The trade that ruins an account is almost never the entry. It is the trade taken after four losers in a row, when someone decides the way back to even is to double up and win it all back at once. I have watched this happen enough times, to other people and once or twice to a younger version of me, that I stopped trusting the part of my brain that generates the idea. So the whole point of a drawdown policy is to take that decision away from you before the drawdown makes you dumb. You write the rules while you are calm, and then you follow them while you are not.
Why anti-martingale wins and the gut loses
Martingale is the instinct to size up after losses. You lost, so you risk more on the next one to recover faster. It feels like courage. Mathematically it is the fastest known route to zero, because the exact moment your bet is largest is the moment your losing streak is longest, and streaks in trading run longer than anyone plans for. You do not need a bad strategy to blow up this way. You just need a normal run of variance and the willingness to keep adding to it.
Anti-martingale does the opposite. You risk less when you are losing and more when you are winning. The logic is not emotional, it is arithmetic. When your equity is falling you want smaller bets so the losing streak does less damage and you survive to trade the recovery. When your equity is rising the bets grow off a bigger base, so the good runs compound. The uncomfortable part is that this feels backwards in the moment. Cutting size when you are already down feels like giving up. It is the opposite. It is the thing that keeps a 20 percent drawdown from becoming a 50 percent one, and a 50 percent drawdown needs a 100 percent gain just to get back to flat.
The throttle: cut size as the hole gets deeper
Here is the mechanical version I actually use, and you should tune the numbers to your own strategy and volatility. The idea matters more than the exact thresholds.
- From the equity high, while drawdown is under roughly 10 percent, trade full size. Nothing changes. Normal drawdowns are just the cost of doing business and you do not want to flinch at every wobble.
- At roughly 10 percent down, halve the risk per trade. If you normally risk 1 percent of equity per position, you are now risking half a percent. Same setups, same rules, smaller bets.
- At roughly 15 percent down, halve it again. You are now trading at a quarter size. At this point you are mostly playing defense and staying in the game.
- At roughly 20 percent down, stop. Flat. No new positions until you do a review.
The halt at 20 percent is not a punishment, it is a circuit breaker. A drawdown that deep usually means one of two things is true. Either the market regime shifted and your edge is not working right now, or something in your own execution has quietly broken and you have not noticed. Both of those are reasons to stop clicking and start looking, not to force another trade. The stop buys you the time to figure out which one it is.
One detail that trips people up. Measure the drawdown from your equity peak, not from where you started the month or the year. The peak is the honest reference point. Measuring from an arbitrary start date lets you hide from a drawdown that already happened, and the whole system depends on the numbers being real.
Equity-curve filters for systematic strategies
If you run a systematic strategy, there is a cleaner version of the same throttle that does not require you to eyeball a percentage. You treat your own equity curve like a price series and put a moving average on it. When your live equity is above its own moving average, the strategy is in a working regime and you trade full size. When equity drops below that moving average, you cut size or go to paper until it crosses back above.
The length of the average is the whole design choice. A short lookback reacts fast and whipsaws you in and out on normal noise, which racks up its own kind of cost. A long lookback rides through the noise but leaves you fully sized deep into a bad run before it finally tells you to cut. There is no clean answer here. I lean longer rather than shorter, because the point of the filter is to catch the genuine regime breaks, not to trade my own equity curve like a scalp. Backtest the filter on your strategy's historical equity before you trust it, and be honest that it will occasionally pull you out right before a recovery. That is the price of the protection, and it is worth paying.
The failure mode to watch for is optimizing the filter length until it perfectly dodged your worst historical drawdown. That number is fitted to the past and tells you almost nothing about the next one. Pick a length that is roughly reasonable across several assets and time periods, and resist the urge to tune it to the decimal.
The part everyone forgets: how you scale back up
Most people write down the rules for cutting and never write down the rules for adding size back. So they cut at 10 percent down, the account recovers, and then they freeze. They stay at half size for months because scaling back up feels risky and there is no rule telling them it is allowed. That is its own kind of leak. You spent the drawdown protecting capital and then you failed to let it work when conditions were fine again.
Make the scale-up mechanical and symmetric. If you halved size at 10 percent down, you restore full size when you climb back above that same 10 percent line from the peak, ideally after it holds for a few trades rather than a single lucky day. For the equity-curve version, the cross back above the moving average is your signal to re-size. The criterion should be a level and a bit of persistence, never a feeling that things seem better now.
The written policy does not need to be long. Mine fits on one card, and I would rather it be boring and followed than clever and ignored.
- Reference point: drawdown measured from the all-time equity peak.
- Throttle levels: the size you trade at each drawdown band, written as exact percentages.
- Halt level: the drawdown at which you take zero new positions.
- Review trigger: what you check during a halt before you are allowed to restart.
- Scale-up rule: the exact level and holding period that restores each size tier.
The value of all this is not that it makes you money on any single trade. It is that it removes the one decision that historically does the most damage, the decision to size up into a hole to get even fast. Take that option off the table in advance, and the worst version of a bad month stays merely bad instead of turning terminal. Write the card, keep it where you can see it while you trade, and let it argue with you when you are about to do something you will regret.