Industry leading modelling is at the heart of the operation. Vast amounts of betting and score related data is captured, cleansed and stored before being interrogated by a specialised team of QA’s. Prototype models are built, tested and then either iteratively improved or sent back to the drawing board. Game theories and market interdependencies are analysed. It is only when the modellers are confident in the underlying strength of an algorithm that both the methodology and outputs are passed to the trading team for refinement or rejection. Only once there is acceptance by the traders is a full model build commissioned.
It is a quality control process that makes all of the 20+ models supplied to date “best of breed”, in both operation and output. Understanding the difference between the “right price” and the “tradable price” is a trader based skill – algorithms don’t know when it has become overcast at Lords and how much it will make the ball swing.
Traders pre-empt subsequent market movements, so removing the costly time-lag from just tracking prices. Models are built on underlying probabilities from which any variety of output (spreads or fixed odds, for example) can be produced simultaneously. The traders refine the model outputs in-play to maximise value. This allows partners to benefit from the same respected in-play trading skills as Sporting Index, and they have comfort in the knowledge that Sporting Index uses the identical technology and decision making to underpin its own business.