How bettingByte works

Edge is the gap between the model's estimated probability for an outcome and the implied probability baked into the market line.

If the model estimates a 58% chance and the book implies 52%, the edge is +6%. A positive edge means the bet is priced in your favor relative to the model — it does not guarantee a win.

Signals are bucketed as Moderate or Strong based on the size of that gap. We only surface signals with meaningful edge; thin-edge picks are discarded.

Confidence measures how reliable the underlying data is — sample size, consistency across comparable situations, and absence of confounding factors (injury, pace outlier, rest mismatch).

A signal can have strong edge but low confidence if the supporting sample is thin. A signal can have moderate edge and high confidence if the pattern is well-established. Both dimensions matter; neither alone tells the full story.

CLV measures whether a signal was published at a better number than where the market ultimately settled at close. Consistently beating the closing line is the most widely accepted indicator that a betting approach has genuine long-run value.

If a signal is published at O 221.5 and the line closes at O 223.5, the CLV is +2.0. This means the pick got value — the market moved in agreement.

Win rate and flat ROI are visible to all users. CLV tracking is a Pro feature and shown in the Performance tab alongside cumulative averages over time.

The Kelly Criterion is a formula that calculates the mathematically optimal stake size given a known edge and bankroll. Full Kelly maximises long-run bankroll growth but produces large swings that most bettors find uncomfortable.

bettingByte simulates two fractional Kelly multipliers in the Performance tab:

These are simulations for illustrative purposes. They assume no slippage and exact line prices. Real results will vary based on the lines available when you bet.

Every published signal is logged at the time of publication with the line price at that moment. Results are recorded only after the game has completed and the outcome is unambiguous.

Pending picks are excluded from all win rate, ROI, and Kelly calculations. Including unsettled picks would allow cherry-picking and inflate apparent performance — so we only count what is settled.