Hi Mad Gambler,
Prior to doing to much of anything else, what you NEED to do is to start by developing a model for figuring out simple straight WIN odds. This is easy, because you actually HAVE the real win odds for every dog in every race. By modeling these odds, you'll be able to learn (and learn much quicker) how the crowd thinks and behaves.

So, to start moving down that path, you might want to think of a logical approach, in general. My first suggestion would be a series of goals, each one getting a little harder, and a little more useful.

First, you might consider predicting nothing more than which dog in each race will be the crowd favorite. Period. Get really good at this, because it's a MAJOR key to your success. You need to predict who the CROWD will choose as a favorite, because eventually you will find that the crowd bases their betting on the strength of this particular dog. If you don't know which one is going to be the crowd favorite, you'll never be able to develop ANY kind of odds model.

After you get REALLY good at picking the crowd favorite, you then go to the next level, which is predicting which dog the crowd will identify as the WORST in the field. Eventually this will be important as well, because you'll need to understand how punters ELIMINATE dogs, and what that does to betting habits.

After you get really good at identifying (simultaneously) the crowd's #1 pick and #8 pick, then you'll want to go for the whole shebang... ranking all eight dogs in the same order as the crowd ranks them. This is nothing except a ranking. At this point, you just want them in order, and you don't care about the actual odds.

After you get really good at the rankings, ONLY THEN will you start to look at actually predicting odds. At this point, start the process over. Start predicting the odds on JUST THE CROWD FAVORITE. Then go to the crowd's last pick.

Now you need to catorgise the type of race as determined by the punters,
"what will the crowd do with this thing." And I mean quite specifically.
This is a very important part for your future studies, you need to make sure you only ever catergorise races only based on what the punters can see in front of them(formguides).....
You have to look at wagering from a psychological viewpoint. People can't look at very many things simultaneously. The human brain isn't wired like that. People tend to scan over a race, and make some very broad judgments about what KIND of race they are looking at. Then from that bias, they go on to make decisions about what further information they will seek out. The broad view dictates both the kind of information they will look for, and how they will interpret what they see or find. It's like people naturally determine what's important, and that then colours their view of how to handicap and then how to bet on the race.
It's VERY important to be able to size up a race and determine what the crowd will immediately see about the race. And what particular factor is likely to be overplayed
What you are looking for are obvious factors that will get too much play, Remember, too, that a strong attention grabber may NOT make the dog in question into a crowd favorite. That obvious early speed runner in a good box may not run the distance, so it's not assured to be the favorite. But this kind of dog will siphon money AWAY from the dog who may STILL be the favorite, and a favorite that you AGREE with. In that case, you will be given an overlay on a favorite that is also your top-ranked pick. And that overlay will be generated because of excess play on another slightly weaker contender.

hope this helps