Trade Audit Replay

Side-by-side classifier verdicts · Generated 2026-05-16T11:16:38-04:00 ET · Source: synthetic examples (built into audit_trades.py)

What every classifier (current production v1, Pine-aligned v2, schema-driven classify_v2, live evaluator) says about the same trade. Use this to validate the cutover BEFORE wiring classify_v2.py into gen_scorecard.py or swapping regime_classifier.py → regime_classifier_v2.py.

Summary

5
Trades
1
v1<->v2 regime diverges
0
v1<->v2 playbook diverges
1
classify_v2 OFF-BOOK
1
v1 TREND tags
2
v2 TREND tags

Per-trade verdicts

If the column "v2 Pine" disagrees with "v1 current", that's a real divergence the cutover will introduce. Inspect each before flipping.

Trade Side P&L v1 (current, deg) v2 Pine (σ/bar) v2 legacy (deg) classify_v2 strategy_id evaluate grade
SOXL
synth-soxl-1
SHORT
2026-05-13T08:30:36-04:00
$8.62 regime: ROTATION
playbook: OFF_BOOK
slope=0.0° rate=5.0%
regime: ROTATION
playbook: OFF_BOOK
threshold: 0.05
regime: ROTATION
playbook: OFF_BOOK
vwap-pull-in-short B
action: scale_at_band
QQQ
synth-qqq-1
LONG
2026-05-13T10:15:00-04:00
$158.00 regime: ROTATION
playbook: VWAP_PULL_IN
slope=0.0° rate=4.0%
regime: ROTATION
playbook: VWAP_PULL_IN
threshold: 0.05
regime: ROTATION
playbook: VWAP_PULL_IN
vwap-pull-in-long C
action: scale_at_band
SNOW
synth-snow-1
LONG
2026-05-13T11:30:00-04:00
$575.00 regime: TREND
playbook: TREND_TRADE
slope=0.1° rate=35.0%
regime: TREND
playbook: TREND_TRADE
threshold: 0.05
regime: TREND
playbook: TREND_TRADE
trend-trade-long OFF-BOOK
action: exit
RMSG
synth-rmsg-1
SHORT
2026-05-13T10:42:00-04:00
$65.00 regime: ROTATION
playbook: OFF_BOOK
slope=0.1° rate=25.0% (borderline)
regime: TREND
playbook: OFF_BOOK
threshold: 0.05
v1=ROTATION -> v2=TREND
regime: ROTATION
playbook: OFF_BOOK
backside-scalp-short C
action: alert
AAPL
synth-chase-1
LONG
2026-05-13T13:15:00-04:00
$-70.00 regime: ROTATION
playbook: CHASE_EXTENDED
slope=0.0° rate=4.0%
regime: ROTATION
playbook: CHASE_EXTENDED
threshold: 0.05
regime: ROTATION
playbook: CHASE_EXTENDED
off-book —
action: —

How to read this

  1. v1 current = exactly what regime_classifier.py says today. Used by gen_scorecard.py nightly. SLOPE_TREND_THRESHOLD = 3.0 (units claimed as degrees).
  2. v2 Pine = regime_classifier_v2.py with slope_units="sigma_per_bar" threshold 0.05, matching TraderDale_VWAP.pine. If this is your TRUE intent, swap v1 for v2.
  3. v2 legacy degrees = v2 with slope_units="degrees" threshold 3.0 — same logic as v1, included as a sanity check. Should match v1 exactly.
  4. classify_v2 strategy_id = single canonical strategy_id from /system/strategies.json. No dual labels possible.
  5. evaluate grade = live grade (A/B/C/F/OFF-BOOK) from evaluate.py if classify_v2 found a strategy. Quality + validation + invalidation all in one number.

Synthetic trades use illustrative band data. To audit your REAL trades, export per-trade band data to a CSV and pass --csv path. Schema in audit_trades.py docstring.

Re-run: py "Z:\setup_daily_reporting\reports\audit_trades.py" [--csv path]. Source: /system/trade-audit-replay.json. Back to strategy-system.html.