Drawdown - The biggest obstacle preventing traders from achieving profits!

In the previous article on Edge and Expected Value, we learned that the way to achieve long-term profitability in trading is by using a strategy with a positive expectancy. However, when implementing such a strategy in reality, it may encounter consecutive losing streaks due to random distribution, causing our capital to decline—this is referred to as the drawdown of a strategy.

Drawdown is the difference between the peak and trough of capital value within a given period. During trading, we will experience numerous drawdowns. If left unchecked, we might abandon the strategy out of fear when the account suffers significant losses, even though such fluctuations are merely natural random variations that any trading system must endure before accumulating a sufficiently large sample size for its edge to take effect. This is the greatest challenge we face and must overcome to achieve profitability. It is the key to stepping into the realm of successful traders.

It is crucial to keep drawdown within an acceptable level where we can maintain the psychological confidence to continue following the strategy. For example, if we experience a 10% drawdown, we only need an 11.1% profit to break even. However, if the drawdown reaches 50%, we would need a 100% profit just to return to the starting point.

When a strategy with an edge is in play, drawdowns often last significantly longer, sometimes up to 2–3 years, leading us to mistakenly believe that the strategy has lost its edge. During such times, continuous account declines create psychological pressure, prompting us to abandon the strategy. The larger the drawdown, the easier it is for fear to take over. Ironically, it is often right at the moment when we feel the most psychological pressure and decide to give up that the strategy begins to perform again. This is a common paradox that traders frequently encounter.

This often leads many traders into an "infinite loop of strategy switching." Each trading strategy performs differently depending on the market phase and conditions. A trader, even with multiple positive expectancy strategies at hand, can still fail completely if they keep switching strategies continuously. Whenever they encounter a drawdown phase, they switch to a new strategy, unaware that the drawdown is simply a result of the current strategy not aligning with the market conditions. If they switch strategies at that moment, the market might also transition to a new phase, and this new phase could be incompatible with the new strategy they adopt but perfectly suited to the old strategy they just abandoned. As a result, they fail to let any strategy run long enough to demonstrate its effectiveness. 

The image below illustrates the psychological journey of a trader applying a strategy with positive expectancy in practice. Although the strategy was promising, the trader abandoned it just before it began to deliver results. This lack of patience led them to miss the entire profit opportunity. 

Drawdown is an inevitable consequence of the market's uncertainty. It cannot be avoided. We must face it and manage it effectively through stop-loss orders and adjusting trade volumes. A system without drawdown is not truly a good system because it lacks the necessary barriers in strategy execution. This is akin to a business without any barriers to entry, which struggles to sustain itself and grow. If everyone can easily adopt the strategy, who will profit from it?

A good trading system, capable of maintaining a long-term edge, is one that inherently includes drawdown. Drawdown acts as a barrier, preventing everyone from adopting the same strategy. Many will lose patience and abandon the system when it encounters a losing streak. They may switch to other strategies or even trade against the previous system. This natural process of attrition allows the strategy to continue being effective and sustain its profitability over the long term. 

From my experience, it is essential to keep drawdown below 20% at all times to maintain psychological stability. To control drawdown effectively, we must employ strategies that include stop-losses for all trades and then evaluate the maximum possible drawdown that could occur. For strategies without stop-losses, such as Dollar-Cost Averaging (DCA) or martingale-style money management, this type of risk cannot be controlled because they do not cap the maximum risk, often letting losses run indefinitely. As a result, the maximum drawdown for such strategies becomes unpredictable and unmeasurable. Therefore, it is best to avoid using such strategies altogether.

Each level of risk per trade results in different drawdown levels. The higher the risk per trade, the larger the drawdown and the greater the account volatility. Drawdown is influenced by the percentage risk per trade, the frequency of trades, and the correlation between trades. To determine the appropriate level of risk to take, we need to measure the historical performance of the strategy and then run a Monte Carlo simulation based on those performance metrics. This helps us estimate potential drawdowns and make informed decisions about risk management.

For those unfamiliar, Monte Carlo simulation is an algorithm that uses randomness to solve problems, providing a range of possible outcomes and the probability of each outcome based on a given dataset. Named after the famous Monte Carlo casino in Monaco, these simulations work on the principle of repeated random sampling. Monte Carlo simulations allow the generation of multiple trading outcome sequences based on existing data. By creating and analyzing these simulated outcomes, we can gain deeper insights into a variety of potential scenarios that may occur in the future.

Keep in mind that everything we observe in historical data does not necessarily represent the worst-case scenario that could happen in the future, as it is not the only possible scenario. In the article on The Illusion of Control, we learned that a probability distribution function can produce various price scenarios over time. Trading a system works the same way; due to random distribution, the equity growth curve we observe in reality is merely one single scenario out of many possible ones. It would be inaccurate to estimate risk solely based on the maximum drawdown observed in this single scenario. Instead, we need to account for the broader range of potential outcomes to better understand the risks involved.

A strategy with positive expectancy can provide profits and meaningful expected value over a sufficiently large sample size. However, in reality, we only observe and react to outcomes from a small sample size. Thinking deeply about this can feel overwhelming because our daily actions don’t operate in this way—our behavior and decisions are influenced by the "single path" we experience and perceive. This creates a disconnect between the theoretical understanding of probability and how we intuitively respond to the outcomes we encounter in practice.

Suppose you have a strategy with a win rate of 50% and a risk-reward ratio (R:R) of 1.1. A quick calculation using the expectancy formula shows that this is a strategy with an edge. You decide to use this system over 1,000 trades and evaluate the results. What you get is just one of countless possible outcomes. Looking at the figure above, I ran a Monte Carlo simulation for this strategy 100 times. To make this simulation easier to understand, imagine these results represent 100 people in 100 parallel universes, all trading the same strategy. Due to randomness, after 1,000 trades, they end up with different results.

The initial capital here is $50,000. You can see that the worst-case scenario after 1,000 trades ends with a capital of just $11,100—a loss! At this point, you might wonder, “What on earth is going on? I thought we were using a system with an edge!” And you’d be right—this is indeed a system with an edge. However, the trader in this particular universe is losing simply because of extreme bad luck! Even after trading 1,000 times, their expectancy has not yet converged to the mean. If they continue to trade this strategy, convergence will almost certainly happen, but whether they have the psychological stamina to persist is another matter entirely. This demonstrates that you can completely go broke trading a system with positive expectancy if you fail to manage risk properly.

Mathematical formulas and expectancy calculations provide us with a number that closely approximates the average outcome when the sample size is sufficiently large. However, they fail to account for the fluctuations in capital that may occur when the system is applied in real-world scenarios. It’s hard to believe that a trading system with an edge, thoroughly tested, could still result in losing up to 80% of an account simply due to bad luck. Yet, this is entirely possible—and in some universe, it is guaranteed to happen. What we must do is strive to control this possibility in the best way possible.

To achieve this, we use Monte Carlo simulations to evaluate the maximum possible loss in the worst-case scenario. Based on my experience, adjust the risk percentage so that the maximum loss in the worst-case scenario is capped at 50%, and the average max drawdown across all scenarios is less than 15%. This approach helps determine a reasonable risk percentage that allows you to maintain psychological stability while achieving meaningful profits. If the risk percentage is set too low, the profits you earn may be insignificant. Along with strict risk management, we also need to strike a balance to ensure profitability.

For individual strategies, I limit the maximum risk to 0.5% per trade. If multiple strategies are combined, this percentage is reduced, resulting in a maximum drawdown of around 15%. Because my strategies incorporate robust tail risk management, the actual drawdown in practice doesn’t deviate significantly from this level. I allow it to fluctuate up to 50% in the worst-case scenario, but such cases are highly unlikely to occur.