A from-scratch backtesting engine for an Opening Range Breakout retracement strategy on NQ E-mini futures (NAS100), built in Python.
The strategy trades a retracement into the Opening Range after a confirmed breakout. The retrace to a VP level provides a defined-risk entry point to participate in the continuation of the breakout beyond the original ORB extreme.
Opening Range: 09:30–09:45 ET (15 minutes). Volume Profile (VAH, POC, VAL) is calculated from the ORB bars.
Entry logic:
- Price breaks out beyond the ORB extreme by a minimum threshold
- Confirmation requires a 1m bar close beyond the threshold — no wick entries
- After confirmation, the strategy waits for price to retrace to a VP level inside the range
- Entry fills when price touches the level
Exit logic:
- Target beyond the ORB extreme
- Stop at the ORB extreme
- EOD force-close and entry cutoff applied
Results (2021–2026, filtered, 1% risk/trade):
| Direction | Trades | WR | Expectancy | R/yr |
|---|---|---|---|---|
| Long | 198 | 42.9% | +0.417R | +13.8 |
| Short | 80 | 42.5% | +0.558R | +7.4 |
| Combined | 278 | — | — | +21.2 |
NQ.v.0 continuous contract, 1m OHLCV (Databento)
The project is being extended to cover additional instruments and exchanges.
$100,000 starting equity, 1% risk per trade, compounding:
The backtest engine (backtest/engine.py) is built for correctness and speed:
- Vectorised bar scanning — all entry/exit logic uses NumPy array operations. No
iterrows(). - Day context pre-computation — ORB, VP, breakout flags, and post-ORB bar arrays are built once per day and reused across all variant evaluations (
build_day_context). - Fast timestamp comparison — timestamps converted to int64 nanoseconds for O(log n) searchsorted lookups rather than repeated DatetimeIndex comparisons.
- Correct entry-bar handling — target cannot fire on the same bar as entry; only stop is valid at bar 0 of an entry.
- MFE/MAE tracking — Maximum Favourable/Adverse Excursion recorded per trade in R units for post-hoc analysis.
| Long — Target | Long — Stop |
|---|---|
 |
 |
| Short — Target | Short — Stop |
|---|---|
 |
 |
Full sample set: public_charts/samples/
- Python 3.13
- pandas, numpy, matplotlib
- Data: Databento NQ.v.0 continuous contract (not included — see
data/README.md)
pip install -r requirements.txt
All strategy design, methodology, and analytical decisions are my own. Claude (Anthropic) assisted with Python implementation.