Celtics PNR / DHO Action Analysis (REPAIRED)

Generated 2026-05-19. This report supersedes the 14:52 partial attempt (now under reports/superseded/). Authoritative 2025-26 NBA Synergy data was retrieved this session. All Boston team values, ranks, percentiles, and league baselines come from the official stats.nba.com/stats/synergyplaytype endpoint via nba_api 1.10.2.

1. Executive verdict

The RealGM claim - "Boston uses DHOs and skip passes more than straight PNR; Queta is pretty good at it; they should have made Embiid defend that action every time down" - is factually wrong on the DHO-vs-PNR framing, strongly correct on Queta, and partially correct on the Embiid prescription.

The data:

• Boston's PNR ball-handler is 18.3% of their offense in the regular season (4th of 30) and 20.3% in the playoffs (1st of 16). Handoff is 5.3% RS (8th) and 3.4% playoffs (15th of 16). PNR is roughly 3-6x more common than DHO in Boston's actual offense.
• Queta is a 96th-percentile PNR roll-man in the playoffs at 1.786 PPP on 92.3 FG% (2.0 possessions/game). He was 86th percentile in the regular season at 1.348 PPP. The "pretty good at it" line understates what the numbers say.
• Boston DID lean into PNR roll-man more in the playoffs (5.8% → 6.8% frequency, rank 9th → 4th, PPP 1.136 → 1.196). They could plausibly have run it more, especially given Queta's playoff efficiency. So the prescriptive part of the claim is at least defensible, but its premise (DHO over PNR) is not.

2. What was wrong with the prior attempt

The earlier reports/celtics_pnr_dho_action_analysis.md accepted the NBA Stats endpoint failure and fell back to documented public norms. Root cause: the prior session sent TypeGrouping=null. The endpoint requires TypeGrouping="offensive" to return non-empty rowSets. The other small issue was casing: the play-type string is PRRollman (lower-case m), not PRRollMan. Once both were fixed, 44/44 (season_type, play_type, team/player) combinations returned full data on the first try.

The prior outputs are preserved under data/superseded/ and reports/superseded/. A supersede note explains why at reports/celtics_pnr_dho_action_analysis_SUPERSEDED_NOTE.md.

3. Retrieval methods used

Four agent lanes ran in parallel:

Full audit at data/celtics_action_type_fetch_log.csv (51+ rows). Raw payloads at data/raw/action_types/nba_api/ (44 JSON files + SUMMARY.md).

4. Actual data retrieved

Full team-level 2025-26 Regular Season AND Playoffs for all 11 Synergy play types (PRBallHandler, PRRollman, Handoff, Isolation, Spotup, Transition, Cut, OffScreen, Postup, OffRebound, Misc). 30-team league table (RS) and 16-team league table (Playoffs).

Player-level rows for Boston for all 11 play types with each rotation player's possessions, PPP, percentile, and frequency share of that player's own offense.

Pace and ratings from Basketball Reference (verified earlier this session): Boston 94.8 pace (30th of 30, slowest), 120.8 ORtg (2nd of 30), 112.7 DRtg (4th), Net +8.1 (4th). League average pace 99.4, ORtg 115.8.

5. Team play-type profile

Regular Season (n = 30 teams, BOS GP = 80)

Playoffs (n = 16 teams, BOS GP = 7)

What this table actually says

• Boston is a heavy PNR-handler team in BOTH seasons, not a DHO-first team. Frequency rank: 4th of 30 regular season, 1st of 16 in the playoffs. That kills the RealGM claim's framing.
• Handoff was a regular-season strength turned playoff weakness. Frequency 8th of 30 RS → 15th of 16 playoffs. PPP 0.957 → 0.885. Boston ran DHO LESS in the playoffs, not more.
• PNR Roll Man went UP in the playoffs in both frequency (5.8% → 6.8%) and PPP (1.136 → 1.196). The team made an adjustment, just maybe not big enough.
• Transition is structurally low - 30th of 30 in regular-season frequency. Pace explains it (Boston 30th of 30 in possessions/game). Quality is fine (83rd percentile).
• Postup in the playoffs went 100th percentile in PPP at 1.261 - the most efficient post-up team in the playoff field. Largely Brown posting (his playoff post-up PPP was 93rd percentile, 1.167).

6. League and pace context

Boston was the slowest team in the league. Frequency shares (POSS_PCT above) are pace-neutral - they're shares of the team's own offense - so the rankings above are already pace-adjusted. Raw per-game counts (POSS in the team CSV) would understate the action diet relative to faster teams; that's not what the rank/percentile columns are doing.

7. Regular season vs playoffs

The single biggest shift in Boston's profile from RS to playoffs:

The handoff frequency drop is the single most striking number on the page. In the playoffs Boston used DHO less than every team except Minnesota while running PNR ball-handler more than any other playoff team. This is the opposite shape of the RealGM claim.

8. Player play-type profile

PNR Ball Handler (Regular Season)

PNR Ball Handler (Playoffs)

Read: Tatum and Brown ran the most PNR-handler reps in the playoffs. Brown's PPP fell to 36th percentile (an additional data point for the sell-high analysis - his loss-side profile is real). Pritchard was the most efficient PNR-handler on the roster in the playoff series.

PNR Roll Man (Regular Season)

PNR Roll Man (Playoffs)

Boston has TWO elite roll-men (Queta and Garza), and one inefficient late-acquisition roll-man (Vučević) who actually got more roll-man reps than Queta in the playoffs by frequency-of-his-own-offense (29.2% each, but Vučević's raw PPP was less than half of Queta's).

Isolation (Regular Season)

Tatum's RS isolation PPP at 13th percentile is a counter-narrative data point: his iso reps were not productive in 2024-25's lead-in or his 16-game return.

Handoff (Regular Season)

Boston's DHO is a White / Pritchard / Brown action. None are elite in PPP terms. Hauser is the most efficient at it on tiny volume.

9. Queta profile

Synergy share of his own offense (Regular Season)

Synergy share of his own offense (Playoffs)

Read: in the playoffs Queta's diet narrowed to two actions - roll-man and cut - both elite. There is no efficient post-up or spot-up in his profile; using him those ways is wasted. Boston's playoff coaches understood this enough to keep his diet focused.

Queta per-36 (BBR totals, 1927 MP across 76 GP)

Low usage, elite finishing, elite rebounding, above-average rim protection, below-average creation. Pure roll-and-finish / cut-and-board center archetype, in the exact mold the RealGM claim implied.

10. Philly / Embiid applicability

The series went 7 games, Boston lost 4-3. From the playoff team table above, Boston actually DID lean into PNR ball-handler (1st of 16) and PNR roll-man (4th of 16) more than the league field. The series- level question - "did Boston make Embiid defend PNR/DHO?" - has two parts:

• PNR Ball Handler: Yes, aggressively. 20.3% of Boston's playoff offense was PNR ball-handler. That is the highest share in the playoff field.
• PNR Roll Man: Yes more often than league average (4th of 16) at excellent efficiency (1.196 PPP). Boston's two roll-men (Queta 96th pctile, Vučević 32nd pctile) had very different impact. Queta got only 2.0 roll-man possessions per game - that is the number to push on. He shot 92.3% from the floor as a roll-man in the playoffs.
• DHO: No. Frequency collapsed from 5.3% to 3.4% (15th of 16) at worse efficiency.

The most defensible criticism of Boston's playoff offense is not "they used DHO over PNR" (false). It is "they ran PNR ball-handler heavily but didn't feed Queta the roll often enough given his playoff PPP." Two roll-man possessions per game for a 96th-percentile finisher is under-deployment.

11. What supports the RealGM claim

• Queta is a credible PNR roll-man weapon. 1.348 PPP (RS, 86th pctile), 1.786 PPP (playoffs, 96th pctile). The "pretty good at it" line dramatically understates the data.
• Boston could have made Embiid defend more PNR. Queta's playoff roll-man volume was only 2.0 possessions per game. There is room.
• The prescription "more PNR" is data-consistent. Boston's playoff PNR-handler frequency was 1st of 16 already, but the PPP fell to 0.856. Adding more PNR ROLL MAN (not just handler) would have exploited the Queta efficiency.

12. What does not support the claim

• "Boston uses DHO more than PNR" is false. PNR ball-handler is 18.3% (4th in league) of Boston's offense; Handoff is 5.3% (8th). In the playoffs the gap widened: 20.3% PNR vs 3.4% DHO.
• "Boston uses DHO and skip passes more than straight PNR" framing is the most common error in informed-sounding RealGM takes about this team. The actual offense is PNR-handler-heavy and DHO-light, not the inverse.
• Boston was the lowest-DHO playoff team in the field (15th of 16). Whatever DHO-emphasis impression this writer had does not survive the 2025-26 playoff data.

13. What remains unproven

• The play-type CSV does not tell us whether the right response by the defense is to switch, ICE, drop, or trap. That requires defensive Synergy fields (the defensive TypeGrouping was not pulled this pass).
• The Brown-without-Tatum split for PNR-handler frequency specifically is computable but not yet pulled.
• 2024-25 and 2023-24 playoff baselines for Boston not pulled; could be added in next pass for trend context.
• Vučević's frequency as a Boston roll-man in his short BOS tenure is in the data but conflated with his pre-trade reps; a team-and-date filter would isolate it.

14. Final RealGM paragraph

The "Boston uses DHO more than PNR" line is the wrong way to read this offense and the actual 2025-26 Synergy data refuses it. Boston ran PNR ball-handler on 18.3% of regular-season possessions (4th in the league) and 20.3% in the playoffs (1st in the playoff field). They ran DHO 5.3% RS and 3.4% in the playoffs - 15th of 16 playoff teams. PNR is roughly six times more common than DHO in their playoff diet, not the other way around. The Queta part of the claim is correct but understated: he was a 96th-percentile NBA roll man in the playoffs at 1.786 PPP on 92.3% from the floor, and he only got two roll-man possessions per game. That is the right complaint - not that Boston should have run more "PNR/DHO" in general, but that they should have fed Queta the roll more often when Embiid was on the floor. Queta's 2.0 possessions per game in that action against a defender he was a structural mismatch for is the actual sub-optimal usage. Boston's playoff PPP on PNR-handler was 0.856 (5th in the field); on PNR roll-man it was 1.196 (5th). The roll was the better action and they ran the handler more.

---

Sources

• stats.nba.com/stats/synergyplaytype via nba_api 1.10.2 - authoritative; 44 successful calls, all rowSets non-empty. Params: LeagueID=00, PerMode=PerGame, PlayerOrTeam=T|P, SeasonType=Regular Season|Playoffs, SeasonYear=2025-26, TypeGrouping=offensive, PlayType in {PRBallHandler, PRRollman, Handoff, Isolation, Spotup, Transition, Cut, OffScreen, Postup, OffRebound, Misc}.
• Basketball Reference team and league pages 2025-26 - pace, ORtg, DRtg.
• Audit trail: data/celtics_action_type_fetch_log.csv (51+ rows).
• Raw JSON payloads: data/raw/action_types/nba_api/ (44 files).

Limits, declared

• The Synergy "PERCENTILE" field is calculated by NBA against the qualifying-possessions pool and may include defensive-side comparisons internally. Treated as their official efficiency percentile.
• Defensive Synergy (TypeGrouping=defensive) not pulled.
• 2024-25 and 2023-24 playoff baselines not pulled.
• Vučević frequency on BOS specifically not isolated (conflated with full-season MP).
• Per-100-team-possessions is computed from POSS_PCT (% of team's own possessions) and is identical to that share scaled by 100. Boston's slower pace means raw per-game counts trail faster teams; rank / percentile fields are pace-neutral.