Methodology
The scored version was scrapped before launch.
The first build of this tool produced a rotation score from zero to a hundred. Testing showed it rated false starts more highly than the real rotations it was meant to find. No member ever saw it. What shipped instead answers a narrower question and reports one of four plain situations.
Rotation is one of the loosest words in this market. It gets used for capital moving into alts, for Ether outperforming Bitcoin, for money leaving one sector for another, sometimes for all three in the same sentence. A tool cannot answer a question that vague.
So this one was narrowed until it could be answered. It looks at Ether relative to Bitcoin, and only at that. It asks whether today's conditions resemble the conditions that genuine past ETH against BTC rotations had in common. It says nothing about any altcoin, and nothing about capital rotating anywhere else.
On the name: the tool carries the word Index because that is what members have always called it and renaming a live tool creates more confusion than it removes. It does not output an index value. It outputs a state, and the four are described below.
The original design led with momentum and produced a single score. In testing it ranked false dawns above the genuine rotations, which is worse than useless in a tool meant to separate the two. It was scrapped rather than adjusted until the numbers improved. Tuning a broken design until it passes is how you build something that only works on the data you tuned it with.
The whole tool rests on four historical episodes. A score to two decimal places built on four observations would imply a confidence that does not exist anywhere in the evidence. Four named situations are honest about the resolution the data actually supports. The precision of the output should match the precision of the input.
Two historical false starts still read as strong setups under the current design. They could have been engineered out. Doing so would have meant fitting the rules to two specific dates and calling the result an improvement. They are documented on the tool instead, which is the version that tells you something true about how often this kind of setup misleads.
Four episodes is a thin base and the tool says so on screen
Everything this tool knows about rotations, it learned from four of them. That is enough to describe what they had in common. It is not enough to make claims about frequency, reliability or what happens next. The board names the episodes it rests on and carries that caveat in the interface, so a member is never reading a state without also seeing how little sits behind it.
It confirms a turn, it does not forecast one
Part of what the tool measures is whether the ratio has actually moved. That makes it late by construction. It will never tell you a rotation is about to begin, and any version that claimed to would be making up the part it cannot see.
Aligned is the state most likely to be misread
Aligned means both measured conditions are present. It carries no view on how far, how fast, or whether this instance resolves like the four that came before. Given the sample it rests on, treating it as an instruction rather than a description is the main way a member could misuse this tool.
Aligned has never meant buy.
Members see the current state, the conditions behind it, and the four episodes the whole thing rests on.
Discussion