Session 3: Why we don’t teach a “winning setup” — The reason USDJPY Ultimate publishes parameters (All 6 parts)

In the previous articles, we talked about why USDJPY Ultimate chose the “trending” method as the classic approach, and how we construct the “entry conditions” and the design for losses. From the reasons for choosing the classic approach to the entry conditions, and the design when losses occur, this time we will discuss the extension of that line: the disclosure of parameters.

There are likely questions like “which settings produce the most profits in the end?” and “why don’t you tell us the optimal parameters?” In this article, we will explain why we disclose the types and mechanisms of the parameters while not providing a single answer of “this value makes you win.”

Why we don’t tell you the “winning settings”

Many commercial EAs are “black boxes” with internal logic and parameters hidden. This is a natural stance from the developers’ perspective, because if all logic and values are disclosed, there is a risk of imitation. Black-boxing itself isn’t inherently bad.

However, I see another issue. Market conditions are always changing. A setting that worked well in the past may not perform the same if price movements, volatility, or the presence of trends change. If users know nothing about how the EA works, it becomes difficult to reason about “why performance changed” or “which part no longer fits the market.” Ultimately, the only option could become “find another EA.”

Therefore, USDJPY Ultimate emphasizes leaving room for users to understand and verify the inner workings.

“Disclosing mechanisms” is not the same as “telling the exact correct numbers”

There is one point that tends to cause confusion. Disclosing the types and mechanisms of parameters is completely different from telling you the exact correct numbers that guarantee “you will win.”

If I were to claim definitively, “these values are optimal,” it might just be showing the most convenient values for a past period after the fact. This would essentially reproduce the same issue discussed in previous articles: overfitting to past data. That’s why what’s published is only the mechanism, not a “correct answer.”

You have the option to verify it yourself

With parameters disclosed, users can do the following:

Of course, some people may find this verification burdensome. Basically, it’s designed so you can use the default settings. The default values themselves were set after extensive backtesting. However, using the default without understanding why a particular value was chosen, versus using it with an understanding of the mechanism, changes how you approach the market when conditions change.

Next time, we will discuss how to judge performance declines after you actually start using the EA, and the mindset for doing so.