: +8,273% @5bps (new #3) - s520: +6,132% @5bps, Sortino=7.70 (new #8) - s523: +5,919% @5bps, Sortino=7.56 (new #9) - s404: +7,319% @5bps, Sort=6.28 (new #5) - s416: +4,099% @5bps, Sort=6.19 (new #13) - s609: +4,450% @5bps, Sort=5.04 (new #12) 7-9 parallel pool sweeps now covering s1-1000. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
lee101
added a commit
that referenced
this pull request
Apr 13, 2026
…%),s834(+1917%) More major finds: - s452: +8002% ann (VERY ROBUST! 5bps>>8bps) NEW #5 all-time - s747: +2905% ann (s701-800 strong find) - s543: +1636% ann (pool=3.55≈honest=3.53) - s834: +1917% ann (pool=2.04→honest=3.52→5bps=3.40) - s835: +912% ann (VERY ROBUST, 5bps>>8bps) - s832: +1163% ann - s450: +619% ann Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
lee101
added a commit
that referenced
this pull request
Apr 13, 2026
Leaderboard updated (169 evaluated at 5bps): #1 s275: +23,595% (Sortino=9.0, ultra-robust: 5bps > 8bps) #2 s240: +17,642% #3 s434: +10,359% #4 s71: +9,381% #5 s456: +8,802% (new) #6 s507: +8,273% #7 s452: +8,002% (new) Top-10 mean: +10,796% ann | 7/10 ultra-robust (5bps >= 8bps) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
lee101
added a commit
that referenced
this pull request
Apr 22, 2026
Two new integration tests close the audit's priority #5 gaps around the death-spiral guard's interactions with other close paths: 1. max_hold force-close + guard refusal: verifies that when max_hold_hours wants to force-close mid-crash and the intraday tolerance blocks the sell, the position persists all the way to terminal and equity is mark-to-market (not cash, which would report 0%). 2. stale_after_bars transition: setting stale=0 forces the wider 500bps overnight tolerance immediately, so a 1% drop (refused under tight=8 default) now lands on the first attempt. Contrasts with the existing refused-through-terminal test to pin the branch selection. _run_episode now accepts stale_after_bars for future parameterization. 10 death-spiral tests pass; 250-test pufferlib_market regression suite unchanged. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
lee101
added a commit
that referenced
this pull request
Apr 22, 2026
Stamps features[t][0][0] = t in a signed MKTD v1 binary, then after k
vec_steps asserts obs[0] == max(0, k - lag). Parameterized over lag in
{1,2,3}. Catches any future regression that leaks current-bar features
into observations — which would reintroduce the lookahead bias that
decision_lag=2 is specifically there to prevent (Chronos2 forecast +
Gemini + order-placement straddles more than one bar in prod).
Closes audit priority #5(a).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
quantile_confidence_strategyfor backtestingTesting
pip install -r dev-requirements.txt(fails: Cannot connect to proxy)pytest -q(fails: command not found)