TRACE · the trajectory model
A career is one curve.
TRACE reads it at three distances.
Trajectory Regime Analysis, Change & Early-warning follows one number per player per game — his fantasy value that night against the league of that season — and answers three questions from it: how good is he right now, has something actually changed, and is something about to.
The first answer is a level with an uncertainty, updated after every game. The second is a regime — Stable, Rising, Breakout, Sliding or Bust — which only changes after seven games of movement and only confirms after thirty, so the words are rare and mean something. The third is a set of calibrated probabilities on the next thirty games: that a breakout-bound rise begins, that a bust-bound decline begins, that a rise already underway will hold, and that a fresh breakout will be lost again.
Availability is kept out of the measurement by construction: the axis is the games a player actually played, so an injury is a gap in the record, not a decline. And no probability is ever labelled with a cause, because none has earned it.
Jokić and Maxey are both Stable at a career high, but Maxey’s level is sagging below the level he just reached, so his downside probability is 31% — and the risk that his January breakout is lost within a season is 49%. Alexander-Walker is in a transition: thirteen games into a rise, with a 92% probability it confirms, and no upside or downside figure because he is no longer pre-departure.
For scale: from a stable level, a breakout-bound rise and a bust-bound decline each begin within thirty games about 8.6% of the time; the models roughly double that in their top decile and still discriminate 21 to 30 games ahead. Of confirmed breakouts, about 29% are lost again within a season.