You cannot select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
465e248cb4
Fix EarningsRunup trailing config that was previously captured in
Candidate.features only. Add activation-gated pct trailing to
update_trailing_stop with entry-relative giveback semantics
("lock in peak − giveback% of entry"). Plumb through
Candidate → ExecutionConfig → simulator. Default None preserves
legacy pct_X behavior — no impact on engines that don't opt in.
Add earnings_runup_poc_v2_tuned config: per_trade_risk_pct 0.65→0.30,
max_positions 30→8, max_positions_per_sector 30→4, macro_vix_max=30,
trailing_warmup_days 7→0 (activation gate replaces warmup).
v2 backtest (4y, 121 trades) vs v1 baseline (119 trades):
Total return: +37.27% → +254.69%
Max drawdown: 53.23% → 23.97%
SQS: 45.2 → 69.7
Robustness: 50.1 → 100.0
Risk score: 23.5 → 35.0
STOP r-mult: −0.41 → +0.65 (trailing-locked winners)
Promotion thresholds met: MDD < 25%, return preserved, SQS > 55.
Recommend integration as PEAD sleeve adjunct (orthogonal entry timing:
pre-print attention runup vs post-print drift) rather than standalone.
7 new pct-trailing tests: activation gate, ratchet, no-ratchet-down,
reversal-stop, legacy compat, end-to-end engine wiring. 28 passed.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
|
3 months ago | |
|---|---|---|
| .. | ||
| fixtures | 5 months ago | |
| integration | 4 months ago | |
| replay | 5 months ago | |
| unit | 3 months ago | |
| __init__.py | 5 months ago | |
| conftest.py | 5 months ago | |