Even changing the timeline didn't change the final wall
The scenery you see may change, but the wall you eventually hit remained in the same place.
About this series (Verification Journey Part 4 of 6)
- Episode 1: The signal of the beginning of Part 4 (released)
- Episode 2: The more you raise the time axis, the easier it is to pass? (released)
- Episode 3: The same answer came up again, a sense of mismatch (released)
- Episode 4 (this article): The wall did not disappear even if you change the time axis ← right here
- Episode 5: Found again in the intervening time axis
What remained unchanged even in the "gap" time axis
In the first two episodes, I’ve told you that even in the time axis of the "gap" (H1), several candidates were found, and among them there were some that matched conditions previously seen. This time, I’ll share one more realization.
That is,the candidates that "almost made it but fell at the final hurdle" remained numerous even when the time axis was changed, which is a fact.
【Image: 29_near_miss_distribution.png to be inserted here (owner pasting manually)】
Numbers that look good crumble a little when slightly disturbed
As mentioned several times in previous series, statistical validation is not about deciding pass/fail based on a single criterion. We test whether a candidate is truly robust from multiple angles—slightly shifting parameters, adding small noise to price data, and so on—to verify.
Even candidates that seemed to show good numbers can easily crumble under these checks. This wall has stood in front of us since Part 1. And this time, even when exploring the "gap" by changing the time axis,this wall did not disappear.
【Image: 29_timeframe_vs_wall.png to be inserted here (owner pasting manually)】
“Changing the time axis” is not magic
What I felt again this time is thatchanging the time axis shows us a "different scenery," but it is not a magic that eliminates the wall of verification itself.
Even with a short time axis, the wall is strict; with a longer time axis it eases somewhat but does not disappear. Trying the "gap" time axis yielded the same result. The candidates that fall at the cusp continue to exist in large numbers no matter which time axis you try.
【Image: 29_same_wall_every_timeframe.png to be inserted here (owner pasting manually)】
What I wanted to convey in this article
We had hope that "if we change the time axis, we could find more easily and more in number." That feeling exists within us as well. However, when we actually tested it, reality was not so sweet.
The fact that many near-miss candidates remain can be read asthe candidates that passed the test have nevertheless wound through tougher hurdles, which reaffirms the essence of validation that does not change with time axis, and this was a moment of renewed realization.
Disclaimer
- The content shared in this article does not constitute an endorsement of specific trades or investment advice. It is for information sharing only.
- Validation results based on past data do not guarantee identical results in the future. Markets always change.
- Whether you actually trade or not should be decided at your own risk.
Thank you for reading this far. If you’re curious, please also check out the products page of Indicator Laboratory.