Gold glossary

    WHAT IS SAMPLE SIZE?

    Sample size is the number of trades a performance statistic is calculated from. It determines how much of a result is signal and how much is noise. Most gold strategy claims fail not because the numbers are false but because the sample behind them is far too small.

    HOW SMALL SAMPLES DECEIVE

    Twelve winners out of twenty is a 60% win rate, but the confidence interval around it comfortably spans systems that are excellent and systems that lose money. Nothing has been proven. The same twelve-of-twenty ratio across four hundred trades is a different statement entirely.

    Low win rate strategies need larger samples still, because the profitability depends on rare large winners that may simply not have appeared yet.

    SampleWhat it can support
    Under 30 tradesProcess review only, no performance claim
    30-100 tradesEarly indication, high uncertainty
    100-300 tradesUsable for expectancy with caution
    300+ tradesSegmentation by session and setup becomes credible

    THE SEGMENTATION TRAP

    Every filter you add divides the sample. Two hundred gold trades split across three sessions and four setups gives cohorts of a dozen or so, each of which is noise. Segment only as far as the data supports.

    This is also how strategies get overfitted: enough slicing will always surface a rule that "works" on history and fails immediately in forward testing.

    RISK NOTE

    Leveraged gold carries a high risk of rapid loss. Everything on this page is education, not financial advice or a projection of returns.

    GOLD TRADING FAQ

    How many trades do I need before trusting my gold results?
    At least a hundred for a broad read, and several hundred before segmenting by session or setup with any confidence.
    Does backtest sample size count the same as live?
    It is weaker evidence. Backtests exclude real spread behaviour, slippage and the trader, so live or forward-tested samples carry more weight.
    Why do low win rate systems need more trades?
    Because their profitability depends on infrequent large winners, and a small sample can easily miss them entirely.
    Can I speed this up with a lower timeframe?
    You gain trades but change the strategy, and costs consume a larger share of each result. It is not the same system tested faster.

    RELATED READING

    PUT THE DEFINITIONS TO WORK

    Live XAUUSD signals on a public ledger, a free gold trade journal that syncs with MetaTrader 5, and the macro pulse before the London open.

    THE FULL GOLD GLOSSARY SET