T-635 tables (node tools/diagnose.js --pooled)

From reports/tiers/2026-10-04-variables-pooled-diagnosis-tables.md , rendered when the site is built.

16 armies, 618 graded units, 341 variables; units held out 5 × 5 folds; 2 scrambles.

The armies:

Army	Lists	Units	Graded	Ceiling	Ledger in sample
Tyranids	auspex, hivemind, second, maelstrom, astrategas	52	52	88.5%	73.3%
Space Marines	auspexFull, tobias	84	84	91.0%	60.4%
Adeptus Mechanicus	belisariusPaul	34	32	–	68.0%
Agents of the Imperium	varik	29	26	–	44.8%
Astra Militarum	auspex	72	61	–	66.0%
Blood Angels	grimdarkChronicle	20	19	–	54.0%
Dark Angels	tacticalrock	93	93	–	54.8%
Death Guard	disgustinglyResilient	36	36	–	47.1%
Drukhari	deployOnTheLine	37	24	–	46.1%
Emperor's Children	joys	23	23	–	56.5%
Genestealer Cults	rey, vonCarmian	90	24	84.9%	68.8%
Leagues of Votann	list2025, discussion	22	20	59.3%	61.0%
Orks	spruesandbrews	55	45	–	70.0%
Space Wolves	artOfWar	21	21	–	46.9%
Thousand Sons	redundantRolling	35	28	–	45.6%
World Eaters	redPath, tacticalSugar, exalted	30	30	89.3%	61.5%

The headline (pairwise against each army's consensus; pooled: every within-army pair once; army mean: the 16 armies equal):

Model	Leave one army out: pooled	army mean	Tyranids	Space Marines	vs lists (pooled)	Units held out: pooled (± SE)	Tyranids	Scrambled, left out: pooled (SD)
(a) the ledger as it is	60.0%	57.8%	73.3%	60.4%	62.3%	60.3% ± 0.2	73.2%	49.1% (0.2)
(a1) the ledger re-fitted	59.9%	57.8%	73.2%	60.4%	62.2%	60.1% ± 0.2	72.9%	49.4% (0.2)
(b) linear, every variable	62.3%	61.7%	66.0%	65.6%	61.9%	64.7% ± 0.5	69.4%	47.8% (1.7)
(b2) linear, core 25	58.6%	57.7%	69.5%	58.6%	59.4%	60.9% ± 0.4	68.7%	48.1% (0.8)
(c) interactions (kernel)	66.3%	62.7%	64.8%	69.4%	64.5%	68.0% ± 0.8	71.3%	49.3% (0.6)
(d) boosted trees	64.8%	61.8%	64.8%	66.5%	63.9%	67.1% ± 0.8	67.1%	48.8% (0.4)
(e) neighbours	65.4%	58.6%	57.7%	71.2%	62.9%	68.4% ± 1.0	66.1%	50.0% (0.7)

Ceiling (each list against the mean of the army's other lists; 5 armies with 2+ lists): Tyranids 88.5%, Space Marines 91.0%, Genestealer Cults 84.9%, Leagues of Votann 59.3%, World Eaters 89.3%.

Leave one army out, per army (pairwise against the consensus):

Army	Graded	Ceiling	a	a1	b	b2	c	d	e
Tyranids	52	88.5%	73.3%	73.2%	66.0%	69.5%	64.8%	64.8%	57.7%
Space Marines	84	91.0%	60.4%	60.4%	65.6%	58.6%	69.4%	66.5%	71.2%
Adeptus Mechanicus	32	–	68.0%	68.0%	52.6%	50.6%	53.5%	58.2%	42.8%
Agents of the Imperium	26	–	44.8%	44.8%	56.5%	47.5%	57.0%	55.6%	52.0%
Astra Militarum	61	–	66.0%	66.0%	56.9%	64.6%	63.4%	67.2%	62.4%
Blood Angels	19	–	54.0%	54.0%	63.7%	53.2%	61.3%	62.1%	46.0%
Dark Angels	93	–	54.8%	54.8%	61.2%	54.0%	70.0%	66.0%	72.7%
Death Guard	36	–	47.1%	47.1%	52.5%	51.3%	52.9%	55.5%	56.9%
Drukhari	24	–	46.1%	46.1%	65.2%	55.2%	63.5%	60.0%	68.3%
Emperor's Children	23	–	56.5%	56.5%	64.9%	56.5%	64.4%	57.6%	61.3%
Genestealer Cults	24	84.9%	68.8%	68.8%	57.1%	59.2%	59.6%	66.3%	49.2%
Leagues of Votann	20	59.3%	61.0%	61.0%	59.1%	57.1%	57.1%	64.9%	54.5%
Orks	45	–	70.0%	70.0%	65.3%	70.5%	68.3%	69.6%	62.0%
Space Wolves	21	–	46.9%	46.9%	61.9%	66.9%	58.1%	46.9%	51.2%
Thousand Sons	28	–	45.6%	45.6%	77.0%	53.1%	77.0%	69.6%	67.6%
World Eaters	30	89.3%	61.5%	61.5%	61.5%	54.8%	63.0%	58.3%	61.0%

The Space Marine family left out together (each of the four scored with all four out of the training; the main run's figure in brackets):

Army	a	a1	b	b2	c	d	e
Space Marines	60.4% (60.4%)	60.4% (60.4%)	57.4% (65.6%)	51.9% (58.6%)	57.3% (69.4%)	59.9% (66.5%)	58.9% (71.2%)
Blood Angels	54.0% (54.0%)	54.0% (54.0%)	50.0% (63.7%)	58.1% (53.2%)	54.0% (61.3%)	53.2% (62.1%)	58.9% (46.0%)
Dark Angels	54.8% (54.8%)	54.8% (54.8%)	50.4% (61.2%)	47.1% (54.0%)	50.9% (70.0%)	48.8% (66.0%)	57.0% (72.7%)
Space Wolves	46.9% (46.9%)	46.9% (46.9%)	65.6% (61.9%)	68.8% (66.9%)	56.9% (58.1%)	48.1% (46.9%)	49.4% (51.2%)
pooled over the four	57.1%	57.1%	53.8%	49.9%	53.9%	53.8%	57.7%

Units held out within armies, per army:

Army	a	a1	b	b2	c	d	e
Tyranids	73.2%	72.9%	69.4%	68.7%	71.3%	67.1%	66.1%
Space Marines	61.1%	60.9%	64.0%	58.3%	68.3%	67.6%	71.9%
Adeptus Mechanicus	70.1%	69.8%	60.2%	56.3%	60.2%	64.4%	55.3%
Agents of the Imperium	43.3%	43.4%	54.4%	51.9%	54.0%	45.2%	42.4%
Astra Militarum	66.5%	66.4%	71.6%	74.8%	72.9%	78.8%	78.2%
Blood Angels	50.3%	50.3%	67.5%	68.6%	68.5%	68.7%	55.8%
Dark Angels	54.9%	54.8%	63.1%	56.7%	68.4%	67.6%	69.8%
Death Guard	46.8%	46.6%	52.8%	52.2%	53.1%	56.9%	61.8%
Drukhari	49.1%	49.1%	68.2%	49.7%	73.8%	63.2%	79.5%
Emperor's Children	53.1%	52.3%	62.7%	54.4%	64.4%	59.1%	63.4%
Genestealer Cults	69.7%	69.7%	65.3%	66.2%	61.5%	60.2%	55.9%
Leagues of Votann	65.4%	65.4%	58.2%	56.6%	59.8%	66.1%	58.0%
Orks	67.9%	67.7%	69.8%	69.9%	71.0%	62.7%	59.2%
Space Wolves	49.8%	49.8%	60.9%	67.8%	60.0%	44.8%	43.9%
Thousand Sons	46.5%	46.5%	71.3%	53.8%	73.2%	66.5%	74.5%
World Eaters	61.9%	62.2%	59.6%	60.3%	63.9%	68.0%	62.9%

Paired, leave one army out (per-army gain in points; SE over the 16 armies; armies better; pooled gain; scrambled spread of the army-mean gain against (a1)):

Pair	Mean gain ± SE	Armies better	Pooled gain	Scrambled SD
b − a1	+3.9 ± 3.1	9 of 16	+2.3	1.5
b2 − a1	−0.1 ± 2.1	6 of 16	−1.4	2.6
c − a1	+4.9 ± 2.9	10 of 16	+6.4	0.7
d − a1	+4.0 ± 2.2	10 of 16	+4.8	1.7
e − a1	+0.8 ± 3.6	8 of 16	+5.4	1.8
a1 − a	−0.0 ± 0.0	0 of 16	−0.0	–
b − a	+3.9 ± 3.1	9 of 16	+2.3	–
b2 − a	−0.1 ± 2.1	6 of 16	−1.4	–
c − a	+4.9 ± 2.9	10 of 16	+6.4	–
d − a	+4.0 ± 2.2	10 of 16	+4.8	–
e − a	+0.8 ± 3.6	8 of 16	+5.4	–

Settings chosen, leave one army out (most often):

a: {} ×16
a1: {"lam":10} ×13, {"lam":0.03} ×1, {"lam":0.01} ×1, {"lam":0.1} ×1
b: {"kind":"ridge","lam":0.3} ×5, {"kind":"ridge","lam":0.1} ×4, {"kind":"ridge","lam":0.03} ×3, {"kind":"ridge","lam":1} ×2, {"kind":"lasso","lam":0.002} ×1
b2: {"kind":"ridge","lam":3} ×8, {"kind":"ridge","lam":0.003} ×4, {"kind":"ridge","lam":1} ×2, {"kind":"ridge","lam":0.3} ×2
c: {"lam":0.0003} ×9, {"lam":0.001} ×5, {"lam":0.003} ×2
d: {"depth":2,"colsample":1,"trees":400} ×4, {"depth":3,"colsample":1,"trees":400} ×3, {"depth":2,"colsample":1,"trees":25} ×2, {"depth":3,"colsample":1,"trees":200} ×2, {"depth":3,"colsample":1,"trees":25} ×2
e: {"k":3} ×10, {"k":5} ×2, {"k":20} ×2, {"k":30} ×1, {"k":8} ×1

Does pooling help Tyranids? Held-out pairwise on Tyranid units against their consensus:

Model	Tyranids alone (their own folds)	From the other 15 armies only	The other 15 plus their own folds
(a) the ledger as it is	73.2%	73.3%	73.2%
(a1) the ledger re-fitted	71.5%	73.2%	72.9%
(b) linear, every variable	70.7%	66.0%	69.4%
(b2) linear, core 25	66.5%	69.5%	68.7%
(c) interactions (kernel)	72.1%	64.8%	71.3%
(d) boosted trees	67.0%	64.8%	67.1%
(e) neighbours	68.0%	57.7%	66.1%

The learning curve (trained on m random armies at the model's most chosen setting, scored on the army left out; pooled / army mean):

Model	3	6	9	12	15
c	59.4% / 58.7%	61.6% / 60.4%	59.9% / 61.9%	66.8% / 63.3%	66.3% / 62.7%
a1	59.9% / 57.8%	59.9% / 57.8%	59.9% / 57.8%	59.9% / 57.8%	59.9% / 57.8%
b2	52.4% / 53.3%	59.7% / 58.3%	56.6% / 57.3%	58.0% / 57.1%	58.6% / 57.7%
d	58.0% / 58.2%	61.5% / 60.8%	56.7% / 61.0%	67.0% / 62.3%	64.8% / 61.8%
a	60.0% / 57.8%	60.0% / 57.8%	60.0% / 57.8%	60.0% / 57.8%	60.0% / 57.8%

Permutation importance, model c (left-out army, its fit on the other 15; drop in pooled pairwise, points; armies where it drops by more than 0.5):

Family	Drop	Armies
all T-636 additions	8.71	13
flags	3.62	12
keywords	2.25	9
datasheet	2.04	11
size	1.72	10
rules	1.60	9
loadout	1.00	7
weapons	0.95	7
killclass	0.90	6
ledger	0.53	5
weaponkw	0.50	7
interaction	0.48	6

T-632 candidate	Drop	Armies
1. Threat on arrival	0.96	8
2. Scarce answers (anti-monster and anti-vehicle reach)	1.30	6
3. Objective bodies (Score, OC per point)	0.43	6
4. Character sniping (Precision)	-0.04	1
5. Transports	0.89	9
6. Psychic attacks and psykers	0.12	4
7. Cheap reach: actions and indirect fire	0.28	5
8. Value at the cheapest size	0.51	9
9. Melee Soak	-0.10	4

Variable	Family	Drop	Armies
rangedVeh100	weapons	0.55	8
kwu_Captain	keywords	0.52	3
ptsAtMax	size	0.43	7
ptsAtMin	size	0.39	6
ptsCheap	datasheet	0.36	5
kwu_Centurion	keywords	0.35	2
carriedTotal	loadout	0.34	7
k6_heavy_infantry	killclass	0.34	5
lands	ledger	0.33	4
OC_max	size	0.33	6
maxA_any	loadout	0.32	4
lr_lone	flags	0.30	5
anti_character	weaponkw	0.30	2
kwu_Support	keywords	0.29	2
pts	datasheet	0.28	8
bestM_AP	weapons	0.27	4
ix_WxSave	interaction	0.26	3
has_loneOperative	flags	0.26	5
OC	datasheet	0.25	5
fx_aura	flags	0.25	3

The T-636 additions alone (top 15):

Variable	Family	Drop	Armies
kwu_Captain	keywords	0.52	3
ptsAtMax	size	0.43	7
ptsAtMin	size	0.39	6
kwu_Centurion	keywords	0.35	2
carriedTotal	loadout	0.34	7
OC_max	size	0.33	6
maxA_any	loadout	0.32	4
anti_character	weaponkw	0.30	2
kwu_Support	keywords	0.29	2
firingDeck_n	rules	0.24	4
formShare	size	0.21	4
has_firingDeck	rules	0.20	4
ppmAtMin	size	0.20	2
kwu_Terminator	keywords	0.20	2
ppmAtMax	size	0.18	1

Permutation importance, model d (left-out army, its fit on the other 15; drop in pooled pairwise, points; armies where it drops by more than 0.5):

Family	Drop	Armies
all T-636 additions	7.33	13
ledger	5.11	11
size	3.62	14
loadout	2.56	12
flags	1.23	8
weapons	0.78	6
killclass	0.66	5
rules	0.34	8
keywords	-0.08	6
interaction	-0.27	2
datasheet	-0.31	2
weaponkw	-0.39	3

T-632 candidate	Drop	Armies
1. Threat on arrival	-0.10	3
2. Scarce answers (anti-monster and anti-vehicle reach)	1.26	4
3. Objective bodies (Score, OC per point)	0.23	6
4. Character sniping (Precision)	-0.01	2
5. Transports	0.07	6
6. Psychic attacks and psykers	0.10	3
7. Cheap reach: actions and indirect fire	0.21	4
8. Value at the cheapest size	-0.03	1
9. Melee Soak	-0.06	0

Variable	Family	Drop	Armies
ptsAtMax	size	1.60	11
c_kill	ledger	1.28	12
line_killMelee	ledger	0.67	9
rangedVeh100	weapons	0.61	6
lfx_effects	flags	0.58	6
ptsAtMin	size	0.57	10
dsRanged	loadout	0.57	10
shortRanged	loadout	0.50	4
has_loneOperative	flags	0.48	5
sizeRatio	size	0.47	5
dsProfiles	loadout	0.40	2
dsWeapons	loadout	0.38	4
maxAP_any	loadout	0.36	5
carriedTotal	loadout	0.30	9
lands	ledger	0.30	4
formShare	size	0.29	6
k6_heavy_infantry	killclass	0.28	3
ppmAtMin	size	0.26	3
meleeChaff100	weapons	0.26	4
kwu_Terminator	keywords	0.25	1

The T-636 additions alone (top 15):

Variable	Family	Drop	Armies
ptsAtMax	size	1.60	11
ptsAtMin	size	0.57	10
dsRanged	loadout	0.57	10
shortRanged	loadout	0.50	4
sizeRatio	size	0.47	5
dsProfiles	loadout	0.40	2
dsWeapons	loadout	0.38	4
maxAP_any	loadout	0.36	5
carriedTotal	loadout	0.30	9
formShare	size	0.29	6
ppmAtMin	size	0.26	3
kwu_Terminator	keywords	0.25	1
carriedDistinct	loadout	0.21	5
kwu_Tacticus	keywords	0.20	2
abilitiesOwn	rules	0.15	4

Each variable alone, within army (share of within-army consensus pairs it orders the same way, ties half; armies with Spearman above +0.1 / below −0.1):

Variable	Family	Concordance	Armies +	Armies −
c_kill	ledger	60.1%	11	0
value	ledger	60.0%	10	2
valueCheap	ledger	59.6%	10	1
k6_heavy_infantry	killclass	58.8%	10	0
k6_cavalry_and_beasts	killclass	58.6%	11	1
carriedTotal	loadout	58.5%	8	4
ix_killxSoak	interaction	58.1%	8	4
ptsAtMax	size	57.7%	9	1
c_hold	ledger	57.1%	8	3
lands	ledger	56.7%	8	4
ix_bodiesxOC	interaction	56.7%	9	5
pts	datasheet	56.5%	10	2
M	datasheet	56.4%	7	4
k6_light_vehicle	killclass	56.3%	8	2
rangedMeq100	weapons	56.1%	7	7
OC_max	size	56.0%	8	4
OC	datasheet	56.0%	8	4
carriedDistinct	loadout	56.0%	7	6
ptsCheap	datasheet	55.7%	10	1
ptsAtMin	size	55.7%	10	1
rangedVeh100	weapons	55.5%	7	6
rangedChaff100	weapons	55.3%	5	7
line_killRanged	ledger	54.9%	7	7
defaultWeapons	loadout	54.8%	7	4
k6_character	killclass	54.6%	5	3
ledByCount	rules	54.6%	6	3
k6_elite_infantry	killclass	54.5%	6	2
meleeChaff100	weapons	54.5%	7	1
rangedAttacks100	weapons	54.4%	5	8
c_score	ledger	54.3%	8	4

The T-636 additions alone, within army (top 20 by distance from 50%):

Variable	Family	Concordance	Armies +	Armies −
carriedTotal	loadout	58.5%	8	4
ptsAtMax	size	57.7%	9	1
OC_max	size	56.0%	8	4
carriedDistinct	loadout	56.0%	7	6
ptsAtMin	size	55.7%	10	1
defaultWeapons	loadout	54.8%	7	4
ledByCount	rules	54.6%	6	3
sizeMin	size	54.2%	6	3
carriedRanged	loadout	54.1%	6	8
sizeMax	size	54.0%	6	3
maxAP_any	loadout	54.0%	6	4
sizeRatio	size	53.9%	9	1
sizeSteps	size	53.7%	9	2
bestWS	loadout	53.6%	10	2
carriedMelee	loadout	53.4%	9	3
minRange	loadout	46.7%	2	9
modelProfiles	size	53.3%	6	4
dsMultiProfile	loadout	53.2%	6	1
kwu_Grenades	keywords	53.0%	8	3
dsProfiles	loadout	53.0%	7	6

T-632's candidates, their variables alone:

1. Threat on arrival: arrives 54.1% (+9/−2); has_deepStrike 52.6% (+7/−3); has_scouts 50.3% (+3/−2); has_infiltrators 51.5% (+6/−2); fx_arrival 50.9% (+6/−2); kw_is_burrowers 50.4% (+1/−0); kw_is_vanguard 50.4% (+1/−0); ix_arrivexMelee 54.3% (+9/−2); scouts_inches 50.3% (+3/−1)
2. Scarce answers (anti-monster and anti-vehicle reach): k6_monster 54.3% (+6/−3); k6_heavy_vehicle 52.3% (+7/−4); k6_light_vehicle 56.3% (+8/−2); rangedVeh100 55.5% (+7/−6); meleeVeh100 52.3% (+8/−2); kw_heavy 51.8% (+5/−3); kw_melta 51.9% (+6/−2); kw_anti 48.5% (+3/−7); scarcity 53.1% (+8/−1); anti_vehicle 49.0% (+4/−3); anti_monster 50.3% (+4/−0); wk_lance 50.1% (+4/−2); wk_meltaN 51.9% (+6/−2)
3. Objective bodies (Score, OC per point): c_score 54.3% (+8/−4); ocPer100 53.7% (+8/−5); OC 56.0% (+8/−4); ix_bodiesxOC 56.7% (+9/−5); kw_is_battleline 51.0% (+8/−4)
4. Character sniping (Precision): kw_precision 49.7% (+7/−4); k6_character 54.6% (+5/−3)
5. Transports: kw_is_transport 48.8% (+1/−6); kwu_Transport 48.8% (+1/−6); kwu_Dedicated_Transport 48.8% (+3/−6); transportCapacity 48.9% (+1/−6); capacityPer100 48.6% (+1/−5); has_firingDeck 50.0% (+4/−4); firingDeck_n 50.0% (+4/−4)
6. Psychic attacks and psykers: kw_psychic 51.8% (+6/−3); kw_is_psyker 52.1% (+7/−3)
7. Cheap reach: actions and indirect fire: kw_indirect 49.2% (+2/−3); c_actions 47.2% (+4/−7); ptsCheap 55.7% (+10/−1)
8. Value at the cheapest size: valueCheap 59.6% (+10/−1); ptsCheap 55.7% (+10/−1); modelsCheap 54.2% (+6/−3)
9. Melee Soak: soakFight 49.0% (+2/−8)

The trees on all 16 armies (depth 3, 100 trees; share of split gain; median threshold in SDs within the army):

c_kill: 7.4% of the gain, threshold ~-0.02 SD, 7 splits
ptsAtMax: 6.0% of the gain, threshold ~-0.78 SD, 15 splits
line_killMelee: 4.9% of the gain, threshold ~-1.44 SD, 6 splits
OC_max: 4.1% of the gain, threshold ~-1.17 SD, 3 splits
c_score: 3.8% of the gain, threshold ~-1.20 SD, 4 splits
choiceEntries: 3.7% of the gain, threshold ~-0.19 SD, 8 splits
ptsAtMin: 3.0% of the gain, threshold ~-1.05 SD, 11 splits
dsRanged: 2.8% of the gain, threshold ~1.19 SD, 12 splits
carriedTotal: 2.4% of the gain, threshold ~-0.64 SD, 8 splits
kw_is_fly: 2.0% of the gain, threshold ~2.30 SD, 11 splits
kw_devastating: 1.9% of the gain, threshold ~1.20 SD, 12 splits
has_loneOperative: 1.7% of the gain, threshold ~3.43 SD, 9 splits
maxAP_any: 1.7% of the gain, threshold ~1.79 SD, 10 splits
dsWeapons: 1.6% of the gain, threshold ~1.42 SD, 12 splits
c_actions: 1.6% of the gain, threshold ~-1.85 SD, 11 splits

Interactions (one split under the other, share of gain):

c_kill × ptsAtMax: 2.7%
OC_max × c_kill: 2.3%
carriedTotal × line_killMelee: 2.1%
c_score × ptsAtMax: 1.9%
choiceEntries × ptsAtMin: 1.6%
choiceEntries × ptsAtMax: 1.6%
carriedTotal × dsRanged: 1.4%
kw_is_fly × ptsAtMin: 1.4%
line_killMelee × ptsAtMax: 1.3%
c_kill × line_killMelee: 1.0%
dsRanged × maxAP_any: 0.9%
kw_devastating × kw_is_psyker: 0.8%

Partial dependence (letters, consensus centred within army; biggest step in SDs within the army):

c_kill: range 0.33; biggest step +0.31 between -0.05 and 0.08
ptsAtMax: range 0.38; biggest step +0.18 between -0.80 and -0.45
line_killMelee: range 0.48; biggest step +0.48 between -3.72 and -1.11
OC_max: range 0.29; biggest step +0.29 between -2.02 and -0.73
c_score: range 0.39; biggest step +0.39 between -3.38 and -0.72
choiceEntries: range 0.28; biggest step -0.17 between -1.30 and -0.93
ptsAtMin: range 0.31; biggest step +0.19 between -1.15 and -0.75
dsRanged: range 0.53; biggest step -0.11 between 1.82 and 2.89
carriedTotal: range 0.24; biggest step +0.20 between -0.94 and -0.44
kw_is_fly: range 0.31; biggest step -0.30 between 2.18 and 2.46

Missed by every flexible model the same way, leaving the army out (mean predicted letter minus consensus; the ledger's beside it):

Rated higher by the reviewers than any model gives:

Unit	Army	Consensus (letters)	Models	Ledger
Miasmic Malignifier	Death Guard	5 (S)	-4	-4
Suppressor Squad	Dark Angels	5 (S)	-3.8	-3
Stormraven Gunship	Dark Angels	4 (A)	-3.8	-4
Venom	Drukhari	5 (S)	-3.4	-5
Biovores	Tyranids	5 (S/S/S/S/S)	-3.227	-0.6
Librarian in Phobos Armour	Dark Angels	5 (S)	-3.2	-4
Skitarii Marshal	Adeptus Mechanicus	4 (A)	-3	-4
Chaos Rhino	Thousand Sons	5 (S)	-3	-4
Navigator	Agents of the Imperium	5 (S)	-2.8	-3
Lieutenant in Phobos Armour	Dark Angels	3 (B)	-2.8	-2

Rated lower by the reviewers than every model gives:

Unit	Army	Consensus (letters)	Models	Ledger
Pteraxii Skystalkers	Adeptus Mechanicus	1 (D)	+3.4	0
Company Heroes	Dark Angels	0 (F)	+3.4	+3
Tzaangor Enlightened	Thousand Sons	1 (D)	+3.4	+4
Heavy Intercessor Squad	Dark Angels	1 (D)	+3	+3
Serberys Sulphurhounds	Adeptus Mechanicus	0 (F)	+2.8	+4
Iron Priest	Space Wolves	0 (F)	+2.8	+5
Maulerfiend	Thousand Sons	2 (C)	+2.8	+3
Primaris Psyker	Astra Militarum	2 (C)	+2.6	+1
Hierophant	Tyranids	1.333 (D/C/–/D/–)	+2.583	+0.667
Fulgurite Electro-Priests	Adeptus Mechanicus	1 (D)	+2.4	+1

Missed the same way, per army (higher / lower / graded): Tyranids 4/5/52, Space Marines 18/14/84, Adeptus Mechanicus 7/7/32, Agents of the Imperium 4/3/26, Astra Militarum 1/7/61, Blood Angels 4/2/19, Dark Angels 15/13/93, Death Guard 5/7/36, Drukhari 2/5/24, Emperor's Children 4/3/23, Genestealer Cults 5/7/24, Leagues of Votann 2/2/20, Orks 7/4/45, Space Wolves 2/3/21, Thousand Sons 3/5/28, World Eaters 8/8/30

Traits of the 30 most under-rated (mean SD within army): ix_killxSoak -0.69, ix_durHeavy100 -0.68, c_hold -0.65, c_kill -0.64, ptsAtMax -0.62, value -0.58, valueCheap -0.56, pts -0.56, OC_max -0.56, OC -0.55

Traits of the 30 most over-rated: kw_psychic 0.60, kw_is_psyker 0.56, k6_heavy_infantry 0.47, bestM_D 0.43, rulesCount -0.38, bestR_A 0.35, rangedChaff100 0.34, value 0.32, c_kill 0.32, meleeVeh100 0.31

Run: 1892 s.