Frequently Asked Questions
Everything you need to know about the platform, methodology, and how to use Olympus Bets for quantitative sports analysis.
What is Olympus Bets?
Olympus Bets is a quantitative sports betting analytics platform that uses Monte Carlo simulation, Kelly Criterion bankroll management, and Bayesian probability calibration to identify edges across 9 professional sports leagues. It is a research and analytics tool that provides probability distributions, edge calculations, and optimal bet sizing. Every prediction is tracked and verified against actual outcomes with a transparent, immutable track record that anyone can audit. The platform is designed for analytical bettors who want data-driven insights rather than subjective opinions.
How do the predictions work?
Each game is simulated 10,000+ times using a league-specific Monte Carlo engine. Every iteration models the full game trajectory with randomized inputs drawn from calibrated probability distributions based on real player and team data. The aggregated results across all iterations produce win probabilities, spread cover probabilities, and total projections. These raw model probabilities are then calibrated using Bayesian methods to correct for overconfidence, compared against live sportsbook odds to calculate true edge, and sized using the Kelly Criterion for optimal bankroll allocation. Only picks that pass multiple independent quality gates are surfaced as recommendations.
What sports does Olympus Bets cover?
The platform covers 9 sports with dedicated simulation engines for each: NBA (possession-by-possession simulation), NHL (zone-based xG modeling), NFL (play-by-play with drive sequencing), College Basketball (player-level matchup simulation with EvanMiya data), College Football, Soccer (FBref xG play-by-play), MLB (pitcher-vs-batter matchup engine), League of Legends (Glicko-2 rating system with meta-aware modeling), and Esports. Each engine is calibrated against thousands of historical games specific to that league.
What is the Kelly Criterion and how is it used?
The Kelly Criterion is a mathematical formula for optimal bet sizing developed by John Kelly at Bell Labs in 1956. It calculates the fraction of your bankroll to wager based on the size of your edge and the odds being offered: Kelly % = (bp - q) / b, where b is the decimal odds minus 1, p is the probability of winning, and q is the probability of losing. Olympus Bets uses fractional Kelly with a unit mapping system (0.5u to 3.0u) and league-specific caps. A Bayesian shrinkage step is applied before Kelly sizing to prevent oversizing on overconfident probability estimates.
How are projections graded and what do the confidence tiers mean?
Every pick is assigned a unit size derived directly from the Kelly Criterion calculation. This is not a subjective assessment — it is mathematically derived from the difference between the model's calibrated probability and the market's implied probability. A 0.5-unit pick reflects a small but real edge, while a 3.0-unit pick reflects a large, high-confidence edge. League-specific caps ensure appropriate sizing: NHL and CBB are capped at 2.5 units, soccer at 2.0 units. The system also assigns descriptive tiers (Speculative, Standard, Confident, Strong, Very Strong, Maximum) corresponding to the unit ranges for quick reference.
What is the difference between free and premium projections?
Free picks are a curated daily selection of 3-5 picks chosen for diversity across leagues and bet types. Premium members get access to the full recommendation slate — every pick that passes the system's quality gates, typically 8-15 picks per day. Premium also includes detailed game-by-game analysis with probability distributions, key matchup factors, Kelly-optimized unit sizing, and access to the full historical performance database with league-level and bet-type-level breakdowns. Both tiers use identical simulation engines and methodology. The free tier is designed to let you evaluate the platform before subscribing.
How often are predictions updated?
The platform generates predictions 5 times daily, with league data refreshes happening throughout the day. NBA data updates at 9:30 AM EST, NHL at 12:00 PM EST, and CBB between 3-5 AM EST. Odds from major sportsbooks are refreshed multiple times per day. The system also runs injury-triggered re-simulations when significant lineup changes are detected between scheduled runs. Morning simulations are treated as authoritative — afternoon runs only override them if material new information (like a key injury) has emerged. Automated health checks run hourly to verify data freshness.
What is Bayesian probability calibration?
Bayesian calibration is a statistical technique that corrects for systematic overconfidence in model probabilities. Raw Monte Carlo output tends to overstate confidence: when the model outputs 70%, historical data may show the actual win rate is closer to 64%. The calibration system groups all historical predictions into probability bins and computes the actual win rate within each bin. It then maps future raw probabilities to these historically observed frequencies. The calibration is updated daily as new results are resolved, creating a self-correcting feedback loop. This is applied separately by league and bet type for precision.
Can I see the historical performance of the predictions?
Yes. The track record page displays complete, verified performance data including overall win rate, units profit and loss, and return on investment. Results are broken down by league, bet type, confidence tier, and time period. Every pick is recorded at the moment of publication with a timestamp and the odds available at that time. Results are resolved automatically using official scores — there is no manual editing, no retroactive changes, and no deleted losses. The immutable ledger can be audited by anyone, including the full chronological history of every pick ever made.
What data sources does the platform use?
Olympus Bets uses exclusively verified, real-time data sources. Live odds come from The Odds API, covering major US sportsbooks. Schedules, scores, and game results from ESPN's API. Player statistics from league-specific providers: NBA Stats for basketball, EvanMiya BPR ratings for college basketball, MoneyPuck for NHL expected goals and Corsi data, FBref for soccer expected goals, and Statcast for MLB advanced metrics. Injuries are aggregated from ESPN, RotoWire, and official team injury reports. No data is estimated, synthesized, or approximated. If a required data point is unavailable, the system skips the pick rather than using a fallback estimate.
How does Olympus Bets compare to other betting analytics platforms?
Most sports analytics platforms use regression models or ELO-based rating systems that produce single point estimates. Olympus Bets runs full Monte Carlo simulations producing complete probability distributions that capture the inherent variance in sports outcomes. Most platforms do not publish verified, auditable track records. The Kelly Criterion provides mathematically optimal bet sizing rather than subjective star ratings or "lock" labels. Self-learning subsystems (regime calibration, profitability zone analysis, cyclical intelligence) continuously adapt to changing market conditions. Visit the comparison page for a detailed feature-by-feature analysis.
Is there an API for developers?
Olympus Bets provides structured data access for premium members through machine-readable JSON endpoints. The platform exposes simulation results, probability distributions, edge calculations, and recommendation data in a format suitable for programmatic consumption. This enables quantitative bettors and developers to integrate Olympus Bets data into their own models, custom dashboards, or automated workflows. Data is updated on the same schedule as the web interface. Contact support for API documentation, authentication details, and information about rate limits.