Fantasy Hoops HQ

Not affiliated with or endorsed by the NBA. Team names are used descriptively.

How it works

Every number on this site comes from a rule you can check. This page explains each one in plain language, with a worked example. The examples are small made-up rosters, and every figure in them is calculated by the same code the tools use.

Counted games

A scheduled game is a night one of your players' teams plays. A counted game is one whose stats actually go into your total. The two are not the same whenever your league has a games limit: once you reach it, games you play after that add nothing, and the players you started on those nights were wasted.

Platforms enforce the limit in different ways, and a single "games per season" number cannot express either of the big two. So the site asks how your league counts, and works out which of your games count under that exact rule. You set the rule once on Settings.

ESPN: a limit per team, per matchup

ESPN caps the games your whole team may use in one matchup. A standard 7-day matchup allows 42, and each calendar day more or fewer moves that by 6, whether or not the NBA plays that day: a 14-day matchup across the All-Star break allows 84, and a 6-day one allows 36.

The part most tools get wrong is what happens on the day you cross the limit. ESPN's rule, quoted:

"If a team starts the day below the limit, the team can go over such limit till all his players finishes playing for the day. If a team starts the day over or at the limit, stats won't count until end of matchup." ESPN Fan Support, Games Played Limit

So the crossing day counts in full, however far past the limit it takes you, and the penalty starts the next morning and lasts for the rest of the matchup.

Some leagues stop counting exactly at the limit instead, mid-day. If yours does, pick Games stop exactly at the cap under games limits on Settings: the counts then never pass the limit, the day that reaches it counts only up to it, and the strategies and the planner start your best games up to the limit and sit the rest.

Worked example

9 players, a 7-day matchup and a limit of 35. On Friday the team starts on 34, 1 below the limit, so all 6 of Friday's games count and the team finishes on 40, 5 past the limit. From Saturday the team starts the day at or over the limit, so nothing counts for the rest of the week.

ESPN limit of 35, example week
DayCounted at startPlayers playingCounts?Counted so far
Monday09 All 99
Tuesday98 All 817
Wednesday179 All 926
Thursday268 All 834
Friday346 All 640
Saturday405 None, 5 wasted40
Sunday407 None, 7 wasted40
Week5240 counted, 12 wasted40

This is why every tool here plans with counted games rather than scheduled ones: a pickup who only plays on Saturday adds nothing to this team's week.

Yahoo: a maximum per position, per season

Yahoo caps games by position, over the whole season. A public Rotisserie league defaults to 82 games at each position, head-to-head leagues have no maximum unless the commissioner sets one, and a game counts as played after one second on the floor (Yahoo Help, Position maximum usage ).

A game is charged to the position slot a player starts in, and only one game per slot per night can count. Once a position reaches its maximum, the rest of that position's season counts nothing. A player eligible at two positions is charged to the first one listed, and the site tells you so, because moving him changes which position fills first.

Worked example

Two centres share your one C slot. Over 90 nights at least one of them plays, and on 18 of those nights both do. That is 108 games between them, but with one slot only 90 can be started. The C maximum is 82, so the position fills on Jan 10 and the last 8 games do not count.

Leagues with no limit

No limit is a real answer, not a very large number. In a league without one every scheduled game counts, so the site counts them and shows no limit warnings. The ESPN example week above, in a league with no limit: 52 scheduled, 52 counted, 0 wasted.

Sleeper, Fantrax and CBS default to no limit here, because we have not verified their rules. If your league sets one, enter it on Settings and it is applied.

Category values

A player's value in a category says how far his per-game average sits from the average player in the pool, measured in that category's spread (its standard deviation). This is a z-score: 0 is average, +1 is one spread better, and the values can be added across categories because each one is on the same scale. Turnovers count against you, so fewer turnovers is a positive value.

Values are worked out the same way in every window: on past per-game averages, or on our rest-of-season projection, which is the default. How the projection is made, and how well it would have predicted 2025-26.

Worked example

The centre, against a pool of 6 made-up players, on the 2025-26 window. The real pool is every player with a line, several hundred of them.

The centre: per-game value by category
CategoryHis numberPool averageSpreadValue
PTS16.014.674.820.28
REB11.05.832.542.03
AST2.03.172.67−0.44
BLK2.20.800.682.07
TO (counts against)1.61.630.860.04
FG% (made shots added)1.300.000.731.79
Total value5.77

Value is (his number minus the pool average) divided by the spread. Rebounds: (11.0 − 5.83) ÷ 2.54 = 2.03. The rankings sort every player by this total, in the window you pick.

Punts and targets

Most category leagues are won by choosing what to give up. When you punt a category it weighs nothing in a player's value; a target weighs double; every other category weighs one. Weights: target 2, in play 1, punt 0.

Worked example

The centre is below average in AST (−0.44). Balanced, his value is 5.77. Punt AST and that category drops out, so his value rises to 6.21. A punt build lifts the players who are weak only where you have stopped competing.

Why FG% and FT% are weighted by attempts

A team's FG% is its total made shots over its total attempts, so a player who shoots a lot moves it more than one who shoots a little, at the same percentage. The value of a percentage is therefore the made shots a player adds over a league-average shooter taking the same attempts: (his percentage minus the league's) times his attempts.

Worked example

A team shooting 40 of 90 (44.4%) adds one of two 50% shooters. The league shoots 46%.

Same percentage, different volume
PlayerShootingTeam FG% afterMade shots added
Low volume2 of 444.7%0.16
High volume8 of 1645.3%0.64

Both shoot 50%, but 8 of 16 adds 4 times as many made shots over a league-average shooter as 2 of 4, and lifts the team's FG% further. The same rule applies to FT%, 3P% and assists per turnover.

Stat windows

A window is the stretch of games a per-game average is taken over. You pick one window and every ranking on the site uses it. Each window is scored on its own, against the pool in that same window.

The windows
WindowWhat it isMin games
ProjectionOur forecast for the rest of the season, per game: past seasons weighted and aged, blended with this season, the last 14 days and current minutes. The default. Method and backtest on the projections page.0
2026-27 so farThis season's games so far. A player with fewer than 5 games is shown on his 2025-26 line instead, and the row says so.5
Last 14 daysOnly the last two weeks of games: who is hot or cold right now, on a small sample. The same under-5 rule applies.0
2025-26 seasonLast season, start to finish.20
3 seasonsThree seasons together, so one injured or lucky year weighs less.40

Two windows are never blended into one number. A player is usually available because something changed, and a blend hides exactly that.

Worked example

The returning forward averaged 20 points over three seasons but 11 in 2025-26, after an injury. In points his value is 0.82 over three seasons (rank 2) and −0.76 in 2025-26 (rank 5). The gap is 1.58, more than the 1.0 the site uses as its cut-off, so wherever the two windows sit side by side he is marked as a split rather than averaged away.

Safe, Swing and Lost

During a matchup each category gets a mark, read off its win chance, so the mark and the percentage beside it always agree.

  • Safe95% chance or better
  • Swingbetween 5% and 95%
  • Lost5% chance or worse

Win chance

Each side's final is its score so far plus its remaining counted games times its per-game averages. Then the question is how far the rest of the week could wander from that. The spread grows with how much is still to play, a little more than pure chance would say, because box scores run hot and cold. The win chance is the gap between the two finals measured in that spread, on a smooth curve: one spread ahead wins about 85% of the time.

The chance of winning the matchup combines every category's chance, counting the ways you can win more categories than you lose. Categories are treated as independent.

Worked example

Thursday morning in a three-category matchup:

Score so far, rest of week, chance and mark
CategoryYou so far + restOpp so far + restGapSpreadWin chanceMark
REB180 + 60 = 240140 + 55 = 1954513.6over 99% Safe
STL24 + 9 = 3322 + 10 = 3214.659% Swing
BLK8 + 6 = 1421 + 9 = 30−164.4under 1% Lost

Rebounds are settled: a 45 lead is 3.3 spreads. Steals project to a 1-steal lead with 19 still to come, a 59% chance. Blocks are gone. Your chance to win the matchup is 59%, and the expected score is 1.6 to 1.4: the week turns on steals.

Strategies

A strategy is a plan turned into a lineup: the nights each player sits. Nothing else changes. The same counted games rules then decide what counts, so a strategy can only move games, never invent them, and Start everyone is exactly the scoreboard's own projection.

Each strategy ranks players by their per-game line against the average of both rosters, with its own weight on each category. FG%, FT% and 3P% count as shots made above the room's rate, so a 40% shooter's volume counts against FG%. Then, from the next night still to play: the lowest-ranked players sit when more play than your lineup has spots, and under an ESPN-style cap the strategy picks the night to cross it that buys the most, the best games before it and everyone on it.

My best players is the site's own ranking, every category weighing the same. The three matchup plans are built for this opponent: every team is strong in some categories and weak in others, so the board ranks the categories by your chance of winning each against their best players, takes the majorities with the best joint chance, and turns each into a lineup that goes after those categories and lets the rest go. The other side gets three of its own, read from its side. My ranking plays the ranking you made for this matchup with Plan with AI on the Matchup page: its weight on each category and its nudge on each player.

The board scores every strategy of yours against every one of theirs with the win chance. The recommended one is the strategy whose worst case is best, because the other manager can change plans too. On the matchup, the button under each name is that side's strategy: a ready one, or the categories that side should win.

Try in tradeCompare

Per-game numbers in the named window. Games are this season's schedule under your league's games limit.