Guide
Last 7, 14, 30 days or season: which fantasy basketball stat window should you trust?

Short answer
Trust short windows (7 to 14 days) to spot role changes and waiver targets, and season or multi-year averages for draft value and trades. Believe a short window only when minutes and shot volume moved too, because percentages need hundreds of attempts to settle. Never blend two windows into one number.
What is a stat window?
A window is the stretch of games a per-game average is taken over. You pick one, and every ranking that uses it is scored against the players in that same stretch. A wing who scored 28 points twice this week looks very different on the last 7 days than on last season. Both numbers are true. They answer different questions.
Most tools offer some mix of four kinds of window:
| Window | Question it answers | Games in it (roughly) |
|---|---|---|
| Last 7 days | Is this player hot right now? | 3 or 4 |
| Last 14 days | Has a change held for two weeks? | 6 to 8 |
| Last 30 days or season-to-date | Is the new role the story of the season? | 12 to 15 and up |
| Last season or several seasons | What is this player worth over a full year? | 60 to 82 per season |
The game counts are arithmetic, not a schedule: an NBA team averages a little over three games a week. Hashtag Basketball’s Ranking Trends page, for example, filters by the last 7, 14 or 30 days. Fantasy Hoops HQ ships four windows of its own: season-to-date (this season so far), last 14 days, last season (2025-26), and three seasons.
How much of a short window is noise?
Less of a box score is noise than you might think, and more of a shooting line is.
Minutes are the most stable thing in basketball. Kostya Medvedovsky’s study of stabilization rates found minutes “extremely stable”, because they are a coach’s decision rather than luck. Counting stats follow minutes and shot attempts, so a player whose minutes jumped is usually a different player.
Percentages are the opposite. Medvedovsky estimates that a player’s three-point percentage needs about 240 attempts before it is half their own and half league average. Darryl Blackport’s earlier study put the point where skill outweighs noise at roughly 750 attempts. A player who takes six threes a game reaches 240 attempts after about 40 games. A two-week window holds maybe 40 attempts.
A worked example
Medvedovsky’s padding method makes this concrete: add 240 attempts at league average to whatever a player has done, and the result is a fair estimate of their true rate. The league made 36.0% of its threes in 2025-26, per Basketball Reference.
Suppose a made-up wing goes 22 of 42 from three over the last 14 days. That is 52%. The padded estimate is (22 + 240 × 0.360) ÷ (42 + 240) = 108.4 ÷ 282 = 38.4%.
| Line | Makes | Attempts | 3P% |
|---|---|---|---|
| Last 14 days, raw | 22 | 42 | 52.4% |
| League average, as padding | 86.4 | 240 | 36.0% |
| Padded estimate | 108.4 | 282 | 38.4% |
Two hot weeks moved a fair estimate of the player’s shooting from 36% to about 38%, not to 52%. If their minutes also went from 24 to 32, the extra threes per game are real even when the percentage cools off, because they are taking more shots. If the minutes stayed flat, most of the jump is a heater.
Which window should you trust for each decision?
| Decision | Window to lead with | Check against |
|---|---|---|
| Draft | Last season or three seasons | The other of the two |
| First two weeks of the season | Last season | Season-to-date minutes |
| Waiver add or stream | Last 14 days | Last season, minutes and attempts |
| Trade | Season-to-date | Last season or three seasons |
| Weekly matchup | Your ranking window | Games played this week |
| Playoff push | Last 14 days and season-to-date | Minutes, and who is injured ahead of them |
The draft
Use last season or three seasons. There are no current games yet. Three seasons dampens one injured or lucky year; last season captures the most recent role. Fantasy Hoops HQ’s draft cheat sheet only ranks on finished seasons for exactly this reason.
Early season
Season-to-date is thin for the first couple of weeks. On Fantasy Hoops HQ, a player with fewer than five games this season is shown on their 2025-26 line in both season-to-date and last 14 days, and the row says so. Treat early numbers as news only when minutes moved with them.
Waivers and streaming
Start with the last 14 days to find movers, then ask three questions before you claim:
- Did minutes rise? From 18 to 30 minutes is a role change. From 28 to 30 with a hot shooting night is noise.
- Did shot attempts and assists rise? Usage moves with the ball. A quiet scorer who suddenly takes 15 shots a night has a new job.
- Is the percentage believable? Pad it toward league average, as in the example above.
If the short window and last season disagree and the minutes did not move, believe last season. If the minutes moved, believe the short window for counting stats and stay skeptical on percentages. Our waiver wire strategy covers the claim itself.
Trades
A trade is for the rest of the season, so a two-week window is context, not a price. Lead with season-to-date once it has a real sample, and check last season. The trade value guide covers what each category is worth.
Matchups and the playoffs
Use the same window you rank on, then let games played do the rest. A player with four games this week puts more counting stats on the board than one with two, even on the same average. The schedule grid has the counts for every week, and the playoff weeks page covers the weeks that decide the title.
Why Fantasy Hoops HQ never blends windows
A free agent is usually available because something changed: an injury, a trade, a minutes cut. Multi-year averages are therefore too kind to exactly the players you would add. A single number that averages a weak recent season with two strong older ones hides the one thing you needed to see.
So the site ranks each window on its own, against its own pool, and shows two side by side where it helps. When last season and three seasons disagree by more than a set cut-off, the player is marked as a split instead. The stat windows section works through one made-up player whose value drops from near the top of a small pool to near the bottom between the two windows.
You pick a window once and it carries through: rankings, the category plan and the free-agent lists all use it. Fantasy Hoops HQ has no projections. Every number is a historical per-game average in the window you chose.
| Window on Fantasy Hoops HQ | Fewest games for a ranked line |
|---|---|
| Season-to-date | 5 |
| Last 14 days | 0 (under five games this season shows the 2025-26 line) |
| Last season (2025-26) | 20 |
| Three seasons | 40 |
Common mistakes
Chasing the last 7 days with no minutes change. Three or four games on the same minutes tell you almost nothing.
Dropping a stable starter for a heater. You can get the hot player back cheaply if they cool off. A proven starter, once dropped, is gone.
Drafting from a short window. Draft boards need full seasons. Short windows are for in-season adds.
Ignoring season-to-date in January. By midseason it is often the best single window: big enough to be stable, recent enough to include role changes.
A five-minute weekly check
- Open rankings on last 14 days and note who moved up.
- For each name you do not own, compare minutes with last season.
- If minutes moved, look at whether their games fill your empty nights.
- If only the shooting moved, leave them on the wire.
- Before you claim, switch to season-to-date once and confirm the story holds.
Short windows find the news. Long windows keep you honest.
Common questions
What is a stat window in fantasy basketball?
A window is the stretch of games a per-game average is taken over, such as the last 14 days or last season. Rankings, free-agent lists and matchup plans use one window at a time, scored against the players in that same stretch.
Which stat window should I use for the waiver wire?
Start with a short window (last 7 or 14 days) to find players whose role changed. Then check minutes and shot attempts against last season, so you are not chasing a three-game hot streak with nothing underneath it.
Which window should I use for trades?
Season-to-date once it has a real sample, checked against last season or three seasons. A trade is for the rest of the year, so a two-week window should only be context.
Does Fantasy Hoops HQ blend last 14 days with last season?
No. Each window is ranked on its own against its own pool. Where last season and three seasons disagree by more than the site's cut-off, the player is marked as a split rather than averaged away.
Sources
Checked on .
- Hashtag Basketball: Fantasy Basketball Ranking Trends (last 7, 14 or 30 days)
- Kostya Medvedovsky: NBA Stabilization Rates and the Padding Approach
- Darryl Blackport (FanSided): How Long Does It Take For Three Point Shooting To Stabilize?
- Basketball Reference: NBA league averages per game (2025-26 3P%)