Market Foresight Premium

The closing reference forecast's skill advantage over a specific state-only model, scaled by remaining reference entropy at each checkpoint.

confirmednull (a finding)not testabledescriptivepending

How to read these measurements

At each checkpoint, compare closing reference and state-only model skill against the naive forecast, then divide their difference by reference entropy with a 0.15-bit floor. The observation window is not published.

The closing reference is a forecaster, not a live price. Checkpoints can contain different populations; inspect each observation count, especially late in the game. Endpoint differences alone do not establish a time trend or causal explanation. Model misspecification and information the reference prices remain confounded.

Published measurements

MLB10 published rows

MLB published timing measurements
Inning	MFP	Closing reference skill	State-only model skill	Reference entropy (bits)	Observations	Entropy floor applied
1	0.051	0.0333	-0.0154	0.9553	3,589	No
2	0.1794	0.0805	-0.0805	0.8971	2,803	No
3	0.1434	0.074	-0.0406	0.7993	3,097	No
4	0.181	0.0386	-0.0951	0.7386	3,034	No
5	0.1249	0.1172	0.035	0.6583	3,139	No
6	0.1404	0.1506	0.0648	0.6106	3,579	No
7	0.1571	0.2463	0.1612	0.5414	3,447	No
8	0.0831	0.3969	0.3574	0.4757	2,833	No
9	0.0159	0.5547	0.5474	0.4589	1,110	No
10	0.411	0.2858	-0.0631	0.8488	46	No

International soccer19 published rows

International soccer published timing measurements
Minute	MFP	Closing reference skill	State-only model skill	Reference entropy (bits)	Observations	Entropy floor applied
0	0.6779	0.4883	-0.0643	0.8152	150	No
5	0.7385	0.4673	-0.1496	0.8354	213	No
10	0.7797	0.4788	-0.1725	0.8352	203	No
15	0.8156	0.4403	-0.2155	0.8041	208	No
20	0.7158	0.3952	-0.1972	0.8275	192	No
25	0.7108	0.3094	-0.2715	0.8171	195	No
30	0.8009	0.2872	-0.344	0.7881	192	No
35	0.732	0.262	-0.2964	0.7629	173	No
40	0.8534	0.2329	-0.4017	0.7437	189	No
45	0.8149	0.308	-0.2975	0.743	402	No
50	0.9193	-0.0211	-0.6176	0.6488	189	No
55	0.7635	-0.0626	-0.5354	0.6193	192	No
60	0.796	-0.0523	-0.549	0.624	189	No
65	0.8462	-0.1651	-0.7257	0.6625	154	No
70	0.9234	-0.358	-0.9763	0.6696	152	No
75	1.0634	-0.1693	-0.9452	0.7297	132	No
80	1.3107	-0.2629	-1.1626	0.6865	127	No
85	1.7802	-0.5103	-1.5469	0.5823	98	No
90	1.4861	-0.2424	-0.8324	0.397	33	No

scripts/platformkit/analytics_showcase/out/novel_market_foresight_premium.jsonPublished date 2026-09-17Checkpoint observations shown per row; unique game counts unavailable.

Explore these measurements in the lab