MATRIX CAPTAIN DECISION ARCHIVE

CAPTAIN DECISION ARCHIVE

High-stakes pairing calls from major team events, recorded with their information state, counterfactual analysis, and outcome. The decisions are interesting; the captains are not on trial. Every entry is reviewed for fairness before publication.
14 ENTRIES

SPOTLIGHT DECISION

FEATURED · UPDATED MAY 11, 2026
DECISION #07 · CAPTAIN'S CALL
THE ANCHORED THROW
WESTERN REGIONAL TEAM CHAMPIONSHIP · 5-PLAYER FORMAT · ROUND 5 (FINAL) · NOV 2025 · WRITTEN-SCRIPT BLIND PROTOCOL
PRE-ROUND
4-0 vs 4-0
WINNER TAKES TOURNAMENT

DECISION SUMMARY

Going into the final round, captain A faced an opposing roster of Aeldari, Death Guard, Tyranids, T'au, and Custodes. Her own five lists — Necrons, GSC, Sisters, CSM, Imperial Knights — read favorably across most matchups, with one structural concern: Necrons into T'au at −2 (a calibrated ≈30% win). She had two structurally distinct calls available: anchor the strong matchups by deliberately putting GSC into T'au (her team's only +1 read there) to free Necrons to take the +2 Death Guard call, or spread the pain by accepting a flatter set of mostly-even pairings.

She chose the anchored optimal. Hungarian-optimal under her stated ratings. The team won the round 4-1.

MATCHUP MATRIX · AT DECISION TIME

Captain A's matchup reads on the WTC −2 to +2 scale (see legend below the matrix). = chosen pairing. = Hungarian-optimal under stated ratings (matches in this case).

Aeldari
Death Gd
Tyranids
T'au
Custodes
Necrons
−1
+2
0
−2
−1
GSC
0
0
−1
+1
−1
Sisters
−1
0
+1
0
0
CSM
+1
−1
−1
0
0
Knights
0
0
0
+1
+1
CAPTAIN'S CHOICE HUNGARIAN OPTIMAL RATING SUM: +6 / +10 MAX · SPREAD ALT: +4 · NEXT-BEST: +5
−2 VERY BAD Need variance to win ≈ 30% win
−1 BAD Underdog — uphill ≈ 40% win
 0 EVEN Anyone's game ≈ 50% win
+1 GOOD Favored — should win more often ≈ 60% win
+2 VERY GOOD Strong favorite — should rarely lose ≈ 70% win

► The WTC −2 to +2 scale is what captain A actually used. The percentage column is a calibrated translation for cross-comparison with aggregator data — illustrative until tied to empirical win-rate data.

COUNTERFACTUAL · WHAT ELSE COULD SHE HAVE CALLED

The "spread the pain" alternative was Necrons→Death Guard, GSC→Custodes, Sisters→Tyranids, CSM→Aeldari, Knights→T'au — a rating sum of +4 against the chosen +6. Calibrated, that's 2.80 expected wins versus 3.10. The trade is real: she would have surrendered roughly 0.30 of expected wins to put Knights into T'au at −1 instead of GSC into T'au at +1. In tournament play with rounds scored as 3-2 wins versus 4-1 or 5-0, that's a meaningful structural difference — fewer 5-0 sweeps, more close calls.

Sensitivity: the chosen call is robust. Any single matchup rating could shift by ±2 steps before the optimal flips — meaning even if she had badly mis-read one cell on the five-point scale, the call holds. The Necron-T'au matchup itself is particularly robust to her being wrong about it: it could be a 0 instead of a −2 and the optimal still doesn't change, because the gain from Necrons going into Death Guard at +2 dominates.

The interesting analytical question is whether captain A's ratings were calibrated. The chosen sum of +6 rests on Necrons-Death Guard at +2 (calibrated ≈70% win) and four +1 matchups (calibrated ≈60% each). Aggregator data from the prior month put Necrons-Death Guard closer to +1 (≈60% win) across 200+ recorded games. If +1 is closer to truth, calibrated expected wins drop from 3.10 to 3.00 — still optimal, still the right call, but the upside is smaller than her read suggested.

RESULT

TEAM A LISTTEAM B LISTRATING (CAL. %)RESULT
NecronsDeath Guard+2  (≈70%)WIN
GSCT'au+1  (≈60%)WIN
SistersTyranids+1  (≈60%)LOSS
CSMAeldari+1  (≈60%)WIN
Imperial KnightsCustodes+1  (≈60%)WIN
TEAM RESULT4-1 WIN

The Sisters-Tyranids loss was the variance outcome. At a +1 read, the team-A list was favored but not crushingly so — a 60-40 game can flip in 4 of 10 plays. Captain A took the optimal call and saw it through despite the lost game; the team's record line — 4-1 for the championship — does not communicate how thin the margin actually was. Two more variance hits in the same direction and the line would have read 2-3.

POST-MORTEM

This decision is in the archive because it does two unrelated things well at once. It demonstrates correct throw logic — the difference between a captain who throws because she's quitting on a matchup and one who throws because the resulting assignment dominates the alternative. Captain A is firmly in the second camp, and the math supports her clearly.

It also illustrates the calibration problem. Whether her decision was right is not the same question as whether her ratings were right. The team's 4-1 result is consistent with a much wider range of underlying matchup truths than her stated +6. If her Necron-DG read had been +1 instead of +2, she still calls the same assignment, still likely wins the round, and would have learned nothing about her own calibration. Outcome-driven feedback in pairing decisions is famously thin — most captains never get the volume of repeated decisions needed to know if they're reading matchups well on the 5-point scale, let alone calibrating to win-rate ground truth.

One thing the archive cannot answer: whether captain B's reciprocal pairing was avoidable. Under blind-script protocols both captains commit simultaneously, so captain B was making her own optimization independently. If captain B had a different read of the matchups — perhaps undervaluing the Sisters-Tyranids matchup as a giveaway and consequently positioning Tyranids elsewhere — the assignment captain A faced would have been different and her optimal would have flipped. Without captain B's matrix it's impossible to say what her read was.

TOTAL RATING
+6 / +10
Calibrated 3.10 / 5 expected wins (≈62.0% implied team win rate).
ROBUSTNESS
±2 STEPS
Any single matchup re-rated by up to ±2 leaves the optimal assignment unchanged. Classified ROBUST.
VS SPREAD ALT
+2 RATING
The "spread the pain" alternative scores +4 (≈2.80 wins). The captain's call gained +2 rating / +0.30 calibrated wins by accepting GSC into T'au instead of Knights.
"Someone on my team was going to take T'au. GSC had the only +1 read on it, so GSC took it. That freed Necrons to go into Death Guard, which is where the +2 lived. The rest of the team got their preferred matchups out of the bargain."
— CAPTAIN A, POST-EVENT INTERVIEW
TAGS
throw matchup anchor pairing blind protocol robust optimal variance survived calibration question
SOURCES
Event coverage streamVOD · 2:14:30
Captain A interviewPODCAST · WK AFTER
Goonhammer recapEXTERNAL
TRY IT YOURSELF
Load this scenario into the Pairings Tool and stress-test the matchup probabilities.
OPEN IN PAIRINGS TOOL ›

ALL DECISIONS

14 ENTRIES · NEWEST FIRST
DECISION #07★★★★★
The Anchored Throw
WESTERN REGIONAL · ROUND 5 (FINAL) · NOV 2025
Captain anchors GSC into the T'au matchup at +1 to free Necrons for the +2 Death Guard call. Hungarian-optimal; robust through ±2 rating steps.
throw matchup anchor robust
WON 4-1 · OPTIMAL CALL
DECISION #06★★★★
One Cell Off
ATLANTIC OPEN TEAM · ROUND 4 · OCT 2025
A knife-edge call where two assignments tied on rating sum and differed by one cell. Captain picked the lower-variance assignment; lost on the high-EV game.
knife-edge variance hit brittle
LOST 1-3 · DEFENSIBLE CALL
DECISION #05★★★★★
The Sequential Trap
NORTH-EAST INVITATIONAL · ROUND 5 · SEP 2025
Under ABBA sequencing, captain B placed first and committed his weakest list into the open. Captain A read the trap and ignored bait pairings.
information asymmetry sequential captain read
WON 5-0 · MASTERCLASS
DECISION #04★★★
Variance Over Optimal
PACIFIC TEAM CHAMPIONSHIP · FINAL · AUG 2025
Captain accepted lower expected sum (+4 vs +6) for a higher 5-0 sweep probability. Calculated a tournament-tiebreak need for the sweep.
tiebreaker pressure EV trade contextual optimal
WON 4-0 · UNCONVENTIONAL CALL
DECISION #03★★★★
Reading the Captain
CENTRAL TEAM OPEN · ROUND 3 · JUL 2025
Under open-declaration protocol, captain A leveraged knowledge of captain B's publicly stated faction preferences to predict the opening pairing.
open declaration meta-read behavioral
WON 3-1 · NON-MATH FACTOR
DECISION #02★★★★
Information Asymmetry
MID-ATLANTIC CUP · ROUND 6 · JUN 2025
A hybrid protocol where two pairings were revealed early. Captain B made an optimal call under partial information that captain A could not have known.
hybrid protocol partial information structural advantage
LOST 1-3 · PROTOCOL DRIVEN

LESSONS · CROSS-ENTRY PATTERNS

UPDATED ON EACH NEW ENTRY
PATTERN A14 / 14
Throw-matchup decisions trend positive
Across 5 archived throw-matchup decisions, captains who threw to anchor won the team round in 4 of 5 (80%). Captains who refused to throw and accepted flatter assignments won 51% across 9 entries.
SMALL N · DIRECTIONAL ONLY
PATTERN B7 / 14
Brittle optimal often loses
Decisions with sensitivity of ±1 step — the optimal flips on a single rating mis-read — won 43%. Robust decisions (±2 steps or wider) won 71%.
CONSISTENT WITH CALIBRATION THEORY
PATTERN C5 / 14
Last placement wins in sequential
In ABBA-style protocols, the captain placing the last commit won their round 5 of 5 times. Small sample, but consistent with the theoretical information-state advantage.
DIRECTIONAL · WATCHING FOR REVERSAL