range breakout 15m backtest

What the numbers mean

This one lost money over the tested window, and the mechanism is not mysterious. The entry buys the close of the bar that has just made a new high for the session, which is the moment intraday enthusiasm is most fully expressed and most likely to be given back. The exits then clip whatever survives: a close under the 20-bar EMA is roughly five hours of leash, and a two-day clock closes the trade regardless. So the losses arrive at full size while the wins are cut before they can grow into anything.

The parameters sharpen that shape. A 1.5% stop sits inside ordinary 15-minute noise on NVDA and TSLA, so it fires on wiggle rather than on the idea being wrong. The 26-bar lookback spans about one session including extended hours, so the range being cleared is partly thin premarket trade. And the universe is four tickers, two of which, SPY and QQQ, move together most days, so there is less diversification here than the count suggests. With a quarter of the account per position and two positions at most, the portfolio drawdown stays milder than the per-trade experience.

Read profit factor against its breakeven of one rather than against the daily templates elsewhere on this site; it asks only whether the winners paid for the losers. Then compare the worst drawdown with the whole-period result: if they match, the equity peak came early and nothing after it recovered. And treat the annualized rate and the Sharpe as arithmetic on a few weeks of bars, not a description of a year.