Free checklist
The Pre-Trade Backtesting Checklist
12 checks before you risk a dollar. Run them on any backtest — yours, a template's, or a vendor's demo — and you will know whether the equity curve is evidence or decoration.
A backtest earns your trust one check at a time. Work down the list; if a check fails, fix the test — not the position size. Nothing here requires a specific platform, and nothing here is a performance claim: these are the questions, your run supplies the answers.
Data & premise
Point-in-time data
Confirm every input was knowable at decision time — publication timestamps, not period labels. If a feature could have been computed with future information, the test is invalid.
Survivorship
Check that delisted and bankrupt symbols are still in the universe. A dataset containing only companies that exist today flatters every long-only strategy.
Corporate actions
Verify splits, dividends, and ticker changes are adjusted consistently on both entry and exit legs — not just in the price series you plotted.
The premise, stated
Write the market behavior your strategy assumes (e.g. 'prices oscillate around a mean'). If you cannot state it in one sentence, you cannot check when it stops being true.
Execution & costs
Fill model
Ask how orders fill: at the close, next bar, or with a spread? A test that fills at the price you saw is measuring a market that does not exist.
Costs, itemized
Commissions, spread, slippage, and financing — set explicitly per instrument, not a hidden default. Then double them and re-run: if the edge dies, it was the costs all along.
Capacity & liquidity
Compare position size against average volume. A strategy that works at $5K may be untradeable at $50K, and the backtest will not tell you unless you make it.
Out-of-sample behavior
A holdout the search never touched
Reserve a stretch of history that parameter tuning had no access to. If the tuning process saw it, it is in-sample, no matter what you call it.
Walk-forward, not one split
Run the tune-freeze-score loop across multiple windows. You are looking for consistency across windows, not one spectacular segment.
Overfitting accounting
Adjust for how many variants you tried (deflated Sharpe or equivalent) and read the overfitting-risk grade. Twenty tuned parameters on ten years of data will fit anything.
Risk & regime
The drawdown you would actually live through
Find the worst peak-to-trough in the results and the worst single window. Decide — before capital — whether you would have held through it or abandoned the strategy at the bottom.
Regime coverage
Confirm the test includes at least one major drawdown, one low-volatility stretch, and one regime change. A strategy that only works in one environment needs a regime filter or a partner strategy.
Keep the checklist
Get it in your inbox
Leave your email and we'll send the Pre-Trade Backtesting Checklist — plus your 14-day free trial details if you want them.
One email, no sequences. Unsubscribe any time.
The checklist is easier to run when the platform does half the work — walk-forward, cost profiles, and overfitting grades built in.
Start free trial