Translate the HTML to English, keep the HTML format, do not use code blocks or insert line breaks, and decode entities before translating. Original: EAの『カーブフィットの有無』の見抜き方 Translated: How to identify the presence or absence of curve fitting in EA

Here, we offer several EA products with high win rates and even past undefeated records.
High-win-rate EAs, past undefeated EAs, and the like.

When you see such EAs described as “super high win rate” or “past undefeated,”
you might think that they are simply curve-fitted (over-optimized) EAs,
and some of you may feel that way.

However,
from the early stages of EA development,
we keep in mind the idea of
“as much as possible avoid curve-fitting,”
and the construction of a logic that is less prone to curve-fitting,

and proceed with development accordingly.
And,
whether an EA is curve-fitted (over-optimized) or not

can actually be judged quite simply.
That is,
“comparing backtest results with forward-test results.”

As a result of this comparison,
if the backtest and forward results show no difference (or only a small difference),
then that EA is not overfitted, and
it can be judged that its good performance is simply due to the EA’s intrinsic advantages.
And such EAs are likely to maintain similarly good performance in the future as well.

We believe that this is what enables sustained performance going forward.
For example, among the EAs we offer,
“Revolution,” the “ED version,” and “5pair,”

all show backtest and forward-test results that do not diverge significantly, and all yield favorable outcomes.
By combining multiple such EAs that have a low likelihood of curve-fitting (i.e., robust logic) and potential future profitability,
we aim to create a diversified portfolio.
This, we believe, is the true meaning of a portfolio.

Thus, we recommend using a mix of these.
A portfolio that combines such EAs with strong logic and future potential.

That is the intended purpose of a portfolio, we think. m(_ _)m

Neko Hakase