Rough stretches have days in front of them
A bad week is easy to date afterwards and almost impossible to watch arriving. By the time it is obviously a bad week, the interesting days are already behind you, overwritten by what came after them.
This isn't a failure of attention. The days before a rough stretch are, by definition, days when nothing had gone wrong yet. They looked like any other days at the time, and they were stored that way.
What a forecast would need, and why this is not one
Prediction requires a claim about the future, and nothing here makes one. There is a weaker claim still worth having: in your own history, what has tended to sit ahead of your rough stretches, and how many days ahead it tended to arrive.
That is a statement about the past. It can be checked, it comes with a count, and it doesn't tell you that anything is coming. An app announcing that a bad week is on the way would be selling you a certainty it does not have.
The two checks that look backwards
Sequence is the direct one. It takes a run of rough days and asks what tends to be logged in the days leading up to it, across every such run in your history. What comes back is an ordering with a gap attached.
Lagged correlation is the general form of the same idea. Rather than asking whether two things move together on the same day, it asks whether one on Monday lines up with the other on Wednesday, and it checks a range of gaps rather than a single one.
This looks like a real pattern.
After days when you logged more sleep than usual, your logged energy tends to be lower about 2 days later.
Drawn from 48 pairs of days, between March and August 2026.
About two days, in that example, with 48 pairs of days behind it. The gap is the part that is invisible while you live it. The count is the part that lets you decide how much to make of it.
Why the ordinary days are the ones that matter
A check like this needs a denominator. A tag that appears in the run-up to every rough stretch is only interesting if it does not also appear in the run-up to everything else. Logging the ordinary days is what makes that comparison possible at all.
This is where most journalling breaks down. People write on the days worth writing about, which produces a record of the loud days and no baseline. A twenty second entry exists to make the ordinary days cheap enough to keep.
A build looks different from an event
Two shapes turn up. One is a level that shifts on a particular date and stays shifted, which the change point check is for. The other is a run of days that line up with something logged earlier, which is the lagged shape.
Telling them apart matters more than it sounds. A build and a single event feel identical in memory, since memory files both under the moment you noticed. A log keeps them distinguishable.
- EmergingEarly signal. This might be a pattern, I need more to be sure.
- ProbableThis looks like a real pattern.
- StrongThis is a consistent pattern in your history.
Every finding arrives at one of those three levels, and the level moves as the log grows. Something that stops holding up loses its level rather than sitting on screen at the one it earned last month.
What this cannot do
It cannot tell you why any of it happened, and it does not attempt to. It has nothing to say about a stretch you did not log. It needs a few weeks of entries before a gap can separate from noise, and longer gaps need more days than short ones, since each extra day of gap leaves fewer usable pairs in the same history.
Life Pattern Journal is not released yet. The waiting list on the front page sends one email now to confirm the address, then two more, ever: one when the beta opens and one when it launches.