How to handle drawdown when the purchased EA experiences drawdown

I found a promising EA and started it after purchasing, but as soon as it began operating, a drawdown occurred.

It didn’t recover, and I ended up with only losses and it was shelved—an experience I think everyone has had.

In the past, about half of the EAs I bought were unusable.

If you say I simply lack the eye, that’s another matter.

In fact, even developers say it’s difficult to predict forward-test results.

If forward results could be predicted, the EA creator should just manage it themselves.

From my experience of making over 100 EAs, no matter how confident an EA is, I can’t be confident until I see the forward results.

The moment confidence becomes conviction is when the forward results match expectations or perform better than anticipated.

Now, let’s consider the causes of drawdowns.

There are three possible reasons.

Reason 1: there is no real edge in the logic to begin with

Reason 2: it’s just a drawdown, and over the long term it will win

Reason 3: the logic has expired

Reason 1 occurs when the logic’s edge is not actually present, often caused by over-optimizing backtests.

It is common in anomaly-type methods.

Imagine a treasure-hunt game where treasures lie along the road.

The goal is to collect as many treasures as possible up to the destination.

You try various routes by brute force, and you take the route with the best result as the correct one.

As you can imagine, the result is that the route with good results happened to be the optimal solution, but there is no real basis for the answer.

If you find some rule for where the treasures lie and base the route on it, then there is a basis.

In EA terms, it’s like using technical indicators or time-of-day biases.

Most EA creators set the default based on the best performance from countless parameter values and don’t adopt the worst-performing settings.

Some say they avoid excessive optimization as part of their policy, but the methods that performed well in past markets are optimized for those past markets, so there’s no real difference.

Whether there is an edge or not can only be inferred by looking at forward results.

Reason 2: it’s just a drawdown.

This is a drawdown that the developer expects.

In this case, continuing to run the EA can push performance into positive territory.

Reason 3: expiration of the logic’s edge

This is the most troublesome case.

It performs well for a while, then at some point the income and expenditure stagnate.

If luck is against you, you may continue to lose.

Possible causes include three items.

① The edge is originally weak, yet market and logic fit well so backtests and forward tests show better performance than the underlying reality.

② The user base increased, erasing the edge.

③ Over time, the market and method no longer fit.

Judgment is very difficult.

Even the EA creator probably doesn’t know.

Whether this is useful is uncertain, but when I run an EA, I consider the market conditions it excels in and those it struggles in.

It’s impossible to predict whether the market will rise or fall.

Also, predicting whether the market is ranging or trending is difficult.

However, I can predict volatility (the range of price movement).

Therefore, during periods of high market movement, I build portfolios mainly with EAs that excel in high-volatility markets,

and during calmer markets, I primarily use contrarian strategies, using volatility as a factor.

■ Conclusion “How to handle drawdowns when they occur”

1. Stop running

2. Lower the lot size and observe

3. Review parameter settings and change currency pairs

I hope this helps your investing.