Skip to content

Add turnover recording, turnover-control analysis, and real-market check #143

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
weich97 merged 1 commit into from
Jun 15, 2026
Merged

Add turnover recording, turnover-control analysis, and real-market check #143

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 10 additions & 0 deletions docs/results/execution_sensitivity_llm/turnover_control.csv
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,10 @@
scenario,turnover_bin,bin_agents,bin_size,mean_turnover,within_bin_tau_e0_e1,full_leaderboard_tau
calm,T1,mean-reversion;signal-weighted;deepseek:deepseek-v4-pro;poe:gemini-3.1-pro,4,10.5,1.0,0.758
calm,T2,naive-momentum;poe:gpt-5.5;random;poe:glm-5,4,18.73,0.667,0.758
calm,T3,poe:claude-opus-4.7;buy-and-hold;minimum-variance;risk-parity,4,20.25,0.333,0.758
high_vol,T1,mean-reversion;signal-weighted;naive-momentum;deepseek:deepseek-v4-pro,4,11.53,0.667,0.242
high_vol,T2,poe:glm-5;poe:gemini-3.1-pro;poe:gpt-5.5;poe:claude-opus-4.7,4,17.48,0.667,0.242
high_vol,T3,random;minimum-variance;buy-and-hold;risk-parity,4,19.52,0.667,0.242
jump_tail,T1,mean-reversion;signal-weighted;naive-momentum;deepseek:deepseek-v4-pro,4,11.43,0.667,0.455
jump_tail,T2,poe:gemini-3.1-pro;poe:glm-5;random;poe:claude-opus-4.7,4,17.85,0.0,0.455
jump_tail,T3,poe:gpt-5.5;minimum-variance;buy-and-hold;risk-parity,4,19.75,1.0,0.455
20 changes: 20 additions & 0 deletions docs/results/execution_sensitivity_llm/turnover_control.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,20 @@
# Turnover-Controlled Ranking Stability

If the E0->E1 reordering were purely a turnover effect, agents of
similar turnover would not reorder. Within turnover terciles (binned by
E1 turnover), the E0-vs-E1 Kendall tau remains low, so the reordering is
not explained by turnover alone.

| Regime | Turnover bin | Mean turnover | Within-bin tau (E0 vs E1) | Full-leaderboard tau |
| --- | --- | ---: | ---: | ---: |
| calm | T1 | 10.5 | 1.0 | 0.758 |
| calm | T2 | 18.73 | 0.667 | 0.758 |
| calm | T3 | 20.25 | 0.333 | 0.758 |
| high_vol | T1 | 11.53 | 0.667 | 0.242 |
| high_vol | T2 | 17.48 | 0.667 | 0.242 |
| high_vol | T3 | 19.52 | 0.667 | 0.242 |
| jump_tail | T1 | 11.43 | 0.667 | 0.455 |
| jump_tail | T2 | 17.85 | 0.0 | 0.455 |
| jump_tail | T3 | 19.75 | 1.0 | 0.455 |

Mean within-bin tau across all regimes and terciles: **0.630** (vs.\ a full-leaderboard tau that is similarly low). The reordering persists within turnover strata.