If you slightly shift the numeric values, will it not collapse?

Introduction

In the previous article, we explained the significance of the third test, "Random Comparison," in relation to a guessing comparison. This time, we present the fourth test, "Parameter Perturbation."

What is parameter perturbation

Among the conditions are several "parameters," such as the length of the period and threshold values. In parameter perturbation, we deliberately shift the parameters of the found conditions slightly to see whether similar performance trends are still maintained.

Why deliberately shift numerical values

During the exploration process, sometimes good results occur only for a specific combination of numbers. This merely indicates that there happened to be an optimal point around those numbers, and it does not necessarily imply that the underlying idea of the condition is broadly applicable.

If the results collapse significantly with only a small shift in numbers, it is likely relying on a pinpoint coincidence. Conversely, if similar trends persist despite some numerical shifts, it suggests that the condition does not depend too much on any particular number.

The idea of seeking a "gentle hill"

What we focus on in this test is whether the good performance comes from a sharply peaked point or from a gentle hill. If it is a gentle hill, it is judged that the results are robust to some misalignment. If only a sharp peak is good, we view that as more due to luck than actual ability.

What we have learned so far

Although it is a modest task to repeatedly shift numbers slightly, skipping this step would lead to mistaking "luckily good numbers" for actual ability. In the next article, we will discuss the fifth test, "Noise Tolerance," and how to verify resilience to small market fluctuations.

Disclaimer

・The content shared in this article is not intended to recommend specific trades or provide investment advice. It is merely for information sharing.

・Verification results based on past data do not guarantee similar results in the future. Markets are always changing.

・Whether you actually execute trades is a decision you must make at your own risk.

・The indicators and tools introduced in the future are one of the factors for market analysis, and using them does not guarantee profits.

If you are interested, please also check out the product page of Indicator Research Institute.