In this lesson
- get to know the metrics a strategy is judged by
- see why a single number (total return) misleads
- learn why few trades mean little certainty
"The strategy made 1,747%." That sounds great – and says almost nothing. Over how many years? With what drawdowns? From how many trades? And how much would simply holding have made? This lesson is about the numbers you always need to see together.
Metrics you always calculate#
- Annual return – the average gain per year, so tests of different lengths can be compared.
- Maximum drawdown – the biggest fall from a previous peak. This is what you will actually live through.
- Profit factor – the sum of profits divided by the sum of losses. Above 1 the strategy is in profit; the higher, the better.
- Win rate and average win and loss – the strategy's character from the previous module.
- Number of trades – how much evidence is behind the result.
- Comparison with simply holding – a strategy that does not beat "buy and hold" is not worth the effort.
Our strategy in numbers#
A 40-day average on Bitcoin in 2019–2021 (the period in which we chose the average length):
| Metric | Strategy (40-day average) | Buy and hold |
|---|---|---|
| Total return | +1,747% | +1,148% |
| Annual return | +164% | +132% |
| Maximum drawdown | −48% | −63% |
| Number of trades | 29 | 1 |
| Win rate | 31% | – |
| Average win / loss | +61% / −3.6% | – |
| Profit factor | 7.6 | – |
- Strategy (40-day average)
- Buy and hold
- Logarithmic axis: equal distance = equal percentage change
What the numbers really tell you#
Three trades did most of the work. Of 29 trades, the three biggest brought 75% of all profits. Had one of them failed, the result would look completely different.
29 trades are few. With such a small sample you cannot tell a good strategy from a lucky accident. From the lesson on expected value you know a more reliable picture comes from a hundred or more trades.
A −48% drawdown is reality. Even a "great" strategy would have cost you almost half the account in the test. Would you sit through it – with real money, not knowing it would end well?
Total return is the least important number
Two strategies with the same return can be completely different: one with a 20% drawdown and hundreds of trades, the other with a 60% drawdown and ten trades. Always judge the whole table, not one number.
Common myths#
- "It made more than buy and hold, so it is better." Look at the drawdown and the number of trades first.
- "A profit factor of 7.6 is great." From 29 trades carried by three – lucky, more likely.
- "I can handle the drawdown." On paper, everyone can.
Summary
- Judge a strategy by a table: annual return, maximum drawdown, profit factor, win rate, number of trades and a comparison with holding.
- Few trades = little certainty; a few big trades can make up most of the result.
- Maximum drawdown is what you will actually live through.
- The result on the data you tuned the strategy on is always too optimistic.