2026-27 Yahoo High Score board

Up: 2026-27 Fantasy Basketball Encyclopedia

Related: 2026-27 volume-corrected nine-category board · 2026-27 points-league rankings · Points leagues versus category leagues · 10-team versus 12-team fantasy basketball strategy · Streaming and in-season roster management · Punt strategies for nine-category leagues · Projecting games played and availability · Positional scarcity and eligibility in 2026-27

Why this note exists

The rest of this encyclopedia is built for nine-category leagues. If Andrew joins a Yahoo public league and touches nothing, he will not be playing nine-cat. He will be playing High Score, which Yahoo made the default format for public leagues in 2025-26, and roughly half of the encyclopedia's analytical machinery — FG% leverage, FT% punts, turnover discipline, week-long accumulation — becomes inert. This board is the translation layer.

The rules, verified

Checked against Yahoo's own help page (SLN37162), the NBA.com explainer, and RotoWire's write-up. All four rules given in the brief survive contact with the sources.

Two corrections and one expansion to the brief:

1. Lineups are not literally locked for the week. Yahoo's help page says lineups can be changed daily, but with a punitive asymmetry: benching a player at any point during the week permanently deletes his high score from that week's total, and reactivating him only counts games played after he returns. So the format is functionally set-and-forget. The correct mental model is not "you may not touch the lineup" but "touching the lineup is almost always a mistake." The one legitimate use is benching a starter who has already been ruled out for the remainder of the week and whose games so far were duds — you sacrifice a bad high score for a replacement's chance at a better one.
2. Roster and team-count settings are commissioner-adjustable in private leagues (2–20 teams). The 10-man / 6-starter structure is the default, which is what matters here, but a custom league may differ. Verify before drafting.
3. The Flex slot is variously called "Flex," "Utility," or "sixth man" across Yahoo's own materials. It is position-agnostic.

The board

\[\text{HS} = \text{PTS} + \text{REB} + 2\,\text{AST} + 3\,\text{STL} + 3\,\text{BLK}\]

computed per game from the encyclopedia's 2026-07-18 projections, then a ceiling estimate (method below). The 9-cat column is corrected_rank from 2026-27 volume-corrected nine-category board; Δ is how many slots a player gains moving from nine-cat to High Score. Players with no nine-cat rank fall outside that board's 262-player universe.

Who moves, and why

Of the top 150 in High Score, 135 also appear in the nine-cat top 150 — the population is broadly the same. But 61 of those 150 move by 25 or more slots, and the movement is systematic, not noise. Three mechanisms drive all of it.

(a) Turnovers vanish

Nine-cat taxes every possession a creator ends badly. The top 120 High Score players average 2.26 turnovers per game, and in a nine-cat z-score frame that is a real drag on exactly the players who dominate everything else. High Score deletes the tax entirely.

The clearest beneficiary is Ja Morant, who moves from 9-cat 129 to High Score 28, a 101-slot gain — the largest in the board. He projects 7.3 assists against 3.2 turnovers on a 71% free-throw stroke; nine-cat punishes him twice for the same aggression, High Score not at all. Stephon Castle (111 → 37), Julius Randle (115 → 41), and Paolo Banchero (84 → 17) are the same story: high-usage, high-turnover, mediocre-efficiency forwards and guards who are nine-cat compromises and High Score cornerstones.

(b) Assists are worth double rebounds

The 2× multiplier on assists is the format's structural tilt toward playmaking. Decomposing High Score into its components makes it visible:

Josh Giddey gains 41 slots (53 → 12) almost entirely on this axis. Alperen Şengün gains 44 (59 → 15) as a passing big who is neither a rim protector nor an efficient volume scorer. Trae Young gains 23 (33 → 10) despite contributing essentially nothing on the glass. Meanwhile the pure rebound-and-block bigs slide: Kel'el Ware falls 72 slots (52 → 124), Mark Williams 54 (79 → 133), Daniel Gafford 53 (92 → 145).

Note the countervailing force: steals and blocks are worth 3 each, which is generous in absolute terms. Gobert still gets 15% of his score from blocks. The problem is that 4.5 points of block value cannot compensate for 17 points of missing assist value. Defensive specialists are not worthless here; they are simply capped, and the cap sits around 30–38 HS/g.

(c) No percentage categories — the big one

This is the deepest change and the least intuitive. Nine-cat treats FG% and FT% as two of nine categories, weighted by attempt volume. High Score does not care whether a shot went in except insofar as it produced points. Two consequences:

Bad free-throw shooters are fully rehabilitated. Giannis Antetokounmpo is the canonical case: 66.5% from the line on 11.6 attempts per game, which in nine-cat is the single most destructive FT% profile in the league and drags him to corrected_rank 46. In High Score he is the 4th-best player in fantasy basketball, at 57.5 HS/g — a 42-slot gain and, functionally, a first-round pick instead of a fourth-round punt-anchor. Zion Williamson gains 50 (95 → 45) on the same logic plus turnover relief. Rudy Gobert (56% FT) gains 21, and Nic Claxton (62.5%) gains 13 — smaller gains because they lose more to mechanism (b) than they win back here.

But the "bad FT shooters rise" heuristic is not reliable on its own, and two commonly-cited examples fail it. Ivica Zubac moves only 73 → 77 — essentially flat — because his 69% comes on just 3.0 attempts, too little volume for nine-cat to have punished him much in the first place. And Jalen Duren falls from 28 to 55. His 75.5% on 6.4 attempts is not actually bad; he was never seriously FT-damaged, so he collects no rehabilitation bonus while still paying the assist-weighting penalty. The rule is not "bigs rise" — it is "bigs whose nine-cat value was suppressed by FT% rise, and bigs whose nine-cat value came from percentages and blocks fall."

Elite shooters lose their moat. Lauri Markkanen (89% FT, 2.8 threes) falls from 10 to 50. Derrick White (89.5%, 2.7 threes) falls from 32 to 68. Kristaps Porziņģis falls from 38 to 79. And note that three-pointers are not a category at all in High Score — a made three is worth 3 points, the same as any other 3 points. The entire "3PM specialist" archetype that nine-cat rewards (Mikal Bridges 51 → 99, Cameron Johnson 118 → 159, Reed Sheppard 69 → 105) is simply deleted as a value source.

The single largest faller among genuine nine-cat elites is Walker Kessler, 8 → 61. He is a top-10 nine-cat asset because blocks and FG% are two categories he wins single-handedly. High Score pays him for blocks (18.8% of his score) but nothing for efficiency, and he offers 12.9 points and 4.2 assist-points. 38.2 HS/g is a good sixth starter, not a first-rounder. Chet Holmgren (19 → 62) and OG Anunoby (35 → 93) fall for the same reason.

Ceiling is everything; consistency is worthless

This is the part that most High Score analysis gets wrong, and it inverts a decade of fantasy instinct.

Because only the best game counts, a player's average is not what you are buying. You are buying the expected maximum of his weekly game sample. A player averaging 30 HS with genuine 55-point nights beats a player averaging 34 who has never cleared 40 — the second player's floor, the thing that makes him valuable in every other format, is a stat you are literally never paid for. Every game except one per week is discarded.

Estimating a ceiling from a per-game projection

Model each game's High Score as a draw from a distribution with mean \(\mu\) (the projection) and standard deviation \(\sigma\). If a player plays \(n\) games in a week, the expected weekly contribution is the expected maximum of \(n\) draws:

\[\mathbb{E}[\text{week}] \;\approx\; \mu + \sigma \cdot \mathbb{E}[M_n], \qquad \mathbb{E}[M_n] = \int_0^\infty \!\!\big(1-\Phi(x)^n\big)dx - \int_{-\infty}^0 \!\!\Phi(x)^n dx\]

with \(\Phi\) the standard normal CDF. For an 82-game schedule across roughly 25.5 weeks, a fully healthy player gets \(n \approx 3.2\) and \(\mathbb{E}[M_n] \approx 0.89\); at \(n=3.0\) it is \(0.85\), at \(n=4.0\) it is \(1.03\). So the weekly bonus above a player's average is just under one standard deviation — and \(n\) enters only logarithmically, which is why availability matters much less here than in accumulation formats.

\(\sigma\) is built from components, treating each stat's game-to-game dispersion separately and then applying the scoring weights:

\[\sigma^2 \approx \sigma_{\text{pts}}^2 + \sigma_{\text{reb}}^2 + 4\sigma_{\text{ast}}^2 + 9\sigma_{\text{stl}}^2 + 9\sigma_{\text{blk}}^2\]

using \(\sigma_{\text{pts}}\!\approx\!0.42\mu_{\text{pts}}\), \(\sigma_{\text{reb}}\!\approx\!0.45\mu_{\text{reb}}\), \(\sigma_{\text{ast}}\!\approx\!0.50\mu_{\text{ast}}\), and Poisson-like \(\sigma\!\approx\!\sqrt{\mu}\) for steals and blocks, then inflating the total 10% for the positive covariance between counting stats within a game (big nights are big across the board). These dispersion coefficients are assumptions, not measurements — they are consistent with typical NBA game-log spread but were not fitted to 2025-26 logs, and that is the weakest link in this note.

The output: coefficient of variation lands near 0.30 across every rank band (0.297 for ranks 1–25, 0.301 for 101–150), so the ceiling adjustment is roughly proportional and does not wholesale reorder the board. The mean is a decent proxy for the ceiling. This is the honest finding, and it cuts against the loudest strategic advice in circulation.

Where ceiling actually does reorder things

Two places, both real.

Availability, and less than you would think. Jimmy Butler III at 30 projected games falls from HS 85 to ceiling-rank 132 — with \(n \approx 1.2\), he barely gets a maximum at all, and in many weeks contributes nothing. Anthony Davis (50 GP) falls 11, Joel Embiid (44 GP) falls 10. But notice how modest those falls are relative to what they would be in a nine-cat accumulation league. A 50-game player loses roughly 39% of an 82-game player's counting stats in nine-cat; here he loses about 39% of his weeks entirely but each week he does play is nearly undiminished. High Score is materially more forgiving of injury-prone stars than any accumulation format. Embiid at ceiling-rank 31 is a defensible pick.

Iron men with high variance gain a little. Payton Pritchard (78 GP) gains 9, Kon Knueppel gains 8, Desmond Bane gains 8. Small, but free.

The larger, unmodelled version of this effect is usage volatility that a per-game projection cannot see. A projection gives you \(\mu\); it does not tell you that a player's minutes swing between 22 and 38 depending on matchup, or that his team plays at a pace that produces 110-possession blowouts. Fast-paced offenses and high-usage-when-on-court roles generate fatter right tails than the component model above assumes. Where two players tie on HS/g, take the one on the faster team, the one whose role is more volatile, and the one who is his team's clear first option on nights the other star sits. Injury news that promises a one-week usage spike is worth more here than in any other format, because the spike only has to produce one game.

And the bench becomes lottery tickets

The four non-starters are not depth — they are options. A bench player only ever matters if he is promoted into a starting slot, and once promoted, only his best game counts. A reliable 26 HS/g role player and a 20 HS/g player with 45-point outbursts are both nearly useless as bench pieces, but the second one at least has a scenario where he wins you a week. Stash volatility, not floor. This is the exact opposite of the nine-cat instruction in Streaming and in-season roster management.

Roster math: this league is shallow

A 10-team league rosters 100 players; a 12-team league rosters 120. Nine-cat leagues with 13-man rosters plus IL slots roster 170–200. That difference is enormous and it reshapes the draft.

Replacement level in High Score sits at roughly:

Three consequences:

1. Waiver wire quality stays high all season. In a 10-teamer, a player projected around 30 HS/g — a legitimate starter in a deeper format — is sitting in free agency in week 1. Injuries to starters are far less catastrophic, and the value of hoarding depth collapses accordingly.
2. The gap between a good and bad roster is concentrated in the top 60. Mean ceiling for High Score ranks 1–6 is 72.5; for ranks 55–60 (the last starters a 10-team draft consumes) it is 47.4; for 115–120 it is 36.5. The distance from an elite starter to a marginal one is ~25 points a week per slot; the distance from a marginal starter to waiver fodder is ~11. Stars are worth disproportionately more than in a format where you start 10 players.
3. Punting is impossible and unnecessary. There are no categories to punt. Every player contributes to one number. This deletes the entire strategy layer in Punt strategies for nine-category leagues — do not bring it to this format.

How to draft High Score

• Use this board, not the nine-cat board. They disagree by 25+ slots on 61 of the top 150. That is not a rounding error; it is a different game.
• Take the biggest raw producers early. Rounds 1–2 should be Jokić, Dončić, Giannis, Wembanyama, Cunningham, Gilgeous-Alexander — high-usage stat-sheet fillers, no efficiency discount, no turnover discount.
• Buy the FT%-damaged. Giannis at his nine-cat ADP is the single largest arbitrage in the format. Anyone whose public ranking is depressed by free-throw shooting is mispriced if the room is using nine-cat lists.
• Fade the specialists. Elite-percentage, three-point-heavy, block-only players are systematically overpriced by nine-cat consensus. Kessler, Markkanen, Holmgren, Bridges all cost more than they return.
• Prioritize assists. Second-tier playmakers (Giddey, Şengün, Morant, Castle) go 40–100 slots later than their High Score value.
• Tolerate injury risk more than you are used to. The best-game rule means a star's games are nearly full-value; only his missing weeks cost you. Embiid, Davis, and Kawhi are worth more here than anywhere else. Butler at 30 games is still too far.
• Draft the last four picks as ceilings. No floor, no handcuffs, no "safe veteran." Take young players with unstable roles on fast teams.
• Then leave the lineup alone. Set it before the week and do not touch it unless a starter is definitively out with a bad score already banked.

Open questions

• The dispersion coefficients (\(0.42\mu\) for points, etc.) are assumed rather than fitted. Fitting \(\sigma\) from actual 2025-26 game logs would materially improve the ceiling column, and would likely increase the spread in CV across archetypes — defensive specialists whose value rides on 3×-weighted steals and blocks should be more volatile than the current model says.
• The component model assumes rough independence between stats within a game and corrects with a flat 10% inflation. Real within-game covariance is probably higher for volume scorers (minutes drive everything) and lower for block specialists.
• Does Yahoo's default High Score league use head-to-head weekly matchups with a playoff bracket, and if so what are the playoff weeks? Not confirmed here, and it affects whether to draft for regular-season floor or playoff-week schedule.
• Are there IL slots in the default 10-man roster, and do they sit outside the 10? This meaningfully changes the injury-tolerance advice above.
• The corrected_rank universe covers 262 players; 15 of the High Score top 150 fall outside it, so their Δ is unknown. Worth extending the nine-cat board to 400 to close the gap.
• Unverified: whether games played in a week are affected by the All-Star break and other short weeks, where \(n\) drops to 1–2 and the ceiling premium shrinks toward zero.