Senior Data Scientist — Monte Carlo Sports Prediction Engine (Python, Statistics)
Confidential employer
Australian sports-analytics company. We run a true native-joint Monte Carlo simulation engine — 100,000 correlated simulation paths per game — publishing daily player-prop and game-line predictions across WNBA, MLB and AFL, with NFL and NBA launching in September. The system already has automated calibration, closing-line-value tracking, Brier/reliability reporting and a fully graded public track record. We need a very senior data scientist to make it the most accurate engine in the market.
WHAT YOU'LL DO
- Audit our v3.8 engine end-to-end and put findings in writing
- Build and backtest the NFL model before launch (backtest-before-deploy is law here)
- Own the calibration loop: bias correction, confidence rescaling, category-level performance auditing
- Extend the prediction side: opponent positional-defence modelling, schedule/rest effects, correlation-aware exposure
- Explain everything in plain English to a non-technical founder
YOU MUST HAVE — WITH PROOF
- You have BUILT AND SHIPPED a probabilistic simulation or pricing model in sports betting, quant trading, bookmaking or similar. On our call you will walk through one such model end-to-end: data, joint structure, validation, real-world results, and one way it failed.
- Deep Monte Carlo + joint-distribution craft (copulas, hierarchical/Bayesian structures, correlated outcomes)
- Working fluency in calibration: Brier score, reliability curves, closing-line value, de-vigging
- Small-sample discipline — you know why a 10-2 streak usually means nothing
- Strong Python/NumPy; strong written English; AEST overlap
We screen hard against overclaiming: "my model hits 65%" without sample size, odds and CLV context ends the conversation.
PROCESS: CV review ? technical call ? short PAID trial task (1–3 hrs, no code, tests judgement) ? decision within the week.
TO APPLY: send (1) your CV and (2) one paragraph describing the most similar system you have personally built, including how you validated it. Applications without the paragraph will not be reviewed. Start your reply with the word JOINT so we know you read this.
Track this external role
Sign in to save this listing and apply through the original source.
Sign in to SaveReport JobOpen Original Listing