Validate your strategy before risking capital. Use custom simulations to compare the theoretical best outcome with how our model would have performed over the same period.
Why run a custom simulation
Backtesting is the fastest way to turn intuition into evidence. A custom simulation shows you two clear paths for any historical window: the perfect path (the theoretical maximum earnings) and the prediction path (what our model would have selected under your rules). Comparing both builds trust, highlights trade-offs, and helps you tune thresholds and filters to match your risk appetite.
What you can filter and set
Our simulation controls let you narrow the universe and tune the rules so results reflect your real trading preferences:
Tip: narrow by exchange or watchlist to focus on the tickers you actually trade: this makes the results more actionable and realistic.
Stay tuned - more filters and customization options are on our roadmap.
What the simulation returns
Each run produces two primary outputs:
- Perfect path - evaluates all possible one‑trade‑per‑day combinations and finds the sequence that would have maximized returns for the period.
- Prediction path - applies your chosen probability threshold and filters, then simulates trading only the model’s same‑day predictions.
From these we compute metrics like cumulative return, capture rate (prediction ÷ perfect), number of trades, and win/loss breakdown. You get a clear, side‑by‑side comparison that shows how close your rules come to the theoretical ceiling.
How to use results to validate your strategy
Run a few short simulations with different thresholds and filters to see how the prediction path moves relative to the perfect path. Useful experiments include:
- Vary the probability cutoff (50 → 60%) to trade fewer, higher‑confidence events.
- Exclude very low or very high yields to remove noise and outliers.
- Limit to your watchlist to measure performance on the stocks you actually monitor.
Look at the capture rate and trade count together: a high capture rate with very few trades may be less useful than a slightly lower capture rate with a steady stream of opportunities that fit your time budget.
Practical example and interpretation
Imagine you run a two‑week simulation on your watchlist with a 65% probability cutoff and €10,000 initial capital. The perfect path returns 18% while the prediction path returns 10% — a capture rate of 56%. That tells you the model found more than half of the real upside while filtering out lower‑probability picks. Use that insight to decide whether to raise the threshold for fewer false positives or lower it to increase trade volume.
| Metric | Perfect path | Prediction path |
|---|---|---|
| Cumulative return | 18% | 10% |
| Capture rate | — | 56% |
| Trades | 14 | 8 |
Best practices
- Check the system generated simulations. They use default filters and give valuable information.
- Start narrow: run simulations on a single exchange or your watchlist before scaling up.
- Use short windows (max 2 weeks) to test tactical rules and avoid overfitting long, noisy periods.
- Compare multiple thresholds and pick one that balances precision and recall for your style.
- Remember market surprises: backtests show historical performance, not guarantees.
Unsure which threshold to start with? Review the historical performance analysis: The Perfect Hunting: Finding Your Optimal Threshold & Filters. It breaks down accuracy, precision, and recall by threshold, exchange, and yield tier so you can pick a starting point that matches your risk tolerance.
Run a custom simulation today: validate your rules, build confidence in the model, and discover how many high‑probability opportunities match your strategy. The heavy lifting is done; you decide the filters and the risk rules.
