Strategy backtest
Hypothetical trades
A strategy backtest invents fills on a chart.
You already ran the experiment.This tells you what the other outcome would have been.
A simulation takes the trades you already closed and asks a different question of them. Remove the ones that match a condition, or keep only those.
Move a stop or a target and replay the exit on real candles. You get your book both ways, side by side.
Tessera simulation
Closed trades
- NQLong+$420Kept
- ESShort-$180Removed
- CLLong+$95Kept
- GCShort+$310Removed
- YMLong-$62Kept
A Tessera simulation starts from the trades you already took.
Your trades, not a simulation of a trader
Most testing tools replay a strategy over market history and generate trades that never happened. The results depend on assumptions about fills, spreads and how the strategy would have behaved.
A Tessera simulation starts from your closed trades. The ones that stay keep the profit and loss they actually made. Only the trades you remove, or the exits you move, change anything.
Two questions you can ask
Remove matching trades
Take out everything that matches a condition and see what the book looks like without it. Use it when a report has shown you something that leaks.
Keep only matching trades
Strip the book down to just the trades that match and see what is left. Use it when you want to know how much of your result comes from one kind of setup.
Trades that stay keep their original profit and loss. The simulation does not re-price them.
Replay the exit
Instead of removing trades, change where they ended. Move a stop or a target by a distance or a multiple of volatility, and Tessera replays the exit on the candles after your entry.
You can also test moving to breakeven, trailing the stop, or a time based exit.
Exits are replayed on real candles, but the result is scaled from the exit you actually got. It is not a new broker fill, and it does not model tick by tick behaviour.
Exit replay
Same trade, different end
Original exitReplayed on candles
Filters and exits in the same run
A single simulation can do both. Remove the trades that were not worth taking, and move the stops on the ones that stay. The results separate the two so you can see which change did what.
Removed
The trades that left the book.
Changed
The trades that stayed but ended somewhere different.
Filtered
What your book looks like after both.
| Metric | Baseline | Filtered book | vs baseline |
|---|---|---|---|
| Trades | 90 | 54 | -36 |
| Win rate | 41.1% | 53.7% | +12.6% |
| Total PnL | -$392.81 | +$84.20 | +$477.01 |
| Expectancy | -$4.36 | +$1.56 | +$5.92 |
Line them up
Run a filter one way, then another, then a third. Every run stays in the session so you can put them next to each other and see which one actually helped.
Keep up to 16 on a report at once. Delete one and the slot frees up straight away.
Combine two rules that both held up
Two policies that each worked can be merged into one and run together. Keep rules can require all conditions or any of them. Remove rules work the same way.
If the combination cannot work, Tessera says so rather than running something that will not hold.
Two ways in
From a report
A finding carries the filter that produced it, so it arrives in the simulator already built. This is the quicker path and the one most runs take.
From scratch
If you already know what you want to test, build the policy yourself. Pick the account, the dates, the symbol and the timeframe, then structure the rules directly.
Simulate before you deploy
Get started and pressure-test policy changes against your logged history.