OddsBeacon ingests match outcomes, scores, and player performance data across table tennis, tennis, and soccer — then runs a three-stage ensemble calibrated on forward-chained backtests. No look-ahead. No revision after publish. The pipeline is documented here in full.
PIPELINE · END TO END
From raw match feed to tonight's pick
01
INGEST
242.4K
Table tennis, tennis, soccer — 24 active leagues. 1.5M+ match records across 141K athletes.
02
CLEAN
~9.4M pts
Disambiguate player IDs, normalize set scoring, drop walkovers and retirements.
03
MODEL
3-ensemble
Anchor (baseline) · Edge · Momentum. Final output is probability-weighted Consensus.
04
VALIDATE
212.6K
Backtest forward only. No look-ahead. Calibrated with isotonic regression.
05
PUBLISH
14 min
Median latency from match start to prediction publish. Locked-state recorded at publish time.
BACKTEST · ROLLING MONTHLY · FORWARD-CHAINED
Accuracy curve
"Each month of training sharpens the edge — the models improve continuously."
Every prediction is timestamped at publication and never re-edited. Misses appear in the public record exactly as the hits do. Confidence tiers are calibrated: a tier should hit about as often as the model says it will, and the live data on the left shows each tier against its own record — including the low-confidence calls, which are most of the volume and sit near a coin flip by design.
PREDICTIONS SCORED
212.6K
HIGH CONFIDENCE
77.3%
MED CONFIDENCE
64.7%
LOW CONFIDENCE
52.6%
FORWARD-CHAINED BACKTEST · NO LOOK-AHEAD
MODEL TRAINING ACCURACY · PER LEAGUE · PER GENDER
Training-run performance
Accuracy and AUC recorded when each model was last trained, before any live deployment.
These numbers come directly from the model training pipeline — each row is the held-out test-set result from the most recent training run for that league and gender. AUC above 50% means the model ranks winners better than chance; accuracy shows how often the top-probability player wins.
TABLE TENNIS
LEAGUE
GENDER
ANCHOR SIGNAL
EDGE SIGNAL
MOMENTUM SIGNAL
TEST SAMPLES
ACC
AUC
ACC
AUC
ACC
AUC
TT-Cup
male
57.1%
60.2%
57.4%
60.3%
57.2%
60.2%
83,628
TT-Cup
female
61.2%
65.8%
61.4%
66.7%
60.1%
63.9%
4,456
TT-Elite
male
59.5%
64%
60.1%
64.6%
60%
64.6%
61,371
Setka Cup
male
55.9%
58.7%
56%
58.8%
55.9%
58.8%
114,192
Setka Cup
female
64%
68.5%
63.7%
68.3%
63.6%
68.2%
5,083
TENNIS
LEAGUE
GENDER
ANCHOR SIGNAL
EDGE SIGNAL
MOMENTUM SIGNAL
TEST SAMPLES
ACC
AUC
ACC
AUC
ACC
AUC
ATP Tour
both
67%
73.3%
67.5%
74.3%
67.5%
74.3%
73,864
WTA Tour
both
64.5%
70.1%
64.8%
70%
64.7%
69.5%
13,770
SOCCER
LEAGUE
GENDER
ANCHOR SIGNAL
EDGE SIGNAL
MOMENTUM SIGNAL
TEST SAMPLES
ACC
AUC
ACC
AUC
ACC
AUC
FA Women's Super League
female
51.7%
60.2%
47.6%
58.8%
44.9%
58.2%
292
Premier League
male
49.5%
63%
49.5%
63.1%
49.5%
63.1%
54,472
ACC = held-out test accuracy · AUC = area under ROC curve · Best model per league highlighted