AAlgoTrader AI
Backtest an idea
Backtesting & ValidationAll markets

Backtest to Live: The Complete Pre-Launch Checklist

The gap between a good backtest and a working strategy is filled with specific, checkable items. This is the list.

5 min readIntermediateUpdated September 16, 2026

At a glance

Six categories
Data, methodology, costs, risk, operations, psychology
Rule
Every item must be checked, not assumed
Starting size
Minimum viable, regardless of confidence
Expectation
Live results will be worse than the backtest

Key takeaways

  • Most live failures trace back to a specific unchecked item rather than to a fundamentally broken idea.
  • Data and methodology problems inflate results; cost and operational problems erode them. Both must be checked separately.
  • Write the drawdown at which you will stop before you start, because you will not decide it well later.
  • Start at minimum size regardless of confidence, and scale only on demonstrated rule adherence.
  • Expect live performance to be meaningfully worse than backtested performance, and confirm the strategy is still worth trading at that lower level.

1. Data checks

  • Does the universe include delisted, merged, and failed instruments?
  • Is index or universe membership point-in-time rather than current?
  • Are fundamental and economic figures as-reported, with correct publication timestamps?
  • Is the adjustment methodology documented and appropriate for the rules?
  • For futures, is the roll rule in the data the same as the roll rule you will trade?
  • Has the dataset been audited for outlier bars, zero volume, date gaps, and stale prices?
  • Do bar timestamps mean what your code assumes, in the time zone you assume?
  • Have you cross-checked a sample of bars against a second vendor?

2. Methodology checks

  • Can you state the mechanism, in one sentence, naming who is on the other side and why?
  • Does the strategy execute on the bar after the signal, not the same bar?
  • When a stop and a target fall in the same bar, does the test assume the stop was hit first?
  • Does performance survive a one-bar delay applied to every signal?
  • Does performance hold across a plateau of parameter values, not just at one point?
  • Does the effect appear, in weakened form, in related instruments or markets?
  • How many variants did you test in total, and have you adjusted your confidence accordingly?
  • Has the strategy been validated out of sample or with walk-forward, without iterating on those results?
  • Are the results driven by more than a handful of trades or one favourable period?

3. Cost checks

  • Are commissions modelled with your broker’s actual schedule, including minimums?
  • Is spread cost included on both entry and exit?
  • Is slippage modelled, with a larger allowance for stops and fast markets?
  • Is market impact modelled for your intended order size relative to average volume?
  • Are financing costs included for leveraged or multi-day positions?
  • Are borrow fees included and verified as available for every short?
  • Is currency conversion cost included for foreign-denominated instruments?
  • What is the cost as a percentage of the average winning trade, and is it under 15 percent?
  • What is the annual cost drag, and does the gross edge comfortably exceed it?
  • Does the strategy remain viable if costs are double your assumption?

4. Risk checks

  • What is the 95th percentile drawdown from a Monte Carlo resampling, not just the historical maximum?
  • Is that drawdown tolerable financially and psychologically, at the intended position size?
  • What is the longest simulated time underwater, and would you have continued through it?
  • Is risk per trade defined as a percentage of current equity, with a formula rather than a judgement?
  • Is there a notional cap per position in addition to the risk cap?
  • Are correlation and sector limits defined, and is total open risk capped?
  • What happens if an instrument gaps three times your stop distance against you?
  • Is the drawdown response schedule written down, including the level at which you stop entirely?
  • Does the strategy survive a stress scenario of your three worst trades occurring consecutively?

5. Operational checks

  • Does the data feed you will trade on match the data you backtested on?
  • Have you tested every order type you will use, including brackets and trailing stops?
  • What happens if the platform is unavailable when a signal fires, and is there a fallback?
  • Are stop orders resting in the market, or do they require you to act in real time?
  • Is the position sizing calculation automated, or is it a manual step that can be miscalculated?
  • Is there a trade log that captures signal time, intended price, actual fill, and any deviation?
  • Have you run a forward test long enough to confirm signal counts and slippage match expectations?
  • Do you know how corporate actions, expiries, and halts will be handled?
  • Is there a checklist for the daily routine, so that steps are not skipped under time pressure?

6. Psychology and commitment checks

  • Have you written down what the worst expected drawdown will feel like, before it happens?
  • Do you know the longest losing streak to expect, from the simulation rather than from hope?
  • Have you committed to a minimum number of trades before evaluating the strategy?
  • Is the capital genuinely money you can afford to lose entirely?
  • Is there a rule preventing you from adding funds during a drawdown?
  • Have you defined what evidence would cause you to retire the strategy, before you need it?
  • Are you starting at minimum size regardless of how confident you feel?
  • Is the scaling schedule defined, based on rule adherence rather than on profits?

Frequently asked questions

How long should I wait between backtesting and going live?

Long enough to complete a forward test covering 30 to 50 trades, which for most daily strategies means three to six months. Going live at minimum size during that period is usually better than pure simulation, because it verifies fills and introduces genuine loss aversion at negligible cost.

What size should I start with?

The minimum that the instrument permits: one share, one micro contract, or the smallest lot. The purpose of the first phase is to verify execution and discipline, not to generate returns. Scale only after a defined block of trades with rule adherence above your threshold.

How much worse will live results be than my backtest?

Plan for meaningfully worse: lower returns, deeper drawdowns, and worse fills. A common planning assumption is roughly two thirds of backtested returns and 1.5 times the backtested maximum drawdown. If the strategy is unattractive under those assumptions, it is unattractive.

What if I skip some checklist items?

Each unchecked item is a specific way the strategy can fail that you have chosen not to look for. The data and methodology items inflate your expectations; the cost and operational items erode your results. Skipping the risk and psychology items is how a recoverable disappointment becomes an unrecoverable loss.

When should I stop trading a live strategy?

At the drawdown level you wrote down before starting, or when live results fall outside the distribution your simulation produced. Both criteria must be defined in advance, because in the moment the decision is made under exactly the conditions where judgement is least reliable.

Test this idea before you trade it

Describe the rules in plain language and AlgoTrader AI turns them into a structured strategy blueprint with a configurable historical backtest, cost assumptions, and exportable code.

Build a backtest

Keep reading

Referenced by

Educational use only. This guide explains how a strategy works. It is not investment advice, not a recommendation, and no result described here is a forecast. Test any approach on historical and out-of-sample data, size positions conservatively, and never risk money you cannot afford to lose.