What a drawdown is, and why to look at the decline before the return

When a trading system is presented, the first thing shown is the return: "29% a year", "doubled in three years". The first question should be a different one: how much did it fall at its worst point, and how long did it take to recover? That is the drawdown, and it is the number that determines whether a person will be able to follow the system or will abandon it on its worst day.

What a drawdown is

A drawdown is the decline from an account's most recent peak to the lowest point that followed. If an account reached US$20,000 and then dropped to US$16,000, it is in a 20% drawdown. It is always measured from the peak, not from the starting point: an account can be above its initial capital and still be deep in a drawdown.

The maximum drawdown (max DD) is the worst of those declines over the entire period under review. When a backtest reports "−20%", it means that, at the worst moment of those years, the account was 20% below its best mark.

There is a second measure, less often quoted and just as important: how long it lasted. Time underwater is the period from the peak until the account recovers it. A −20% that lasts three months is experienced very differently from one that lasts two years.

The arithmetic of recovery

Declines and gains are not symmetrical. Recovering from −10% requires +11%. From −20%, +25%. From −50%, +100%. The deeper the decline, the more disproportionate the effort required to recover it.

That means two systems with the same average return are not equivalent. One that returns 25% with −15% declines is better than one that returns 25% with −40% declines, even though the headline number is identical. With the second, there is a year in which almost half of the account disappears, and whoever follows it must decide, with the account at a loss, whether to continue.

What really matters: the person following the system

The annual return is an average that no one experiences as such. What is experienced is the day to day, and the day to day of systematic trading is a sequence of streaks: some good, many mediocre, some bad. The drawdown is the bad streak measured in money.

The practical question is: can you see your account 20% below its peak and still execute the next day's signal? If the answer is no, that system is not for you, whatever it returns on paper. Almost no one abandons a system during a good streak; it is abandoned at the bottom of the drawdown, which is exactly the worst moment to do so.

That is why the first number to look for is the decline, not the return. The return indicates how much could be made if the system is followed to the end. The drawdown indicates whether the end will be reached.

How to compare: the ratio between the two

A simple way to combine the two measures is the MAR: annual return divided by maximum drawdown. A system that returns 29% a year with a −20% maximum drawdown has a MAR of 1.4. One that returns 40% with −50% has 0.8. The first returns more for each unit of decline endured.

It is not a perfect measure (it depends on the period, and a long backtest tends to show larger declines than a short one), but it serves as a first filter: a MAR below 1 means the worst decline was larger than what the system makes in an average year.

The Ecliptic Alerts numbers, with their frame

As an example of how this information should be presented, the numbers of the system Ecliptic Alerts uses are published on the home page of eclipticalerts.com, each with its frame: the backtest (simulated; 2016-2025 for the core, 2022-2025 for the full system) with its maximum drawdown shown before its return, and the live result (own account, real money, since August 2026, unaudited), updated every night. They are not repeated here so that this note does not go stale: a number without its date is exactly what this note asks you not to accept.

One detail from the first days of real trading, in late July 2026: the account fell 6% in three sessions before recovering. That is what a drawdown looks like in practice: it gives no warning, and it often appears shortly after the start. The system itself states in advance that declines of −20% or worse are a normal part of the process. None of this is a promise: the past, simulated or real, does not guarantee the future, and this is not financial advice.

What to ask before following any system

1. What was the maximum drawdown, and over what period was it measured?
2. How long did the longest recovery take?
3. Is the number a backtest or live? Since when?
4. Is the return net of costs?
5. What does the system do during a bad streak: keep trading the same, reduce position size, or stop trading?

If whoever offers the system cannot answer all five, or answers only the one about the return, that in itself is a sufficient answer.

An example day shows what a real signal looks like: https://eclipticalerts.com/?demo=1