EA Craftsman's EA Course 【027】 What is a reproducible EA? Dangers of using indicators in EAs (Part 2)

I don't know the reason, but…

Last time, I asked traders who use indicators...


Among the countless combinations of indicators and parameters, was there a rational reason for choosing that one?

This time, I'll go further and discuss the problems of EAs that use indicators.

By the way, what I’m about to explain also applies to discretionary trading.

When beginners start trading in stocks or FX, many will first learn about moving average lines, Bollinger Bands, MACD, RSI, and the Ichimoku Kinko Hyo, among others.

And for some reason, you’ll see traders who seem to be ahead of you using those tools to explain things, even though you don’t know why.

Many beginners, without questioning anything,

think: “I see! It’s the common sense in the market to analyze with indicators!

If I understand the indicators for unknown reasons, I’ll be able to predict future price moves and win!

The Ichimoku cloud works for unknown reasons!

If price touches the 2σ of Bollinger Bands, I should take the counter-trend, right!

If RSI indicates overbought, I should sell, right!

And this becomes a mistaken common sense.

Where is the mistake, what is the problem, can you tell?

Self-attuned Analysis

The biggest problem is that it completely ignores the psychology and the basic behavior of other participants.


Nowadays, many trades are done by machines or AI, but that doesn’t mean human participants are gone, and machines are ultimately created by humans.

AI learns to outsmart other participants, and its foundation is ultimately other participants’ behavior, so it isn’t independent of human psychology.

With that in mind, what is the basis of indicators’ overbought or oversold signals?

Who decided that RSI above 70 means overbought?

Was there a story where a famously successful short-term trader gathered statistics and found that beyond 70 is overbought?

On the contrary, if a strong trend pushes RSI above 70, wouldn’t it be reasonable to expect the price to rise for a while longer?

When price enters the Ichimoku cloud, why does it stop falling or rising there?

Because, historically, when you fit it to past charts, it empirically seems to behave that way.

In other words, the reason is unknown. A rational explanation cannot be given. Yet in past charts it somehow appears to work.

What is universality?

Yes. If you have thoroughly understood the contents of the EA course so far, you can probably recognize that this is the entrance to overfitting.


If you input this parameter into this indicator and run it, backtesting will show it earns the most!

For unknown reasons.

We covered the method to achieve the best backtest results in the fifth EA course, but if you continue trading with indicators, whether discretionary or automated trading, you will eventually fall into the trap of overfitting.

Because there is no rational reason grounded in the behavior of other participants.

Whether it’s recent highs, ranges, key levels, or places where other participants are watching or placing orders—the places where positions would have been held—

if you base your trade plans and entries/exits on those grounds, it’s hard not to imagine that universality exists there.

In other words, indicators that can change in any way with parameters are only serving your own convenience.

Even a single moving average becomes a completely different thing if you set its parameter to 25, 75, or 200...

What parameters are other participants using?

Are major speculators all using 200?

If so, it might work, but there’s no evidence of that anywhere, and even if it were so, there’s no reason to believe they’d continue using 200 forever.

This proves that indicators lack universality and highlights the problem of continuing to use them.

Therefore, a reproducible EA should be based on the essence and principles of the market that reflect other participants’ behavior, without relying on indicators.

That doesn’t mean you must never use indicators.

If you don’t use them, the logic becomes more complex, so using them as an auxiliary tool is acceptable for convenience.

The problem is trading anchored on indicators.

■ My developed EA concept and operating policy

EA Craftsman’s EA (Three Arrows) is here