About TiltDataLabs
We find mispriced sports lines with machine learning. Public sentiment, recency bias, and line-setting heuristics leave systematic edges; careful modeling exploits them.
Every pick is committed to a public repo and signed to a public transparency log before first pitch — cryptographic proof we said it in advance. No cherry-picking, no hindsight edits, no selective disclosure. The full track record is public.
The Product
Market Edge
Daily pre-game picks with explicit edge and confidence metrics. You see the model's probability, the market's implied probability, and exactly where the disagreement is.
Kelly Sizing
Bankroll-optimal bet amounts grounded in the Kelly criterion — not arbitrary flat units. Each suggestion comes with a recommended stake proportional to the detected edge.
Multi-Sport
Models are done for MLB, NBA, and European soccer (EPL + Serie A). MLB is producing picks daily; NBA and soccer switch on when their seasons start. NFL launches with the 2026 preseason.
Public Proof
Each pick file is committed to a public repo AND signed to a public transparency log before games start, then submitted to OpenTimestamps for eventual Bitcoin anchoring. Two independent layers — anyone can verify the pick was locked pre-game.
Transparency
- •Full track record — performance stats are public on the Performance page, no login required.
- •Downloadable wager files — daily pick sheets are available as JSON for anyone to audit.
- •Independent verification — OTS proofs can be verified with open-source tooling. No trust in us is required.
- •No dollar amounts — all public curves are normalized to 100-unit starting points. Bankroll sizes are private.
Want the technical details?
Read about the ensemble model, Kelly bet sizing, and how the blockchain verification protocol works.
Read the methodology