Two strategies. The first drops 17%, sits there for a few ugly weeks, and makes a new high six months later. The second never loses more than 5%. It also doesn't make a new high for three years.
Ask most traders which one they'd rather run and they'll pick the second without thinking. Ask which one they'd still be running in year three, and the answer flips.
Depth is what you measure. Duration is what you feel.
Every backtest report leads with maximum drawdown. It's one number, it's easy to compare, and it answers exactly one question: how bad did it get at the worst moment? It says nothing about how long you had to live there.
That second number has a name. Time under water is the stretch between one equity high and the next one that beats it. Two systems can share the same max drawdown while one recovers in a week and the other takes half a year, and the report treats them as identical. Your nerves won't.
Public markets have already run this experiment for us. In 2020 the S&P 500 fell 34% in about a month, then closed at a new record on August 18, six months after the February peak. Painful, fast, over.
The Nasdaq's March 2000 peak is the other kind. It didn't reclaim that level until April 2015. Fifteen years under water.
Trend followers know the long kind well. The stretch from late 2010 to early 2014 is still called "CTA winter" in the managed futures world. No crash, just more than three years of choppy markets and flat-to-negative returns, and a steady trickle of investors who had sat calmly through 2008 deciding they'd had enough.
Why the shallow one breaks you
Here's the thing about a -5% that lasts three years. It never gives you a moment of crisis. No single day forces a decision. Instead, it asks you the same question every morning: is this thing still working?
Crypto trades every day of the year, so three years is about 1,100 mornings of that question. A sharp -17% gives you maybe forty bad days and then a recovery that restores your faith. In my experience, people who abandon a system rarely quit at the bottom of a crash. They quit in month fourteen of a flat line, usually not long before it starts working again.
There's also an opportunity cost that makes it worse. While your strategy goes nowhere, something else is going up: Bitcoin, an index, a friend's altcoin. Three years of watching that makes a shallow drawdown feel far deeper than the number on the report.
So the mistake isn't ignoring risk. It's treating max drawdown as the whole risk picture. You're choosing a system by how deep the water is and ignoring how long you'll have to hold your breath.
Put a clock next to the depth
Before you trade any strategy, pull two more numbers out of the backtest: the longest time under water and the median one. Write them down next to the max drawdown, in the same place you keep your rules.
If you want one metric that blends both, look at the Ulcer Index. Peter Martin and Byron McCann published it in 1987 to penalize drawdowns by depth and duration together, precisely because a single max-drawdown figure misses the second part. A strategy with a lower max drawdown and a higher Ulcer Index is usually the harder one to live with.
Then ask yourself one question: could I watch this equity curve go nowhere for that many months without touching it? If the answer is no, the strategy doesn't fit you, however shallow its worst dip looks. Size it smaller, pair it with something uncorrelated, or pick a different system.
My own rule is simple. I'll accept a deeper drawdown if the recovery is fast, and I won't accept a long flat period I couldn't have predicted from the backtest. Knowing the expected time under water in advance turns month fourteen from a crisis into a checkpoint.
Trend following in crypto has deep, fast drawdowns by nature. That's the price of catching the big moves, and I'd rather pay it in weeks than in years. We publish all our backtest data at v33systematic.com if you want to check both the depth and the clock for yourself.