outcome matters: motivation asymmetry + stage behavior

← lab · n with closing odds 5,149 · n late season matches 10,808 · ran 7/5/2026

What this is: A one-off study — its numbers below are exactly what the run reported.

note	market_tests excess = realized home rate minus de-vigged closing market home rate in that situation; \|z\| > 2 is interesting, > 3 is real. URGENT = drop_fight/title_race, DEAD = dead_rubber/mid.
n with closing odds	5,149
n late season matches	10,808

market tests

n	z	excess	situation	draw expected	draw realized	realized home rate
1,071	0.1	0.0015	urgent_home_vs_dead_away	0.2596	0.2605	0.3987
1,037	-0.69	-0.0108	dead_home_vs_urgent_away	0.2478	0.2459	0.4581
1,008	0.04	0.0007	both_urgent	0.2581	0.2639	0.4246
1,294	2.44	0.0335	both_dead	0.2655	0.2233	0.4614
1,785	0.5	0.0056	drop_fight_side_home	0.2724	0.2711	0.3501
463	0.22	0.0049	title_side_home	0.2103	0.216	0.6328

n	stage	avg goals	under 2 5	draw in 90
52,009	league	2.695	0.5067	0.2455
347	group	2.896	0.464	0.2277
79	knockout	3.076	0.443	0.2405

n	round	avg goals	under 2 5	draw in 90
12	quarter_finals	2.25	0.6667	0.3333
16	last_32	2.375	0.5625	0.3125
27	last_16	3.37	0.3704	0.1481
16	playoffs	3.875	0.25	0.1875

n	draw	away win	home win	over 2 5	avg goals	away need
2,520	0.2556	0.2833	0.4611	0.4853	2.6	dead_rubber
1,367	0.218	0.1953	0.5867	0.5267	2.813	drop_fight
1,244	0.2757	0.3939	0.3304	0.4574	2.544	dead_rubber
1,220	0.2926	0.2811	0.4262	0.4525	2.503	drop_fight
456	0.1842	0.2215	0.5943	0.4715	2.559	mid
388	0.2345	0.3995	0.366	0.4897	2.549	dead_rubber
369	0.2005	0.1816	0.6179	0.5718	2.919	dead_rubber
355	0.2817	0.4761	0.2423	0.5042	2.741	title_race
347	0.2478	0.2911	0.4611	0.5072	2.602	drop_fight
331	0.2356	0.3353

AUC	predictability: 0.50 = coin flip, ~0.70 = ceiling for sports
Importance	how much the model leans on this factor (permutation importance)
Direction	sign of the raw correlation with the outcome
Survives all eras	effect points the same way in every historical era

Spec · the reproducible recipe

{
  "kind": "motivation",
  "odds_comps": [
    "EPL",
    "ELC",
    "EL1",
    "EL2",
    "SPL"
  ],
  "late_season": 0.66,
  "fight_margin": 6
}