A backtest applies a set of trading rules to historical data. It can help investigate an idea, but the result depends on the rules, data and assumptions used. An attractive equity curve is an invitation to inspect those details.
1. Are the rules complete?
“Buy the breakout” is not enough. Which price defines the breakout? When is the signal confirmed? What order is placed, and when is it cancelled? A reproducible strategy needs entry, exit, sizing and timing rules. If two people would place different trades from the same instructions, the specification needs more work.
2. Was the information available at the time?
Suppose a rule depends on a five-minute candle’s closing price. The test cannot first observe that close and then assume a purchase at the same candle’s earlier low. That would use future information to obtain an earlier price. Check the distinction between signal time, order submission and execution.
3. Does the dataset match the question?
Ask about missing bars, timezones, corporate actions and adjusted versus unadjusted prices. Testing today’s surviving stocks across the distant past can omit companies that failed or disappeared. For an options strategy, underlying stock data alone omits the actual option premiums and contract behaviour.
4. Are trading costs and fills plausible?
Small edges may disappear after spreads, commissions and slippage. Touching a limit price does not always establish that an order would have filled. A stop price is not a guaranteed exit price through a gap. Ask how ambiguous bars, liquidity and unfilled orders are handled.
5. How many ideas were tried?
Trying hundreds of indicator settings and publishing only the best makes luck harder to distinguish from a durable effect. Ask whether the test period was used to choose parameters. A later, previously unseen evaluation is more informative than repeatedly retuning on the same sample.
6. What happened during difficult periods?
Look beyond total return to peak-to-trough losses, losing streaks and recovery periods. A result covering one favourable market environment may not answer how the strategy behaves elsewhere. Report the number of trades: a very small sample can make a win rate look more convincing than it is.
7. What is the comparison?
Consider a relevant alternative over the same dates, with clearly stated capital and cost assumptions. A strategy that earns a positive return has not necessarily outperformed its benchmark. Differences in leverage, time invested and downside exposure can explain part of the outcome.
A practical reading habit
Read the assumptions before the conclusion. If an important detail is missing, treat the result as incomplete. That is a reason to ask for clarification, not proof that the strategy is bad or its creator is dishonest.
The SEC’s investor education guidance emphasises that back-tested performance is hypothetical rather than actual performance. Use a backtest to structure further investigation, with its limitations visible.
Further reading: SEC Investor Bulletin: Performance Claims.
General education only. Historical results do not guarantee future performance.
