OptiNod Academy

Maximum Drawdown: The Limit That Determines Whether You Can Keep Trading a Strategy

Maximum drawdown is not just a record of past losses. It is the capital and psychological limit that determines whether you can keep trading a strategy. Recoveries are asymmetric, and live drawdowns are usually deeper.

Maximum drawdown, or MDD, is the largest percentage decline from a peak in an equity curve to the next trough. If capital falls from $10,000 to $7,000 before recovering, that segment has a 30% drawdown. The calculation itself is simple: find the deepest trough relative to the prior high.

Most traders glance at this number as just another line in a backtest report. They spend time studying an 80% annual return, then move past a 35% maximum drawdown as if it were only a loss that already happened. This number goes beyond a record of the past. It also determines whether the strategy can be traded going forward.

Drawdown becomes a limit for three reasons. First, the return required to get back to breakeven rises asymmetrically as drawdown deepens. Second, when a drawdown lasts too long, traders abandon the strategy regardless of its long-term return. Third, live drawdowns are usually deeper than the maximum drawdown shown in a backtest. Once you understand these three points, maximum drawdown stops looking like a reference number beside returns and becomes the starting point for sizing and strategy selection.

Required recovery returns are always larger than the drawdown

If you assume the drawdown and the recovery return are the same number, you underestimate the distance back to breakeven. It is tempting to think that after losing 20%, you only need to make 20% to recover. The math does not work that way. After a 20% loss, 80% of capital remains. To get from 80 back to 100, you need a 25% gain. Losses are applied to a larger base, while recoveries are applied to a smaller base, so the required recovery return is always larger than the drawdown.

This gap widens quickly as drawdown deepens. A 30% drawdown needs a +43% gain to recover. A 50% drawdown needs +100%. An 80% drawdown needs +400%. Once the drawdown passes 50%, the required recovery return is more than double the drawdown, and the distance starts moving beyond what ordinary strategy performance can realistically overcome.

BTC shows this math clearly. From the roughly $69,000 high on November 10, 2021, to the roughly $15,476 low on November 21, 2022, BTC fell about 77.6% based on daily closes. From that trough, it needed about +346% to return to $69,000. BTC did not close back above $69,000 until March 11, 2024. The full recovery took about 28 months from the peak and about 16 months from the trough. A one-line 77% drawdown turned into more than two years of recovery time.

The same asymmetry applies at the strategy level. Even with a high annual return, one 50% drawdown forces the strategy to earn +100% just to return to the previous high, flattening the compounding curve for a long time. This is how a strategy chosen only for its return can end up trapped in a recovery zone it cannot realistically escape.

Traders usually tire of the underwater period before depth breaks them

Drawdown has another dimension besides depth: the time spent below the prior high, known as the underwater period. Depth measures how far the equity curve fell. Drawdown duration measures how long it stayed below the high. In live trading, duration is often what causes traders to abandon a strategy, more so than depth.

The reason is psychological. If a 30% drawdown recovers in one month, most traders do not lose faith in the system. If the same 30% drawdown remains unrecovered for ten months, doubt builds every day as the account balance stays below its former high. In a backtest, the recovery appears instantly on the chart. In real time, the trader must endure every trade without knowing whether recovery will come. The most common failure comes from quitting the strategy just before recovery.

BTC holder behavior shows this pressure from duration. The $15,476 low in November 2022 was only a brief event, but BTC remained below its prior high for more than two years. Many long-term holders endured the few days around the low, then sold during the long, dull consolidation through 2023. Most of them sold when the fatigue of doubting recovery grew heaviest, with the heaviest selling clustered well above the cheapest price.

If you look only at maximum drawdown and ignore duration, you miss this risk entirely. Two strategies can both have a 25% maximum drawdown, but one may recover in three weeks while the other takes eight months. In live trading, those are completely different strategies. This is why the longest underwater period should be checked alongside maximum drawdown in every backtest report.

Backtested maximum drawdown is the floor for future drawdown

The most expensive misunderstanding is treating the backtested maximum drawdown as the limit of what can happen in the future. If a report shows a 28% maximum drawdown, traders often allocate capital as if the live bottom will be somewhere near that level. In practice, that 28% is closer to a floor for future drawdown than a limit. Future drawdowns almost always exceed the historical maximum.

The reason is the sample. A backtested maximum drawdown is only the worst segment inside the historical data. Events that did not occur in that dataset are not included in the calculation. If an out-of-sample shock arrives, meaning a shock of a size not present in the test period, drawdown can immediately exceed the historical record. Even if the backtest spans a long period and includes several volatile regimes, its maximum drawdown is still only the worst event observed so far, well short of the worst event that can happen next.

Live trading also adds losses that only appear in execution. Backtests often assume low or no slippage, simplify fees, and treat every order as filled at the intended price. But slippage and missed fills are largest exactly when volatility spikes and drawdowns deepen. Backtests rarely capture that relationship well. As a result, the trough in the live equity curve is usually deeper than the trough in the simulated equity curve.

BTC itself is an example of this out-of-sample logic. From the 2017 high to the 2018 low, BTC drew down about 84%. A system built only on 2017 data might have treated an 80% drawdown as the worst case, but BTC then fell about 77% again between 2021 and 2022. Deep drawdowns were a recurring feature of the asset. Anyone who treated a single historical drawdown as a hard limit was unprepared for the second shock. Live sizing should add a buffer to the backtested maximum drawdown so the strategy can keep running even if drawdown reaches 1.5 to 2 times the backtest figure.

The type of capital determines the drawdown you can actually trade through

The same 30% drawdown can be tolerable for one pool of capital and strategy-ending for another. The difference comes down to the nature of the money behind each pool, with the absolute drawdown number staying the same in both cases. Losing 30% of surplus investment capital and losing 30% of money needed for next month’s living expenses are numerically identical but practically very different losses.

The key issue is recovery time. An equity curve needs time to return to its prior high. If the trading capital includes money needed for living costs or near-term expenses, the trader may not be able to wait. If funds must be withdrawn before recovery arrives, losses are realized during the drawdown. This is where the recovery asymmetry becomes most damaging. If capital is down 50% and part of it is withdrawn, the remaining capital cannot return to the original amount even after earning +100%.

So the tradable drawdown limit should be set by looking at capital along two axes. The first is how much of total assets is allocated to the strategy. The second is how long that capital can remain untouched. If most of a trader’s net worth is in one strategy and that money is tied to expenses within the next year, even a 20% drawdown may be hard to tolerate. If only a portion of total assets is allocated and the capital will not be needed for five years or more, even a 50% drawdown may be something the trader can wait through.

This is where forced liquidations spread most aggressively during the 2022 bear market. Positions funded with borrowed money or short-term capital were liquidated through margin calls and capital withdrawals before prices recovered, triggering cascading liquidations. Other positions saw the same price decline but avoided liquidation because they were funded with their own capital and could wait for recovery. The ability to withstand drawdown comes down to what kind of money is being traded, while the strategy’s statistics stay the same across both cases.

Use maximum drawdown to work backward into position size

Once you treat maximum drawdown as a limit, the sizing process reverses. You first define the drawdown you can tolerate, then work backward into size, reversing the usual order of choosing size first to maximize return and checking risk afterward. The limit comes first. Bet size comes second.

The method comes down to two numbers. First, define the maximum drawdown you can tolerate on the capital allocated to this strategy. Next, estimate the strategy’s expected live maximum drawdown by adding a buffer to the backtested figure, typically using 1.5 to 2 times the backtest drawdown. Divide your tolerable drawdown limit by the expected drawdown, and you get the share of total assets that can be allocated to the strategy. If a strategy’s expected maximum drawdown is 40% and you can tolerate only a 20% drawdown at the total-portfolio level, the maximum allocation to that strategy is 50%.

This sequence also changes how strategies are compared. Suppose one strategy has a 60% annual return with a 50% maximum drawdown, while another has a 35% annual return with an 18% maximum drawdown. If you compare only returns, the first strategy looks better. But if the capital can tolerate only a 20% drawdown, the first strategy breaches the limit during a normal drawdown and cannot be traded through a full drawdown. Only the second strategy can keep operating. A high return from a strategy you cannot keep trading remains just a number on a report.

• [ ] Set the operating limit: Separate the capital allocated to this strategy from total assets, then define the maximum drawdown you can tolerate on that capital, such as 20%.
• [ ] Estimate live drawdown: Check the backtested maximum drawdown, then multiply it by 1.5 to 2 to account for out-of-sample events and slippage. Use that as the expected live drawdown.
• [ ] Calculate allocation: Divide the operating limit by the expected live drawdown to determine the allocation to this strategy. If expected drawdown is 40% and the limit is 20%, the allocation is 50%.
• [ ] Check recovery time: Review the longest underwater period in the backtest and confirm that you can leave the capital untouched for that long. If it is longer than one year, separate it from living-expense capital.
• [ ] Define invalidation: If live drawdown exceeds 2 times the backtested maximum drawdown, treat the strategy as no longer matching the data. Reduce size, stop trading, and review it.

Three common ways traders misuse drawdown

Traders can still misuse the drawdown number even when they have it in front of them. Each mistake leaves out one of the mechanisms described above.

Looking only at returns and ignoring drawdown. If you rank backtests by annual return and choose the highest one, you may never check whether its maximum drawdown exceeds the operating limit of your capital. High returns often come with large drawdowns, and drawdowns that exceed your limit can end the strategy before recovery arrives. Returns and maximum drawdown should be evaluated as a pair, and returns should be compared only among strategies that fit within your operating limit.

Mistaking backtested maximum drawdown for a hard limit. If you allocate all available capital based exactly on the maximum drawdown in the report, you have no buffer when an out-of-sample event arrives. The backtested number is a lower bound, with real drawdowns running above it. Live sizing should always leave room above it.

Missing the recovery asymmetry. If you treat drawdown and recovery return as the same number, you underestimate the distance back from deep losses. Assuming a 50% drawdown can be recovered with +50% ignores the actual +100% required, causing traders to overestimate recoverability and tolerate losses that are too large.

If maximum drawdown is read as a record of past losses, it remains just one line in a report. If it is read as the limit that determines whether a strategy can keep running, it becomes the starting point for sizing and strategy selection. Only traders who factor in asymmetric recovery, psychologically difficult underwater periods, and live drawdowns that exceed backtests at the sizing stage can hold their position through a deep drawdown and remain there for the recovery.