fix: Puell Multiple indicator range regression (#22)

[Emivvvvv]

This PR fixes issue #22

The Issue:
The "High Model" math incorrectly assumed diminishing returns would follow a permanent straight line, resulting in a ceiling that was far too low for the current cycle.

The Fix:

• The model now looks at the actual peaks of the Puell Multiple itself, rather than guessing based on Bitcoin price peaks.
• I added a rule that the "ceiling" can never drop lower than the lowest historical peak we've seen (the 2021 cycle). This prevents the model from crashing into the ground.

Verification:
I compared the old logic vs the new logic on historical data. The chart below shows how the Old Version (Red) spiked to unrealistic levels (because the ceiling model collapsed), while the New Version (Green) remains stable and realistic.

[ColinTalksCrypto]

Appreciate your interest in this metric. I agree that the Puell Multiple has become less useful in recent times due to how steep the upper regression line has become. This is a correct observation.

The thing is, the purpose of CBBI is to attempt to spot Bitcoin price tops (and bottoms) as accurately as possible. As such, CBBI logic intentionally aligns metrics' upper regression lines based on actual Bitcoin price peaks (as you pointed out), so as to maximize top-spotting accuracy. All of the metrics operate in this manner and we want to maintain consistency.

Our conclusion was that we didn't want to get into the habit of customizing individual metrics for "exceptions" (each having its own set of rules), rather than all metrics following the same rule set.

So, yes, I agree that Puell has become less useful. Another option is to remove it from CBBI entirely if it proves un-useful for its intended purpose.

I am open to making things work better, if we can do so without arbitrary rule sets per metric.

Let's see what Kamil says. Maybe he has additional thoughts on this.

Again, thank you for your chart and your code suggestion.

[Zaczero]

We talked with Colin and it turns out we thought of modeling price versus metric peaks some time ago (years back), and the decision was to model the price highs, allowing for metrics to go above 100% and just clip them.

[Emivvvvv]

@ColinTalksCrypto @Zaczero thank you for your clear explanation of your decision making point about Puell Multiple. I agree with you. My goal was just to make it similar to other examples but for CBBI, this makes a lot of sense. Since all of the indicators are optional in the website, the indicator being a little bit outdated should be fine.

On this—and it is a bit out of topic—maybe if the outdated indicator numbers become a couple instead of one, there might be another default just for the "better" indicators. This is a completely new idea, I know, but I just wanted to mention it.

I'm a huge fan of the idea of CBBI and would like to help it become a better platform overall.

Feel free to close this PR if we agreed that the current way is the way to go.

Best!