State of the art models for props using video data.
We generate player prop predictions with data from computer vison for video tracking of basketball games from the NBA, WNBA, and NCAA
We track the expected probabilities of scoring for every shot taken on the field, calculated using player movements, player shooting histories, defensive movement, locations on the court, and much more.
Using the novel tracking data, we generate predictions for all players with a given prop line in sportsbooks. Not only that, but we give model information and confidence metrics as a form of signal to help you make the best betting decisions.
Use our backtesting tool to see how our model has performed in the past. Filter based off model metrics to see the PnL and model win rates. Once you isolate a winning signal, look for the same metrics posted in current predictions.
Use statisticals tests to figure out if players have a positive or negative correlation between their results for points, rebounds, and assists. We give the direction of the correlation and its significance.
A complex video tracking setup that gives us an edge over sportsbook props
We ingest videos of games and use computer vision to track and identify players on the court. From there, we create our proprietary data.
We feed the tracking data with other stats into our prediction engine which creates models for each player. Filtering over 100 features with an advanced modeling process, we outperform current lines.
We backtest our models on historical data to ensure they are profitable. We use statistical tests and filters on model metadata to see where we have the strongest edge.
Sign up for a free account today and access our model and data pipeline.