Lock it in or wait?
Sleeper's default: 0.5 a point, 1 a rebound or assist, 2 a steal or block, −1 a turnover, 0.5 a three, +1 double-double, +2 triple-double, +2 at 40 points and +2 more at 50.
Quick call
Pick a player and enter his score. You get the call, the chance waiting does better, and the line to lock at.
My roster this week
The lock line after each game he has left. If his score clears it, lock it; if not, wait. His last game of the week counts as is.
How lock-in works
The rules. Weeks run Monday to Sunday. Only one game per player counts toward your matchup. After a player finishes a game you can lock that score, but only until his next game starts; then that game is gone. A lock is final, and a locked player can't be moved to another spot that week. Only players in your starting lineup that day can be locked. If you never lock him, his last game of the week counts, even if he doesn't play it (a zero).
The trade-off. Keep a good score now, or gamble that one of his remaining games is better. The more games he has left, and the more likely he plays them, the higher a score has to be to keep. A shaky back-to-back or an injury listing makes waiting worth less, so the line drops.
The lock line. We work backwards from Sunday. Before his last game, waiting is worth his average game times the chance he plays it. Before each earlier game, waiting is worth what you'd get by locking that game only if it beats the line after it. Lock tonight's score if it beats what waiting is worth. That rule gives the most points on average over a season; it won't win every week.
His range. His last 40 games at his current minutes (within 20% of what he's expected to play now), scored with your league's settings, then nudged up or down (at most 25%) to match his projection. The chance he plays each game comes from our Will he play? forecasts, injury report included.
Rules checked in October 2026 against Sleeper's help center (“League formats & game modes”), where lock-in is the default for new basketball leagues, and the FantasyPros and RotoWire guides to the format. The last-game and starters-only rules come from those guides; Sleeper's own page doesn't spell them out.
Does it work?
We replayed every week of 2025-26 for the top 200 fantasy players (3,833 player-weeks), using only what was known at the time.
Using these lock lines, the counted game averaged 21.2 points per player per week, against 21.1 for “lock anything above his average”. That edge is small, about 0.2 points per player a week (1.7 across a 10-man lineup): “lock above his average” is already a good rule of thumb, and most weeks both lock the same game. Both beat locking his first game (18.5) or never locking (16.3) by a lot. Nobody can pick the best game in hindsight (23.6); the lines got 90% of the way there.
- Best game in hindsight (can't be done)23.6
- Our lock lines21.2
- Our lines, assuming he plays every game21.1
- Lock anything above his season average21.1
- Lock anything above his last-10 average21.0
- Lock his first game, every week18.4
- Never lock (his last game counts)16.3
Our lines did better than the average rule in 369 player-weeks, worse in 215, and the same in 3,249 (both locked the same game). Average edge 0.2 ± 0.1 points (95% range).
How the test workedHide
Every Monday-to-Sunday week, for each player who played at least once: after each game, each rule decides to lock or wait. With no lock, his last game counts, a zero if he sat it, as on Sleeper. His range is his last 40 games before that week at the minutes he was playing then (we don't have projections from back then, so no nudging). Play chances are our Will he play? forecasts as they stood the morning after each game, injury report included; that model was fit on the same seasons, so this part of the test flatters us a little.
When the raw numbers said “waiting does better X% of the time”, here is how often it did. They were too sure of themselves at both ends, so the chances shown on this page are corrected with this table:
- said 3%: happened 10% (750 nights)
- said 15%: happened 21% (473 nights)
- said 25%: happened 28% (441 nights)
- said 35%: happened 38% (445 nights)
- said 45%: happened 51% (412 nights)
- said 55%: happened 56% (422 nights)
- said 65%: happened 63% (405 nights)
- said 75%: happened 70% (607 nights)
- said 86%: happened 77% (978 nights)
- said 94%: happened 82% (1,045 nights)
Method and limits
Each player's range is his own recent box scores from our database, scored with the selected settings. Bonuses (double-doubles, 40-point games) count as they happened and aren't scaled. Technical and flagrant fouls aren't in our box scores. Before the season starts, his range is mostly last season.
The line assumes he is in your starting lineup for every game left. If a lineup crunch will keep him on your bench some night, that game doesn't count for him, so you should lock a bit lower than the line.
It treats each player on his own. Sleeper lets you work around a slot (moving players between spots before they lock), which can make waiting a little more attractive than we show.
Play chances for players our forecast doesn't cover use a flat 85% and say “rough”.
These are averages, not promises. A player can always have his best game on Sunday.