Change in historical values (and calculation method?)

[uok]

Hello,

Please note: what I write here is a honest question and not to blame anybody. I want to learn for the future and how to better use this indicator for important Bitcoin price ranges.

While backtesting and looking through historic data from API, I noticed a discrepancy with the indicator values.
In my script I have set 85 as a limit to get notified (for years). At that time the chart on your website showed this as a good area to look more closely into possible Bitcoin tops.

Previously, historical data showed the indicator staying below my threshold.
I found screenshots of my script vs. checking today api data for the same day
2021-10-20: the index was 61.5 vs. 0.8638 (~87 on the chart) today
2025-07-21: 81.5 vs. 0.9342 (~94 on the chart)

Which explains why in the past I didn't get an alert and today I would. While I understand that small rounding errors, etc. happen, this is much more as it makes the difference between an elevated indicator value vs. an useful alert.

This suggests the calculation logic has changed over time.
Could you please look into this and confirm. If yes, where can I get more information?
I'm thankful for ideas on how to capture these changes in the future. Setting the threshold lower would result in many alerts that are too early.

Thanks and have a great weekend!

[ColinTalksCrypto]

Hello, I understand what you are saying. I added the answer to the FAQ a little while ago to explain what is happening:

The CBBI score for last cycle’s top changed. Why?
Most bitcoin metrics fail to account for what is commonly called “diminishing returns” (a decrease in gains/losses from one cycle to the next). This is one of CBBI’s core observations and solutions. To address this, CBBI employs "regression lines" on each metric's chart in order to assist that metric in remaining maximally useful as an oscillator for each upcoming BTC cycle.

These regression lines must be realigned once per cycle to account for the new upper and lower boundaries of the charts. This is one of the core features that has been in place since the beginning, to improve the practical usefulness of each metric in creating a single oscillating score.

Regression lines are shown as the orange and green lines here. This visualization was created so they can be easily seen.

Here’s a short video example of the before-and-after of such a realignment.

Screenshot of the FAQ answer: