Leaderboard

Best performing user generated models

#	Competition	Market	User	Algorithm	ROI	RPS	Log Loss	Predictions (possible)
1	Liga	1X2	Quant Wizard	TabICL	7.28%	0.192	0.940	268 (268)
2	MLS	1X2	Quant Wizard	XGBoost	6.04%	0.231	1.098	483 (991)
3	Liga	1X2	Quant Wizard	XGBoost	5.43%	0.198	0.988	1311 (1311)
4	Liga	1X2	Quant Wizard	CatBoost	5.24%	0.194	0.962	782 (782)
5	Liga	1X2	John W	XGBoost	5.09%	0.200	0.993	957 (957)
6	Liga	1X2	Quant Wizard	CatBoost	4.99%	0.192	0.954	784 (784)
7	Brasileirão Serie A	1X2	Quant Wizard	TabPFN	4.97%	0.199	0.986	322 (322)
8	Ligue 1	1X2	John W	CatBoost	4.74%	0.216	1.007	405 (751)
9	Liga	1X2	Quant Wizard	TabPFN	4.47%	0.195	0.954	300 (300)
10	Liga	1X2	Quant Wizard	CatBoost	4.46%	0.194	0.961	782 (782)
11	Serie A	1X2		Logistic regression	4.28%	0.193	0.981	824 (824)
12	MLS	1X2	Quant Wizard	XGBoost	4.14%	0.232	1.098	484 (991)
13	Ligue 1	1X2	singinanarchist	Decision Trees	4.13%	0.216	1.037	778 (778)
14	Liga	1X2		Random forest	4.09%	0.198	0.964	408 (408)
15	Serie A	1X2	Quant Wizard	CatBoost	3.91%	0.188	0.951	608 (1044)
16	Serie A	1X2	Quant Wizard	TabPFN	3.38%	0.195	0.961	305 (305)
17	Ligue 1	1X2		Decision Trees	2.99%	0.213	1.026	633 (633)
18	Liga	1X2	Quant Wizard	TabPFN	2.81%	0.195	0.956	283 (283)
19	Serie A	1X2		Naive Bayes	2.56%	0.285	6.574	918 (919)
20	Primeira Liga	1X2	Quant Wizard	XGBoost	2.47%	0.180	0.914	1094 (1094)
21	Premier League	1X2		Naive Bayes	2.27%	0.208	1.002	291 (291)
22	Liga	1X2	Quant Wizard	CatBoost	2.26%	0.191	0.958	1055 (1055)
23	Brasileirão Serie A	1X2	Quant Wizard	CatBoost	2.03%	0.198	0.994	803 (803)
24	Ligue 1	1X2		Naive Bayes	1.93%	0.559	4.723	529 (529)
25	MLS	1X2	Quant Wizard	TabPFN	1.71%	0.206	0.992	174 (356)

Only true predictions made before a future game are included (no back test). The models are sorted by ROI using Pinnacle™ odds at the time of prediction. To qualify for the leaderboard, a model should have at least 150 predictions.

Better ROI than betting on the favorite using Pinnacle™ closing odds.

Better ROI than betting on a home team win.

Worse ROI than both baselines.