#21 "All-Win EA" is trash. The true difference between "optimization" and "cheating late-jack" (backdoor or opportunistic) logic.
December 8, 2025: Note
As announced previously, today“Optimization”will be discussed.
Let me get to the point.
If I set my mind to it,“a invincible EA with a 100% win rate for the past 3 years and zero drawdown”is not a difficult feat for me.
But I won’t do that.
Why? Because it is“fraud”and “garbage.”
Today, I will tell you the difference between the widespread lies of so-called “curve fitting (over-optimization)” and the “true tuning” I practice.
Cunning “after-the-fact” Janken
Many of the so-called Holy Grail EAs are created like this.
1. Look at past charts.
2. Find the places where there were big losses.
3. “Program the entry to not occur at that specific date and time.”
4. Or,“Add special indicator conditions that fit only that movement.”
If you repeat this, you get a clean ascending graph on the historical data.
This is“over-optimization (curve fitting)”in its essence.
For the record, this is not programming.
“After seeing yesterday’s Janken result, deciding what to output then.”It’s just a cowardly after-the-fact Janken.
Future markets do not move like the past.
Therefore, such an EA becomes useless trash the moment you put it into actual trading.
What I mean by “Optimization”
I do not change the logic (trading rules) to fit past charts.
My logic is always simple.
“If you break the trace left by the whale’s feeding (the tip of the upper wick), sell.”
That’s all.
The optimization I perform is, while fixing this logic,“selecting the specs of the tools (parameters)”.
・ What number of pips should be the stop loss to be safest?
・ At what pip interval should trailing take profit be pursued to maximize profit?
You could say it’s the act of aligning the rifle’s aiming point.
Without changing the bullet (logic), adjust the scope magnification and powder amount to find the values with the highest hit rate.
Know the weight of “3 years”
If you say that, some pretenders will act smart and say this:
“Even parameter optimization is just fitting to past data.”
…No, it isn’t.
Look at my backtest period.
“Three years.”
I’m not talking about just a few months.
Throughout the long span of three years, in rain or wind,the same parameters (settings)kept winning.
If one says that is “just luck,” you can say so.
But a setting that survived three years and thousands of trials is anything but random.
That is“the market’s nature (physical laws)”.
I’m conducting enormous calculations to uncover that “law.”
Not to predict the future.
“The form of human instinct (fear and greed) that will likely not change in the future.”to be captured in numbers.
That is why my EA is strong.
A mere cosmetically tuned EA has a different skeleton.
Next time:
Actually, I thought the EA was already perfectly finished.
PF 6.46. There is nothing more than this.
I was considering starting the sales prep, when I suddenly realized something.
It’s a matter of physics.
The ball slows down as friction acts on it the more it rolls.
The market is the same.
The longer the trend extends, the more its energy should decay.
If that’s the case…
“Shouldn’t trailing width be shortened with each repetition as well?”
Now that I’ve realized it, I must act.
I will release the seal on the completed “whisker-trimming EA,”and evolve it into an even more formidable weapon.
Next time, I will publish the astounding test results.
Please tremble with anticipation.
※
The EA I am developing is automation of the method introduced in Part 2, which anyone can achieve manually,
“a method that wins even with no scalping”.
150 yen!So cheap!... President, cheap! 9(ˊᗜˋ*)و
× ![]()
