Strategy Backtester for Your Own Price Files

About the Strategy Backtester for Your Own Price Files

Load a price file you exported and test your own rules on it, bar by bar: entries when the close crosses a moving average, a fast average crosses a slow one, the close breaks the high or low of the bars before, the RSI crosses your level or the close passes your price — on all or any of them, on the weekdays you choose. Exits on the opposite signal, a stop (a percentage, a multiple of ATR or a price), a target (a percentage or a multiple of R), a trailing stop or a number of bars. Long, short, or both.

Orders fill at the next bar’s open by default, with your commission and slippage on every fill and sizes rounded to your market’s quantity step. Prices can be in your account’s currency, in pence (GBX) for an account in pounds, or in points with a futures contract’s multiplier. You get the trade list with each trade’s R, bars held and its adverse and favourable moves (MAE and MFE), the equity curve and drawdown, net profit, CAGR, Sharpe and Sortino ratios, win rate, payoff, expectancy, profit factor, exposure and cost drag — for an in-sample part and an out-of-sample part side by side, and against buy and hold of the same file.

The rules are white-box: you set every one. There is no optimiser and nothing that searches for “best” settings. The tool warns when the trades are too few to say much, when the out-of-sample part is much worse than the in-sample one, and when fills could look ahead. Your file is read in your browser and never uploaded.

Common uses

• Check what a moving-average crossover you read about would have done on your own instrument, after costs.
• See how much commission and slippage take out of a rule that trades often.
• Compare a rule’s in-sample result with the out-of-sample part before trusting it.
• Read the drawdowns and losing streaks a rule went through, not just its profit.

The formulas

• A trade’s P&L = side × size × (exit fill − entry fill) − the commission on both fills. A fill is the price moved against you by your slippage; the size is rounded down to your quantity step.
• Stops and targets are checked against each bar’s high and low; a bar that opens beyond a level fills at its open, and when the stop and the target are both inside one bar the stop is assumed to come first. A trailing stop follows the best high (or low) since the entry and never loosens.
• R = a trade’s net P&L ÷ its initial risk (size × the distance from the entry to the stop).
• Equity is the cash plus the open position marked to each close; the drawdown is equity ÷ its running peak − 1. CAGR = (end ÷ start)^(1 ÷ years) − 1, from the file’s dates.
• Sharpe ratio = the mean of the equity’s bar-to-bar returns less the risk-free rate, ÷ their standard deviation, × √(bars a year) (Sharpe, The Journal of Portfolio Management); the Sortino ratio divides by the downside deviation instead (Sortino and van der Meer).
• Profit factor = gross wins ÷ gross losses; expectancy = the mean net P&L per trade (and in R); payoff = the average win ÷ the average loss; exposure = the share of bars with a position.
• Buy and hold puts the starting equity into the file’s first close (in fractions of a unit, with your commission both ways) and holds it to the last close.
• Overfitting: a rule tuned to one stretch of prices tends to do worse on the next (Bailey, Borwein, López de Prado and Zhu, Notices of the AMS). The out-of-sample part is a test the rules were not set on; the warning shows when its average trade return is less than half the in-sample one.

Reading the results

• Too few trades: under 30 closed trades, the figures say little — a few lucky or unlucky trades move them a lot.
• In sample against out of sample: set your rules on the in-sample part, then look at the out-of-sample part. Much worse results there are a sign the rules fit the past rather than a pattern.
• Same-bar fills (at the close of the bar that gave the signal) can look ahead: the rule reads a close that is only known when the bar ends. Fills at the next bar’s open avoid it.
• Simulated past performance (FCA COBS 4.6): these are simulated results on past prices, not a record of trades anyone made.

What it leaves out

Whatever your file does not hold: dividends, splits and corporate actions (unless the prices are adjusted), interest on cash, margin and its costs, borrowing fees for short positions, taxes, partial fills, liquidity and the price moves inside a bar. Real fills can be worse than the open, especially after gaps. Costs you do not enter are not counted.

Limitations

• One instrument and one set of rules at a time; up to 50,000 bars.
• No optimiser and no parameter search: you change a setting and run it again.
• Stops and targets inside a bar are judged from its high and low; the order of prices within a bar is not known.
• Results are simulated on past prices: they do not predict future results.

Privacy

Everything happens in your browser. What you enter or open here is not uploaded or stored by MySmartCoPilot.