I do my own ratings and if I believe that knowing the class of the race compared to the class of the previous races a horse has been in is the most important thing of all. I basicaly only include weight carried, margin from the winner and jockey to calculate the horses future rating and am currently running at a profit from the ratings it produces.
I don't think you can blindly use a rating for a race based on anything as simple as the type of race it is. The strength of a race should be worked out from the horses that are running in that race. I've only been working on my system for about 9 months and look forward to many years of fun fine tuning and tinkering with it.
An interesting thing to consider when putting in filters and whatnot is that you don't only have to filter according to what is going to make a horse more likely to win. You have to balance that against what is more likely to bring a horses price down. For example I added a filter based on prizemoney of a horses previous race. I gave it a bonus based on a higher prizemoney race at it's previous start. Not only did that not work but when I did the opposite and gave it a penalty for competing in a higher prizemoney race last start I got a better return. A slightly diminished percentage of winners but a surprising increase in POT. I can only suggest this is because the average punter places more importance on the prizemoney of a horses last start than is warranted.