Backtesting a pullback rule: what the numbers actually show
A backtest tells you what happened after this pattern in the past. It does not tell you what happens next time. That distinction does more work than most of the actual statistics.
What's actually in the table
For every historical WATCH or STRONG signal on an asset, GoldAlert looks 5, 10, and 20 trading days ahead and records two numbers: the average return, and the win rate — how often the price ended up higher than the signal day's close. They're shown side by side on purpose. Either one alone can tell a misleading story.
How an average lies to you
Nine small losses and one big rebound can still average out positive. That's not a hypothetical — it's just arithmetic. A glowing average return can be hiding a losing track record for most of the individual signals behind it. The win rate sitting right next to it is what catches that.
How a win rate lies to you
Flip it around: a high win rate built from a string of tiny gains can still lose money overall if the rare loss is large enough to wipe them out. Position sizing and exit discipline decide that outcome, and neither is something this simulator does for you.
Check the sample size before anything else
Three or four completed signals isn't a track record — it's a handful of coin flips dressed up as statistics. Any row built from fewer than five completed observations gets a small asterisk in the table for exactly this reason. Read those rows as "here's what happened these few times," not as reliable evidence of anything.
The date range you pick changes the story
120, 250, or 500 trading days — each window can happen to capture mostly a rising market, mostly a falling one, or a stretch that was unusually calm or unusually wild. None of that necessarily represents what's coming. A quick way to sanity-check a pattern: run the same asset through a couple of different range lengths and see whether the numbers hold up or fall apart.
What no backtest can promise you
Markets change character. A pattern that worked cleanly through one stretch of history can behave completely differently once conditions shift — a rate cycle turns, a sector rotates, sentiment flips. A backtest is a place to start digging, not a green light to trade on.
Want to go deeper? GoldAlert's charts are powered by TradingView. For advanced technical analysis, real-time data and professional screening tools, explore TradingView plans → (referral link — we may earn a commission at no cost to you).
This article explains how to interpret historical statistics and is not investment advice. Past performance does not guarantee future results.