Oddara is a probability and risk analytics platform. It does not guarantee outcomes, does not guarantee profit, and should be used responsibly.
1. Ingest odds across books and normalize formats.
2. Estimate fair no-vig probability per market.
3. Apply model probability and compute EV%.
4. Recommend stake sizing via Kelly fraction + cap.
5. Track realized PnL and ROI with full audit trail.
For each outcome line, we convert decimal odds into implied probability: p_implied = 1 / decimalOdds.
We then remove market overround (vig) by normalizing probabilities so they sum to 1.0: p_no_vig_i = p_implied_i / sum(p_implied_all_outcomes).
This provides a cleaner baseline for comparing model estimates and expected value.
EV is computed from model probability and payout odds. A simplified EV form is: EV = p_model * (decimalOdds - 1) - (1 - p_model).
Positive EV indicates a favorable long-run expectation under model assumptions, not a guaranteed short-term result.
Every EV signal is sensitive to data freshness, model quality, and market movement.
Full Kelly fraction is approximated as: f = (b * p - q) / b, where b = decimalOdds - 1, p = p_model, and q = 1 - p.
Oddara applies conservative controls with user-defined Kelly fraction and hard max-stake cap. Recommended stake is bounded by risk limits.
Risk controls reduce volatility but cannot remove drawdown risk.
Model probability represents an estimated chance for the selected outcome based on current market and feature inputs. Confidence reflects signal quality, not certainty.
Higher confidence does not imply guaranteed wins. It indicates stronger internal agreement and cleaner input conditions at decision time.
Model versions and assumptions can change over time; historical and live performance can differ.
No guaranteed profits. No guaranteed win rates. No certainty claims.
Our focus is disciplined probability modeling, risk controls, and auditable performance tracking.