forex-scalping.com

Variance and the Illusion of Skill

Randomness produces patterns that look like skill and patterns that look like failure. Understanding variance protects you from reading too much into results.

Imagine 1,000 traders, each with a strategy that has zero edge, taking a trade a day with a fifty percent chance of winning. After 20 trades, some will have won 14 or more, which is about a 6 percent chance for each trader, so around 60 of them will appear to be successful. After more trades, a few will look like geniuses. They are not. They are lucky outliers in a large population.

The same logic applies to funds, signal providers, and social media profiles. Selection plays a role: you see the survivors, not the many who did equally well by chance and then failed.

In your own results, the lesson is to demand more evidence. A good month is not proof of skill, and a bad month is not proof of failure. Sample size, as discussed, and statistical tests help distinguish.

The practical stance is to treat early success with skepticism, to avoid raising risk after good results, and to avoid abandoning a sound method after a bad stretch.

It also guides how you judge others. A trader with a great record over a short period may simply be lucky. A long, verified record across market conditions is far more informative.

Practical step: simulate a hundred random traders with a coin flip in a spreadsheet, and look at the best and worst results after 50 trades.

The coin flip spreadsheet is worth doing. Watching a few random traders look brilliant after 50 trades is the fastest way to get suspicious of short track records. In a spreadsheet, put RAND() greater than 0.5 in a grid of 100 columns by 50 rows, give each win plus 1R and each loss minus 1R, and sum each column. Several columns will show impressive totals, and a few will look terrible, all from pure chance. Press recalculate and a different set of stars appears. Once you've seen that, a signal seller showing two good months stops being impressive. You start asking for hundreds of trades and verified records instead.

Applies to our own best months too. The temptation to raise risk after a hot streak is strongest exactly when luck might be doing most of the work. I'd decide in advance what evidence is needed before risk increases, something like at least 100 trades with positive expectancy and rule following above 90 percent, then increase in small steps. If a great month makes you want to double size, wait another month and see if the feeling survives. Usually the next month is more ordinary, and you're glad you didn't. The same restraint applies to bad months. Don't cut a strategy because of one losing stretch that fits normal variance.