Methodology
KAIROS marks a window, never a date.
Timing tools fail in a predictable way. They claim more precision than the evidence supports, then quietly bury the times it did not work. KAIROS was built around the opposite habit: bracket the window, score every one of them afterwards, and show the misses next to the hits.
Knowing a market is historically cheap does not tell you when it stops falling. Those are two separate questions, and running them through one tool produces something that answers neither well. KAIROS exists to hold the second question on its own: when have conditions resembled the periods in which cycle lows have historically formed.
The word it deliberately avoids is prediction. A cycle low is only knowable afterwards. What is knowable in advance is that certain configurations of price and cycle timing have preceded lows more often than chance would account for, and that this has never been reliable enough to name a day. A window is a claim you can be measured against. A date is mostly a marketing decision.
KAIROS runs on price and cycle timing alone. It reads no on-chain data of any kind, which is deliberate. A timing tool that shared inputs with the on-chain tools would agree with them for reasons that have nothing to do with the market.
Rather than one cycle length applied everywhere, KAIROS runs a set of detection engines in parallel, each tuned to a different band of cycle lengths. A window is committed only when they align. One cycle read will always find a cycle. Requiring agreement between several is what stops the tool marking a window every few weeks.
Once enough forward data exists, each window is graded against fixed rules as a hit, a partial or a miss, and that record is what the member-facing statistics are built from. The rules were fixed in advance so a disappointing window cannot be reinterpreted later. A timing tool that does not publish its misses is not a timing tool.
The same machinery can be pointed at cycle highs, and it was. Testing showed no demonstrated edge, so the high-side windows are hidden from members by default rather than shipped as a feature with a caveat attached. Something that does not work is not improved by a disclaimer. KAIROS says nothing about tops.
A window opening does not mean the fall has finished. The tool publishes how far price has historically dropped between a window opening and the eventual pivot, as a banded range with levels. It is the least flattering number on the screen, and leaving it out would make the tool look considerably better than it is.
It cannot tell you which windows will work
This is the hardest limitation to sit with and the most important one to state. Windows that failed looked identical at the moment they opened to windows that worked, across every observable that has been tested. The tool can tell you that you are in a period resembling past lows. It cannot tell you whether this one is among the ones that resolve.
Context changes the odds, and bear markets are worse
Windows opening in bear conditions have historically resolved worse than windows opening in bull conditions. The tool labels market context for exactly this reason, and the label should carry real weight in how you read a window rather than being treated as decoration.
The typical-move figures flatter the tool
Historical move sizes are measured from the exact pivot low, which nobody actually buys, and they lean toward the windows that resolved. Read them as an upper bound on what was theoretically available, not as an expectation. The same applies to every statistic on the card: these are historical frequencies from the tool's own archive, not probabilities about what happens next.
Some samples are thin
Certain combinations of market context and timeframe rest on very few past windows. Where that is the case the tool marks the sample as limited rather than presenting a percentage from a handful of observations as though it were a settled figure.
The misses are on the chart too.
KAIROS runs across Bitcoin, Ethereum, the S&P 500 and Gold for members, on the dashboard and as a TradingView indicator.
Discussion