Good idea Josh. My 2 cents worth is this:

1/Just run a bot and collect data of the field at an agreed time (official start perhaps?)

2/ Make use of the ratings that you scrape to find a flaw in the algorithm, which all ratings incorporate.

3/ After collecting price, price rank, volume, volume rank, rating, rating rank for several thousand races, you start to look at what combinations beat the inferred probability and one's that do. A scatter plot usually reveals this quite quickly.

4/ You can also rank or rate track conditions, class of race, distance, number of runners blah, blah.

At the end of the day you've created a monster that requires some number crunching.

I think fellow participants are best using a bot to collect the data themselves and paste into excel. After that, be they backer, layer or anything in between, they're on their own? OR you subscribe to a proprietory program to do all this for you, which this forum is well serviced for.

Nothing new really, just dedication to be patient and see it through. If you're prepared to put in the time you will reap the benefits.