Research Workflow
How to Compare Backtest Results
A professional workflow for comparing backtest results side by side so strategy decisions come from tradeoffs, not one attractive metric.
Traseq··6 min read
A professional workflow for comparing backtest results side by side so strategy decisions come from tradeoffs, not one attractive metric.
Start with a no-code crypto spot strategy, lock the version, run the backtest, and keep the result traceable for comparison.
Compare backtest results by choosing a baseline, changing one meaningful variable at a time, and reviewing performance, risk, conditions, and time periods together. The best-looking return is not always the best research candidate. A strategy variant can improve return while worsening drawdown, reducing trade count, or becoming more fragile after fees and slippage.
In Traseq, comparison sets help crypto spot researchers review multiple historical simulations side by side. Traseq is a research workspace, not live trading or exchange execution software.
Use this comparison order:
If you cannot explain what changed between two backtests, the comparison is weak.
A baseline is the reference result every other result is judged against.
The baseline can be:
The baseline does not need to be perfect. It needs to be stable and easy to explain.
In Traseq, a useful baseline should point to a finalized Ready version and a clear run configuration. That keeps the comparison tied to exact rules and assumptions instead of memory.
Backtest comparison breaks down when too many variables move together.
Weak comparison:
v1: BTCUSDT, 1h, RSI 30, no trend filter, no slippage
v2: ETHUSDT, 4h, RSI 35, EMA filter, stress slippage
If v2 looks better, what caused the improvement? The symbol, timeframe, RSI threshold, trend filter, or cost setting?
Stronger comparison:
v1: BTCUSDT, 4h, RSI 30, no trend filter, Market Baseline costs
v2: BTCUSDT, 4h, RSI 30, EMA trend filter, Market Baseline costs
Now the comparison asks one clear question: did the trend filter improve the research result?
Return is useful, but it should not be the first and last decision metric.
Use a broader comparison set:
A variant that improves return but doubles drawdown may not be an improvement. A variant that reduces drawdown but eliminates most trades may not provide enough evidence. A variant that only works with no costs may be too fragile.
Traseq comparison sets are designed for side-by-side research review.
The comparison workflow lets you review:
This matters because many strategy decisions are visual and path-dependent. A summary table may show two similar returns, while the equity curve reveals that one version suffered a long unrecovered drawdown.
For product steps, use the comparison guide.
A professional comparison asks whether the return path was survivable.
Look at:
If a variant increases return by taking deeper or longer drawdowns, document that tradeoff instead of calling it better.
For deeper risk interpretation, read Maximum Drawdown in Backtesting.
Trade count does not automatically validate a backtest, but it affects how much confidence you can place in the result.
Be careful when:
Sometimes the best next step is not choosing a winner. It is extending the range, comparing another market, or simplifying the rule so the strategy produces enough evidence to review.
Costs can reorder strategy rankings.
When comparing variants, make sure fee and slippage assumptions are controlled. If one run uses no slippage and another uses stress slippage, the comparison may be answering the wrong question.
A clean cost comparison looks like this:
If one version only wins under clean assumptions, treat it as fragile until reviewed further.
Every comparison should end with a decision:
Do not keep every version alive because one metric improved. Research quality depends on deciding what not to pursue.
A useful comparison note can be short:
v2 reduced max drawdown and kept profit factor stable, but trade count fell by 60%.
Advance v2 for a longer-range test before changing exits.
That note is more valuable than a screenshot of the best return.
A backtest comparison is a side-by-side review of two or more historical simulations. It helps you understand whether a change improved the strategy, worsened risk, reduced evidence, or only changed the result under one assumption.
Start with the baseline, then compare profitability and risk together. Do not choose a winner from total return alone.
Change one meaningful variable at a time when possible. If multiple settings change together, the comparison may not explain what caused the difference.
No-code workflows make iteration faster. That speed is useful only if each version remains traceable and each comparison explains a real tradeoff.
No. Traseq helps you compare historical research evidence. It does not provide financial advice, guarantee performance, or make live trading decisions.
| Lens | What to check |
|---|
| Performance | total return, profit factor, expectancy, equity curve |
| Risk | max drawdown, drawdown duration, recovery behavior |
| Trades | trade count, average win, average loss, holding period |
| Conditions | which rule changes created the difference |
| Time periods | whether the result depends on one market regime |
| Costs | whether fees and slippage change the ranking |
Mar 6, 2026