Trading Research Share Report
Batch experiment_five_20260312_breakout_robustness · Last updated 2026-03-12T10:19:17Z
Executive summary
- This batch is an early research readout rather than a final strategy verdict.
- The least bad average segment was Breakout on 15m data at 1.35%.
- The weakest average segment was Breakout on 1h data at -3.52%.
- The best run that actually traded was Breakout · SOL · 15m · range 36 / hold 6 with total return 22.08%.
- Interpret these as research signals about direction and quality, not as production-ready trading systems.
Best trading run in this batch: Breakout · SOL · 15m · range 36 / hold 6
Assets covered
3
ETH, SOL, BTC
Total experiment runs
360
Research batch
Best observed return
22.08%
Including all runs
Scope and assumptions
Exchange: binance
Fee: n/a bps
Slippage: n/a bps
This report summarizes research tests on BTC, ETH, and SOL using 15m, 1h data, with 4h available as higher-timeframe context. The goal is not to prove a strategy works, but to learn which directions deserve deeper study.
Average return by strategy and timeframe
Current research conclusions
This batch was used to decide whether more selective signals and larger timeframes improved the baseline results.
- 5m baseline strategies currently look weak under the present cost assumptions.
- 15m results are generally less bad than 5m results in the current baseline suite.
- Breakout currently appears less bad than momentum and mean reversion in the first pass.
- These are baseline findings, not final conclusions.
Show detailed conclusions and decision trail
- This pass moved away from raw zero-threshold signals and toward more selective entries.
- The strongest average strategy/timeframe bucket was Breakout on 15m at 1.35%.
- The weakest average bucket was Breakout on 1h at -3.52%.
- Across timeframes, 15m held up best on average at 1.35%, while 1h was weakest at -3.52%.
- We now have genuinely positive runs, led by Breakout · SOL · 15m · range 36 / hold 6 at 22.08%. That is a real improvement over the first pass.
- The decision to move forward now comes from selective evidence rather than broad optimism: a few setups are starting to survive trading costs.
- The next direction should emphasize the most promising lanes rather than another broad sweep: 1h breakout and higher-threshold mean reversion deserve more attention than raw momentum.
- This decision trail matters because the project is intentionally trying to learn honestly from negative as well as positive evidence.
Top 10 runs
Best-performing runs in this batch, still subject to the current assumptions and sample window.
| Run | Trades | Total Return | Drawdown | Win Rate | Profit Factor |
|---|
| Breakout · SOL · 15m · range 36 / hold 6 | 52 | 22.08% | -6.44% | 53.85% | 2.015 |
| Breakout · SOL · 15m · range 24 / hold 8 | 74 | 21.99% | -8.09% | 51.35% | 1.632 |
| Breakout · SOL · 15m · range 24 / hold 4 | 80 | 20.33% | -6.90% | 48.75% | 1.693 |
| Breakout · SOL · 15m · range 24 / hold 6 | 76 | 19.27% | -8.59% | 50.00% | 1.493 |
| Breakout · SOL · 15m · range 36 / hold 6 | 19 | 17.15% | -1.38% | 57.89% | 4.400 |
| Breakout · SOL · 15m · range 36 / hold 6 | 52 | 16.49% | -7.20% | 51.92% | 1.700 |
| Breakout · SOL · 15m · range 36 / hold 6 | 19 | 15.16% | -1.51% | 52.63% | 3.597 |
| Breakout · SOL · 15m · range 18 / hold 6 | 114 | 14.53% | -10.42% | 47.37% | 1.257 |
| Breakout · SOL · 15m · range 24 / hold 6 | 25 | 14.39% | -3.10% | 48.00% | 2.437 |
| Breakout · SOL · 15m · range 24 / hold 8 | 74 | 14.13% | -9.41% | 51.35% | 1.381 |
Bottom 10 runs
Weakest runs in this batch. These are useful because they show what clearly did not work in this pass.
| Run | Trades | Total Return | Drawdown | Win Rate | Profit Factor |
|---|
| Breakout · SOL · 1h · range 18 / hold 6 | 63 | -28.20% | -28.40% | 33.33% | 0.450 |
| Breakout · SOL · 1h · range 24 / hold 8 | 49 | -25.87% | -25.88% | 34.69% | 0.422 |
| Breakout · BTC · 1h · range 18 / hold 6 | 55 | -22.22% | -25.80% | 27.27% | 0.310 |
| Breakout · ETH · 1h · range 18 / hold 6 | 58 | -21.72% | -25.88% | 34.48% | 0.541 |
| Breakout · SOL · 1h · range 18 / hold 6 | 63 | -20.59% | -23.95% | 39.68% | 0.574 |
| Breakout · SOL · 1h · range 24 / hold 6 | 52 | -20.17% | -20.25% | 34.62% | 0.499 |
| Breakout · SOL · 1h · range 24 / hold 8 | 49 | -19.82% | -20.73% | 38.78% | 0.530 |
| Breakout · ETH · 15m · range 18 / hold 6 | 105 | -19.64% | -25.90% | 28.57% | 0.610 |
| Breakout · SOL · 1h · range 18 / hold 6 | 24 | -17.77% | -21.60% | 25.00% | 0.335 |
| Breakout · SOL · 1h · range 36 / hold 6 | 37 | -17.34% | -19.61% | 40.54% | 0.486 |
How to read this report
This batch evaluates three strategy families across 15m, 1h, with 4h data reserved for context/regime use.
- Data basis / timeframe: the actual experiments in this report are based on 15m, 1h bar data, with 4h available as higher-timeframe context.
- Momentum means buying after a recent upward move and testing whether that move continues.
- Mean Reversion means buying after a sufficiently large adverse move and testing whether price snaps back.
- Breakout means buying when price clears a recent range by more than a minimum amount, on the assumption that the move may be starting a new leg.
- lookback <number> is how many bars backward the strategy looks when measuring the recent move or condition.
- hold <number> is how many bars the position is held after entry before exit in this simplified backtest.
- Trades is the number of completed trades the strategy actually took during the tested period.
- Total Return is the cumulative strategy result across the full tested period, after the current fee and slippage assumptions.
- Drawdown is the worst peak-to-trough decline of the strategy equity during the run.
- Win Rate is the fraction of completed trades that were profitable.
- Profit Factor is gross profit divided by gross loss; above 1 is generally better, below 1 means losses outweighed gains.
- Thresholds were added because the first-pass raw strategies traded too often and were too easily overwhelmed by costs.
- Fees and slippage are included in all runs, so results reflect friction rather than idealized fills.