Signal Accuracy and Backtest (15D)

Did past signals actually lead to price hitting the target or stop?

What does this page tell me?
  • TARGET_HIT - price reached +3.0xATR above entry within 15 days. Win.
  • STOPPED_OUT - price fell 1.5xATR below entry. Loss.
  • OPEN_AT_EXPIRY - neither hit; valued at close on day 15.
  • EV = win_rate x 3.0 + loss_rate x (-1.5). Positive = edge.

Settings: score≥0.45, stop=1.5xATR, target=3.0xATR, last 12 months, max 500 signals.

Signals: 2025-06-30 to 2026-06-10
500
Total Signals
25.4%
Win Rate
127
Target Hit
287
Stopped Out
6.4%
Avg Winner
-3.7%
Avg Loser
-0.36xATR
EV

Win Rate by Regime

Does signal work better in bull markets?

regimeSignalsWin Rate %Avg delta%
SIDEWAYS95.00031.6000.310
BEAR24.00029.2001.030
BULL301.00024.600-0.110
STRONG_BULL69.00021.700-0.390
STRONG_BEAR11.0009.100-2.850

Win Rate by Score Bucket

Rising win rate with score = real predictive power.

score_bucketSignalsWin Rate %Avg delta%
0.75+500.00025.400-0.080

Win Rate by Month

Any seasonal months that consistently outperform?

Download all 500 evaluated signals
Static snapshot · Generated 30 Jun 2026, 08:03 · Source: Nifty 500 PostgreSQL ML Pipeline · ← Back