10-team versus 12-team fantasy basketball strategy

Up: 2026-27 Fantasy Basketball Encyclopedia

Rebuilt on corrected data 2026-07-18 Every quantitative table below was originally computed from Data/player-projections-2026-07-18.csv, whose FG% and FT% inputs used fabricated shot volume for 429 of 549 players (Projection engine defects and corrections). All of it has been recomputed against Data/player-projections-corrected.csv and the 2026-27 volume-corrected nine-category board. The structural conclusions survived — the replacement gap moved from 0.91 to 0.891 z and the VOR ratio curve is unchanged to two decimals — but the names attached to almost every rank moved, and two recommendations changed materially (centre timing and the punt-FT anchor list). Those are flagged inline.

Correction 2026-07-19: full independent recomputation of every table in this note — replacement levels, VOR curve, sleeper band, waiver-supply counts, centre-eligibility counts — directly from Data/player-projections-corrected.csv and the tier board's position shapes. All reproduce; the July 18 rebuild is confirmed genuine, not cosmetic. Added in the same pass: the ESPN-default points replacement comparison (rank 130 vs 156 from Data/espn-points-board.csv), which had only existed for the Yahoo points model.

Related: 2026-27 volume-corrected nine-category board · Projection engine defects and corrections · 2026-27 fantasy basketball rankings · 2026-27 category rankings · 2026-27 points-league rankings · 2026-27 draft boards — ESPN points and nine-category · 2026-27 fantasy basketball tiers · 2026-27 fantasy basketball player projections · 2026-27 fantasy basketball projection methodology · 2026-27 sleepers, breakouts, and fades · 2026-27 fantasy basketball injury and availability ledger · 2026-27 fantasy basketball evidence and confidence policy · Schedule and streaming desk · ADP and draft market desk

Why this note exists

Everything else in this encyclopedia — the top 200, the tiers, the z-scores in the July 18 projection snapshot — is explicitly calibrated to 12 teams, nine categories, per-active-game value. That is stated in the board contract of the rankings note. If Andrew's league lands at 10 teams, roughly the bottom third of that board becomes free, and a meaningful share of the mid-round advice in the vault stops applying. This is the corrective.

Roster math and where replacement level actually sits

Both major platforms converge on a 13-man active roster. Yahoo's default is ten starters (PG, SG, G, SF, PF, F, two C, two Util) plus three bench and three IL, max roster 16. ESPN's default is also ten starters plus three bench, but with one C and three Util. Yahoo's public leagues default to 10 teams; private and prize leagues default to 12.

So:

\[N_{\text{rostered}} = T \times 13 \quad\Rightarrow\quad 10 \times 13 = 130, \qquad 12 \times 13 = 156\]

Replacement level is the value of the best freely available player, i.e. the player ranked \(N_{\text{rostered}} + 1\). Pulling the corrected nine-cat z-values out of Data/player-projections-corrected.csv:

The replacement gap is \(-0.31 - (-1.20) = 0.891\) standard-deviation-units of total nine-cat value. That is not a rounding error — it is roughly the per-game category gap between a rank-100 player and a rank-130 player. Every single roster spot in a 10-team league starts 0.89 z ahead of its 12-team counterpart, for free.

The gap barely moved under the volume correction (0.91 → 0.891) because the defect displaced players across the replacement band in both directions rather than shifting the band itself. What changed is who sits on it: Christian Braun is no longer the marginal 10-team roster spot, and Jalen Green — 171 on the defective board, free in both formats — is now precisely the last player a 12-team league drafts.

In points format the same computation, using the corrected points model (Yahoo default, no separate 3PM term):

\[\text{FP}_{130} = 26.3 \quad\text{vs}\quad \text{FP}_{156} = 23.9 \quad\Rightarrow\quad \Delta = 2.4\ \text{FP/g}\]

The marginal players are Devin Vassell at 130 and Collin Gillespie at 156.

Over a 13-man roster and a full season, that free 2.4 FP/g floor per slot is the entire margin of most head-to-head weeks.

ESPN's default points model prices the same boundary differently, because it scores shot events directly (FGM +2, FGA −1, FTM +1, FTA −1) rather than folding efficiency into percentages. From the ESPN points board (Data/espn-points-board.csv, per-game column):

\[\text{FP}^{\text{ESPN}}_{130} = 27.0 \quad\text{vs}\quad \text{FP}^{\text{ESPN}}_{156} = 24.1 \quad\Rightarrow\quad \Delta = 2.9\ \text{FP/g}\]

The marginal players are Dillon Brooks at 130 and Jaylon Tyson at 156 — neither of whom sits near the Yahoo-model boundary, because the attempt penalties re-sort exactly the low-efficiency volume scorers who populate this band. The replacement gap itself is slightly wider than Yahoo's 2.4, and the late-round compression is the most extreme of the three models: ESPN points rank 120 (PJ Washington, 27.6 FP/g) is worth 0.6 FP/g over a 10-team replacement but 3.5 over a 12-team one — a 12/10 ratio of 5.83, versus 3.72 in nine-cat and 3.05 in Yahoo points. In an ESPN points 10-teamer, picks past round 10 are almost purely ornamental. One caveat carried over from the ESPN board itself: ESPN H2H points accumulate every game played, so the season-total column, not FP/g, should settle actual draft-day ties there.

The consequence: pick surplus collapses in the late rounds

Value over replacement is \(\text{VOR}(k) = z(k) - z(\text{rep})\). Because the replacement baseline moves but the top of the board does not, league size compresses the late rounds far more than the early ones.

This table is the whole argument, and it is the part of the note the volume correction left almost untouched — every ratio moved by less than 0.25, because the correction reshuffled which player occupies each rank without changing the shape of the value curve. A first-round pick is worth 6% more in a 12-teamer. A round-10 pick is worth 272% more. The points-league version is the same shape (points rank 12: ratio 1.14; points rank 120: ratio 3.05).

Read the corollary carefully: it is not that stars are better in shallow leagues in absolute terms — Jokić is Jokić. It is that in a 10-team league your late picks are nearly worthless relative to free agency, so the fraction of your edge that must come from the top of your roster is much larger. Rounds 12 and 13 of a 10-team draft have VOR near zero by construction: pick 130 is replacement level. Those picks should be pure lottery tickets with zero floor considerations, because their floor is available for nothing all season.

"Sleeper" in one league, undraftable in the other

The 26 players ranked 131–156 are the entire population that changes status between formats. In a 12-teamer they are the last four rounds; in a 10-teamer they are the opening-week waiver wire. Real names from the corrected board, with the defective board's rank shown for comparison:

Gobert is the headline. On the defective board he sat at 99 — a rounds-8-to-9 pick and a plausible 12-team roster staple. Correctly priced against his real free-throw volume he is 138, which puts him outside a 10-team roster entirely and in the last two rounds of a 12-teamer. Three players who were free in both formats on the old board — Jalen Green (171→156), Bennedict Mathurin (157→144) and Davion Mitchell (174→154) — are now inside a 12-team draft.

And the tier just below, free in both formats but which a 10-team manager can churn continuously: Jeremiah Fears (158), RJ Barrett (161, 18.5 pts), Grayson Allen (163), Malik Monk (164), Shaedon Sharpe (169), Jusuf Nurkic (171, 9.0 reb), Brook Lopez (182, 1.2 blk).

A 10-team manager who needs 18 points per game in week 14 can simply add RJ Barrett. A 12-team manager has to trade for it.

Waiver-wire supply, counted

Taking the free-agent pool as ranks 131–220 (10-team) and 157–220 (12-team), and counting how many players clear a useful per-game threshold in each category:

The 10-team wire holds roughly double the specialist supply in every category — the ratio survived the correction intact, and in rebounds and points it widened. That is the mechanism behind every other claim in this note. Note that the single largest addition to the 10-team free-agent pool is Gobert, who now supplies both blocks and rebounds off waivers in a shallow league.

Punting: cheaper to execute, less necessary to attempt

Standard punting guidance — RotoWire's 9-cat guide is representative — describes the two canonical lanes (punt guard stats, punt big-man stats) and stresses that "having a dedicated streaming spot on your roster is a massive advantage," but says nothing about league size. That omission matters, because size is what decides whether a punt is reversible.

The common claim that punting is cheaper in small leagues because roto point totals are smaller does not survive arithmetic. In 10-team roto the max is \(9 \times 10 = 90\) points and average is 5.5 per category; punting concedes ~1.5, a loss of 4.0 points, or 4.4% of the pool. In 12-team, \(9 \times 12 = 108\), average 6.5, loss ~5.0, or 4.6%. Effectively identical.

The real asymmetry is the specialist-supply table above. In a 10-team league a punt is reversible. If you drafted a punt-FT% build around Walker Kessler and Kel'el Ware and the FT% race turns out to be soft, you can add Malik Monk (.860 on 2.1 FTA) and Grayson Allen (.855 on 2.5) off waivers and half-recover the category. Symmetrically, a punt-assists build can stream 6 free players at 4.5+ AST. In a 12-teamer a punt is a commitment: only three free agents clear 4.5 AST, and your opponents will roster them the moment your build is legible.

The correction sharpens this in one direction worth stating. Because the corrected board prices free-throw volume rather than points, the cheap punt-FT bigs are cheaper than the old board showed — Gobert at 138 and Jakob Poeltl at 135 are both late-round or free. A 10-team manager can now assemble most of a punt-FT frontcourt from the waiver wire, which strengthens the "punt tactically and late" recommendation below rather than weakening it.

Net practical guidance: in a 10-teamer, punt tactically and late — draft best-available for eight rounds and let the punt emerge from what you own. In a 12-teamer, punt structurally and early, because you must buy the category concentration at the draft table; there is no second market for it.

Injury absorption

A 13-man roster carrying an injured star loses one active slot. In a 10-team league the marginal cost of that slot is the VOR of pick ~130, which is 0.00 by definition — the replacement is free. In a 12-team league the vacated slot was producing VOR \(\approx 1.22\) (pick 120 level) and the free replacement produces 0. Combined with Yahoo's three IL slots, this is the quantitative case for higher medical risk tolerance in shallow leagues: Stephen Curry (52 projected GP, corrected rank 18), Joel Embiid (44 GP, rank 24), and Anthony Davis (50 GP, rank 23) are all more draftable at 10 teams than the encyclopedia's 12-team board implies, because the cost of the games they miss is a free replacement rather than a real one.

Two of those three got more expensive under the correction — Embiid rose 30→24 and Davis 26→23, both on real free-throw volume the proxy had understated — while Curry fell 11→18 because .920 shooting on 5.0 attempts is worth less than the proxy's implied volume suggested. The direction of the shallow-league argument is unchanged; the prices are not.

Caveat: this is a roster-slot argument, not a games-played argument. It does not make Embiid's 44 games into 60. It only says the opportunity cost of the empty slot is lower.

Positional scarcity, especially center

Counting C-eligible players in the encyclopedia's tier board (position shapes are draft-room archetypes, not platform eligibility guarantees):

On Yahoo's two-center default, a 10-team league needs ~20 starting centers and the top 130 supplies 36 — a surplus of 16. A 12-team league needs ~24 and the top 156 supplies 43 — surplus 19 in raw count, but drawn from a materially worse tail (the last centers rostered are Robert Williams III at 151 on 15 mpg and Morez Johnson Jr. at 155 on 17 mpg, both well below replacement).

The practical difference is not the count but the quality floor. In a 10-teamer you can genuinely wait on center: Myles Turner (117), Jakob Poeltl (135) and Rudy Gobert (138) are all late-round or free. In a 12-teamer, especially on ESPN's single-C setting where C-eligible bigs get funneled into Util anyway, the rebounds-and-blocks supply thins fast enough that securing one of the top ~20 C-eligible players by round 5 is defensible.

Correction 2026-07-18 — centre timing The previous version of this section named Gobert at 98 and Sabonis at 68 as the "rounds 2–3 scarcity insurance" a 10-team manager should decline to pay for. On the volume-corrected board those two moved in opposite directions: Gobert fell 99 → 138 once his .560 FT% was priced against 3.8 real attempts per game instead of the 2.1 the pts/4.8 proxy implied, while Sabonis rose 68 → 58 on efficient interior volume the proxy had understated (2026-27 volume-corrected nine-category board). The recommendation to wait on centre in a shallow league therefore strengthens, but the example changes: Gobert is no longer a rounds-2-to-3 asset anyone would be tempted to reach for — he is a rounds-11-to-12 punt-FT specialist, and in a 10-team league he is a free agent. Sabonis, by contrast, is now genuinely a round-5 price rather than a round-6 one. Recorded rather than silently overwritten because the previous framing was the market belief this desk held on the defective board.

The board changes Andrew should actually make

If 10 teams:

1. Delete ranks 131–200 from your board entirely. Print the top 130 only; everything below is in-season inventory.
2. Weight the first four rounds harder. Break ties toward the higher ceiling, not the higher floor — because your floor is 0.89 z better than the board assumes.
3. Raise availability-risk players by roughly half a tier. Curry, Embiid, Davis, and Tyrese Haliburton's post-Achilles line all get cheaper in expectation.
4. Treat rounds 11–13 as unconstrained lottery tickets: rookies, minutes-battle winners, post-hype swings. Consult 2026-27 sleepers, breakouts, and fades and take the widest variance available.
5. Do not draft a specialist for a category you can stream (blocks, threes, steals). Draft the categories you cannot stream: assists at volume, and elite ratio anchors.

If 12 teams:

1. Use the existing encyclopedia board as written — it is already calibrated here.
2. Pick a punt lane by round 4 and commit. The waiver market will not sell you the category back.
3. Value games played more heavily. A slot lost to injury costs a real \(\sim 1.22\) VOR, not zero.
4. Roster the last useful center earlier than instinct suggests.
5. Round 12–13 picks have real surplus (VOR ≈ 1.2). Take floor over variance there — the opposite of the 10-team instruction.

Either way: confirm before the draft whether the league uses ESPN's one-center or Yahoo's two-center default, and whether IL slots exist. Those two settings move center valuation and injury tolerance more than the 10-versus-12 question does.

Open questions

• The 0.891 z replacement gap assumes all 130 or 156 rostered players are drafted rationally. Real 10-team leagues roster a handful of name-brand players well below the model's top 130, which pushes the true free-agent baseline higher than computed here. The direction of the bias favors 10-team managers even more; the magnitude is unmeasured.
• The projection snapshot's ratio categories are still volume-proxied for many players. Resolved 2026-07-18. Real 2025-26 attempts were recovered for 248 of the 262 pool players and the remaining 14 rookies are flagged estimated; see 2026-27 volume-corrected nine-category board. The residual uncertainty is now concentrated in those 14 rookie rows and in the minutes forecasts that drive the attempt scaling, not in the attempts model itself.
• The corrected board still inherits its games-played and minutes forecasts unchanged, and those drive attempt projections linearly — a player projected 4 mpg too high carries a proportionally inflated FTA and therefore an inflated ratio impact. The replacement-level band, which sits in the 15–25 mpg range where minute forecasts are least certain, is the part of this note most exposed to that.
• Whether the league is H2H or roto matters nearly as much as size and is not modeled here. The streaming advantage of a shallow league is much larger in daily-lineup H2H than in weekly-lineup or roto formats.
• No public source was found that empirically measures how much of a 10-team champion's season value came from in-season adds versus draft picks. The claims above are derived from the encyclopedia's own value curve, not from observed league outcomes.