Quantitative systems
Models, simulators and automation for operators and technical teams working in poker, sports arbitrage and game mathematics.
Illustrative execution example · no performance claim
A quoted margin is not an accepted position.
Two outcomes at decimal odds of 2.10 appear to offer a £10 profit on £200 staked. If only £40 of the second £100 bet is accepted, one outcome loses £56.
- Both bets accepted
- +£10 in either outcome
- Second bet partly accepted
- +£70 or −£56
- Hedge completed at 1.80
- £0 before fees
Synthetic arithmetic: exactly two exhaustive outcomes, identical settlement, no fees. The final hedge requires another £70 of accepted stake.
Follow the stakes, returns and limitsWhat the work produces
- Decision model
- Compare actions using the information available at the decision, including payouts, costs and assumptions about opponents or outcomes.
- Simulator or analysis tool
- Explore a game, test a strategy or review recorded activity against a defined baseline.
- Bot or integration
- Connect decisions to a permitted interface, with exposure limits, monitoring and stopping conditions.
The deliverable can be a report, model, source code or working tool, with the evaluation and operating notes needed to use it.
Choose the problem
Sports arbitrage
Compare prices for outcomes covering the same event under compatible settlement rules. Accepted size, commission and incomplete execution determine the resulting position. Predictive betting asks a different question: whether an estimated probability justifies a price. That estimate can be wrong.
Poker
Strategy models and hand analysis need game format, position, effective stacks, action history and opponent assumptions. Rake affects action value; tournament payouts can change the objective.
Poker models and examplesGames and incentives
Game mathematics examines paytables and configurations. Advantage-play analysis examines a proposed programme of play under eligibility, capital, time and volume constraints.
Where the system runs
Analysis of stored records, simulation and live integration have different requirements. Live automation needs a named environment, authorised users and platform permissions; a successful simulation does not establish permission or performance in a live game.
Stale inputs, rejected actions, exposure limits and stop conditions belong in the acceptance test alongside model quality.