To really gain any value from a Rating Service it needs to provide probabilities for each runner.

One thing about this whole idea that is important. Is the whole concept of overlay "accuracy" comes into play. For example, let's say I create a partition consisting of a group of runners that I/Rating Service say all have a winning probability within some fairly narrow range, say a couple of points around 25%. We can look at the crowd odds on those runners and come up with the crowd's version of a probability expectation for those runners.

The issue is, mine /or the Rating Service methods don't have to be EXACTLY accurate to be VALUABLE. If the crowd gives a runner a 15% chance of winning, and I/Rating Service give him a 25% chance of winning, if his ACTUAL probability is anything greater than the crowd's 15% peg, then I've found an overlay. I don't need him to actually get to my 25% estimate to have done something worthwhile. If he's higher than 15% AT ALL, then I'm doing the job. Obviously the more accurate my/Rating Service estimates are, the better, but all I have to do is push past the crowd probability on my side, and I've won.

The core concept is in identifying overlays. And an overlay can be 1 cent potential profit. After all, if I think a runner is a 20% overlay and I bet on him, and he turns out to only give me a 10% profit, then I've STILL made a nice profit, even though I didn't achieve my 20% estimated target!