How nkosikhona-ai/sa-quant-racing-predictor works
SA Quant Racing Predictor turns a Computaform/PhotoForm race-card PDF into extracted runners, configurable race predictions and betting intelligence. The main Streamlit workflow accepts and segments the meeting, lets users set race context and odds, invokes the prediction engine, and presents rankings, EV/value signals, race navigation, and exotic combinations. README-described model variants and meeting-level bets are represented, though their detailed implementations were not sampled. The app imports the extractor and predictor; relationships beyond the shown code and README are omitted or marked unverified.
Main entry points: app.py Streamlit app · predictor.py Prediction engine · extractor.py PDF parser · index.html Static fallback
Components 19
Race-card ingestion
- Streamlit app
app.py - PDF parser
extractor.py - Race setup
app.py
Prediction engine
- Prediction engine
predictor.py - Track models
predictor.py - ML calibration
predictor.py - Calibration model
calibration_model_v42.pkl
Betting interface
- Odds editor
app.py - Race rankings
app.py - EV and value
app.py - Race navigation
app.py - Exotic bets
app.py - Multi-race bets
app.py - EV meeting matrix
app.py - Copyable tables
app.py
Static fallback
- Static fallback
index.html - Static race data
data.js
Other components
- Bettor
- Race-card PDF
Connections 19
- Bettor → Streamlit app: configures and requests
README.md - Race-card PDF → Streamlit app: uploads
README.md - Streamlit app → PDF parser: extracts PDF
app.py - PDF parser → Race setup: extracts runners
README.md - Streamlit app → Race setup: configures races
README.md - Streamlit app → Odds editor: accepts odds
README.md - Streamlit app → Prediction engine: runs predictions
app.py - Prediction engine → Track models: selects model
predictor.py - Prediction engine → ML calibration: calibrates probabilities
README.md - ML calibration → Calibration model: loads model
- Prediction engine → Race rankings: returns predictions
README.md - Race rankings → EV and value: shows EV signals
README.md - Streamlit app → Race navigation: navigates races
app.py - Race rankings → Exotic bets: shows combinations
README.md - Race navigation → Multi-race bets: shows meeting bets
README.md - Race rankings → EV meeting matrix: summarizes EV
app.py - Race rankings → Copyable tables: formats results
README.md - Bettor → Static fallback: uses fallback
README.md - Static fallback → Static race data: uses race data