NBA DFS Floors, Ceilings and p10–p90 Projection Ranges

What a percentile range is

Our ranges come from the model simulating many possible versions of each game. In the model’s simulated outcomes, about 10 percent fall below p10, and about 10 percent land above p90. These are percentiles of the model’s simulations, not a measured hit rate against real results.

In practical terms, p10 is a realistic bad night. It is not the worst case, because rare disasters like an injury in the first quarter are not the typical downside. And p90 is a realistic big night, not a best case. Between p10 and p90 sit about 80 percent of the simulated outcomes.

Because the simulations are only as good as the model, treat the edges of the range as guides, not boundaries. A player can score below p10 or above p90, and real results can land outside the band more often than the simulations suggest.

Why the same projection can be two different bets

Here is a real example from our projections for November 4, 2026. Jabari Smith Jr. was projected for 30.2 fantasy points with a range of 24.9 to 34.9. Paul Reed was projected for 30.1 with a range of 21.5 to 41.1.

The projections are nearly identical. The ranges are not. Smith’s range is about 10 points wide. Reed’s is about 20 points wide. Reed’s bad night is lower and his big night is much higher.

Wide ranges often come from big-man roles with uncertain foul trouble or from players whose minutes depend on game flow. Narrow ranges often come from starters on steady minutes who score in several categories at once.

Neither is better in the abstract. They are different bets. Smith looks like the steadier choice. Reed looks like the higher-variance choice with more upside, and more downside.

Using ranges in cash games

In a cash game you want to finish above roughly the middle of the field. A bad night costs you more than a great night helps you, because once you are above the cutoff, extra points do not matter.

So a higher p10 is useful. If two players have the same projection and similar salary, the one with the higher p10 is usually the safer cash play. You are paying attention to the bottom of the range.

This does not mean ignoring the projection. A player with a high floor but a low projection is still a weak play. The best cash picks combine a solid projection with a narrow range and a good floor.

Using ranges in GPPs

In a tournament you need a score far above the middle. The top of the range matters more. A higher p90 means a better chance of a very large score.

That is why GPP builders often accept wider ranges. Choose players whose p90 is high, then combine several in ways that could all hit on the same night, such as players in a fast game with a high total. A lineup full of low-variance players rarely reaches the top of a big field.

But more variance is not always better. A wide range from a shaky role, such as a bench player with uncertain minutes, mixes real upside with a high chance of a bad night. Look at why the range is wide.

What makes ranges wide or narrow

• Minutes certainty: stable starters tend to have narrower ranges than players with uncertain playing time.
• Role: high-usage scorers vary more with shooting luck than players who rebound and defend.
• Game script: a likely blowout can cut a starter’s minutes and widen his downside.
• Stat mix: players who earn points from rebounds, assists, steals and blocks as well as scoring often have steadier floors.
• Injury risk: a player returning from injury may have a minutes cap and an uncertain range.

How to read a range in practice

Look at three numbers, not one: p10, the projection and p90. If p10 and p90 are close together, the player is predictable. If they are far apart, the player is volatile.

In TipoffLab DFS, every player shows a p10–p90 band next to the projection. You can use it to sort cash candidates by floor and GPP candidates by ceiling.

A simple test when you are choosing between two players with equal projections is to ask what you need from the lineup. If you need safety, take the higher p10. If you need a chance at a very large score, take the higher p90. Writing that question down before you build keeps you from mixing the two goals.

As with everything in DFS, the range is an estimate based on past patterns. It does not promise any particular outcome.

See it in TipoffLab DFS

Every player on the main slate gets a projection, projected minutes, a floor and ceiling, value per $1,000 of salary and projected ownership, in one sortable table that keeps updating until lock.

Player PoolNov 4, 2026 · sample view

Search playersAllPGSGSFPFCFiltersColumns

Player	Salary	Proj	Val/1K	Own%	Floor	Ceil	Proj min
CNikola JokicDEN v TOR	$12,300	61.6	5.01x	13.3%	53.0	67.9	36.0	+ Lock
PGShai Gilgeous-AlexanderOKC v CLE	$10,300	53.1	5.16x	9.5%	46.5	60.1	34.6	+ Lock
PFAnthony DavisWAS v ATL	$8,400	48.4	5.76x	18.5%	40.2	55.5	32.5	Locked
PGJalen BrunsonNYK v CHI	$8,100	46.4	5.73x	19.1%	39.2	51.0	34.8	+ Lock
PGCade CunninghamDET v PHI	$9,300	45.2	4.86x	5.1%	40.0	52.0	33.7	+ Lock
CJoel EmbiidPHI v DET	$9,000	45.1	5.01x	5.2%	38.0	53.6	32.5	Excluded
SFScottie BarnesTOR v DEN	$8,600	43.5	5.06x	11.0%	37.4	51.5	31.3	+ Lock
PGJeremiah FearsNOP v UTA	$6,600	39.7	6.02x	16.3%	31.6	46.8	31.3	+ Lock
SFSaddiq BeyNOP v UTA	$6,100	34.8	5.70x	23.6%	28.2	40.7	32.3	Locked
SGKyshawn GeorgeWAS v ATL	$5,500	30.5	5.55x	15.2%	23.4	37.2	29.5	+ Lock

More on projections

FAQ