BuilderOU STORY【Chapter 1】Episode 15|The verification never ends. Therefore, it keeps evolving.
BuilderOU STORY|Episode 15
Verification never ends.
Therefore, it continues to evolve.
It's not finished after one creation.
Verify, improve, and confirm again.
That repetition builds trust.
Hello.
From BuilderOU (Builder King),
Mr. Ojun.
In Episode 14,
the theme was
“An era where honest verification can be done by anyone.”
I wrote about not wanting verification itself to be the work of only those with special knowledge.
Choose a period.
Choose a currency.
Choose conditions.
And, see the results.
If you create a system where anyone can check by the same method,
you can judge with numbers, not impressions or biases.
However,
here I noticed one more big thing.
Just because you built a verifiable system,
it doesn't mean it's complete there.
The market changes.
The users change.
The environment changes.
Then,
the verification system itself must continue to change.
What BuilderOU needed was
not a finished product.
A system that continually evolveswas what we needed.
1. The moment you think it is “finished,”
is the starting line for the next step
When you are writing a program,
the moment it runs, you’re really happy.
Errors disappear.
Buttons work.
Signs appear.
Backtest numbers appear.
Results appear in Excel.
“I did it.”
I think so.
But,
once you actually start using it, something always becomes visible.
I want this condition as well.
This display is hard to understand.
I want to separate this time frame.
I want to view HIGH and LOW separately.
I want to exclude conditions with few signals.
I want to compare differences by currency.
Fixing one thing reveals another improvement.
In the past, I aimed for
“to complete it.”
But,
as I kept building, my thinking changed.
Completion is not the end.
At that point, it became usable.
Then actually use it,
verify,
find problems,
and fix them again.
That cycle is what I came to believe is real development.
2. Even if the numbers look good,
don’t get complacent
In backtests, good numbers appear.
Win rate is high.
PF is also good.
Trading count is high.
Seeing that, you can’t help but feel hopeful.
“This might work.”
I’ve thought that many times as well.
However,
good numbers and being usable in the future are different things.
There might have been a period where it performed well.
It might have suited only one currency.
It might have been biased toward certain time frames.
If market conditions change,
the same conditions might yield different results.
Therefore,
I stopped judging by a single result.
Change the period.
Change the currency.
Change the time frame.
Change the hour of the day.
Change the market.
Split TREND and RANGE.
Change the conditions slightly and see how much the results change.
Rather than one high win rate,
it’s important that the system remains stable even when conditions change.
That became more important to me.