Why did I end up analyzing as many as 30,000 charts?
In the previous article,
I talked about starting the development of AutoChannelEntry from the question,
“People differ in how they draw channels.”
Next, I thought about this.
“Will the current channel actually be effective?”
Even if you can draw a channel,
nobody knows whether it will resist strongly,
or break through.
So I asked myself,
“Was there a similar channel in the past?”
I wondered.
For example,
assume there have been 100 past channels that closely resemble the current form.
Among them,
・How many times did it resist?
・How many times did it break?
・How many times did it turn out to be a false signal?
If we knew this,
it could serve as evidence for judging
“How reliable is the current channel?”
I thought.
However,
a few dozen cases are statistically weak,
and perhaps it was just a coincidence.
Therefore I decided to
analyze as much past data as possible.
In the current ACE Gold,
we analyze 30,000 past charts and
identify shapes similar to the current channel and display statistics.
Of course,
the number 30,000 wasn’t chosen on a whim.
We tested the balance between processing speed and accuracy many times,
and arrived at the current specification.
From here on, you can view it after free registration
Next, we will discuss
✅ Why we adopted 30,000 as the number of analyses
✅ How we calculate the relapse rate, break rate, and false signal rate
✅ What we learned from repeated backtests
about these topics.