How NBA steals actually behave

Where the coin-flip line sits in NBA steals

Each player's own most balanced threshold: the half-point line he has finished above in closest to half his logged games, measured from his distribution rather than read off a book. Deepest sample first, not best to worst. The 40 deepest samples of 520 measured; the figures above this table are over all 520.

The steadiest NBA steals records

Standard deviation in the market’s own units, and the same figure as a share of the player’s average so it means the same thing in points as in passing yards. A narrow spread means the average is close to what a typical game actually looked like; it says nothing about what the next one will look like.

The widest NBA steals swings

The other end of the same measurement. A wide spread means the average describes the middle of a broad range rather than a typical night, which is worth knowing before reading anything into that average.

The widest home and away gaps in NBA steals

Both averages are over the games actually played at each venue, and the game count for each is in the cell beside it. A gap measured over a handful of games either way is one you cannot distinguish from noise. This table prints the denominators so you can tell which is which. Three games a side is the floor for appearing here at all.

Ahead of their own baseline over the last five

Last five logged games against the full log, reported only where the gap is more than a tenth of the player’s own average. It is a description of five games that have been played and it is not a forecast of the sixth.

Behind their own baseline over the last five

The same comparison in the other direction, and the same caveat: five games is five games, and a run below a baseline is as often noise as it is a change.

Common questions