Dog Ratings
 

I have been refining this idea for class rating for the dogs and i came up with the following list of class rating for the different grades of races at the dogs.


Grade of Race Rating
Group Races---62
FFA-----------58
Special Event--55
Grade 5-------53
Grade 4-------50
Mixed 4/5-----47
Maiden-------40

These are my thoughts.

Mad Gambler

Why have u rated grade 5 higher than grade 4??? It goes the other way. Maiden then to grade 5, 4, 3, 2, 1. So a grade 4 is for dogs that have won their grade 5 on that track. I think if you lose 2 races in one grade u can drop back a grade also. Also a dogs grade at each track is independent. So a grade 4 dog at Wenti Park could still be a grade 5 dog at Nowra, The Meadows etc.

Also surely the mixed grade 4/5 rating should be inbetween the grade 4 and grade 5 rating. It is slightly stronger than the grade 5 and slightly weaker than a grade 4.

Maiden Qualy
Maiden
Maiden Final
grade 5
grade 4
grade 3
grade 2
grade 1/FFA
Group Races

Youn can then fit Mixed grades inbetween ie a mixed 3/4 would be inbetween a grade 3 and grade 4. Special Events are hard to assess. Rate a FFA as a Grade 1 although it is probably slightly weaker. (They rarely have grade 1 and 2 races, but they do happen at Angle and Albion Park). Another problem is then lining up standard of each track. Certainly races at a metro track are stonger than the country but then the country tracks can be different. For example a grade 4 at Ballarat might be stronger than a FFA at say Wangaratta.

Relative and Absolute measures
 

There is a set of Grading rules at every track that specify how greyhounds move up and down between the grades. If you make a few general observations, you'll easily see what I mean. First, that dogs in higher grades are generally faster than dogs in lower grades. I think that's been proven beyond a shadow of a doubt. On average, the winning times of grade 3 races are faster than grade 5, which are faster than grade Maiden, and so on. The same is true for average running times as well, another well-studied result.
Now, just think about things for a few minutes, MG. Let's say that today's race is grade 4. And I'm looking at a dog with six past performance lines with the following grades: 444333. (The most recent one is on the left.) What happened to this dog? Obviously, he WAS a grade 3 dog, four races back. He shows three G3 races, then he dropped down to grade 4. Now ask youself a VERY SIMPLE QUESTION, Which will be a better PREDICTOR for a dog running in a GRADE 4 RACE? Three races he ran at grade 3, or three races he ran at grade 4?????? I hope you don't have to think TOO long to realize that three grade 4 races will be a better predictor of a dog's performance in a grade 4 race, over three races he ran against TOUGHER competition at grade 3.

Now let's look at what happens when a dog moves UP in grade. Today's race is still grade 4, but the dog's last six lines look like this: 444555. How about this? Well, those most recent three grade 4 lines should be a BETTER predictor than the three grade 5 lines, because in those three grade 5 races, the dog was running against LESSER competitors. Three grade 4 past performance lines are better at predicting performance in a grade 4 race, than three grade 5 lines.

The easiest is to look at predicting a dog's next time, and seeing how accurately you can do that. If you ONLY use lines in the same grade as for the current race at hand, you will find that the most recent three, and the previous three, give nearly IDENTICAL predictive value. You can do this on a dog-by-dog basis, and then combine lots of individual dogs, and you will find that there is NO predictive difference when you eliminate grade changing.

In a VERY HIGH PERCENTAGE of dogs, you will see grade changing within their last six past form lines. And if you see ANY grade changes, the most recent three lines will outpredict the earlier three lines, statistically. but as I've shown here, that is due to the grade changing rules for the track, and the fact that a dog will perform RELATIVE TO THE QUALITY OF THE COMPETITION. The more recent three lines are more probably from the same grade as today's race, than are the three previous lines, and are thus likely to be better predictors.

The universally observed truth that the average of four lines works better than the average of three, and the average of five lines works better than the average of four, points directly AGAINST the existence of "form". The exact same "grade change" forces are working AGAINST that observation (the more formlines in the average, the better) and yet that observation is STILL SEEN, even with as little as four to six lines of past performance history.
Now how do you use Grade in your handicapping, remembering it's a Relative measure and NOT an Absolute measure about "comparative" predictive value, and prediction "relative to today's field." Look at a time statistic like "average time behind the winner." (ATBW).This is a RELATIVE measure, and has a very strong (in-grade) component. If a dog in question is a true grade 4 dog, that sets an expectation for his ATBW average in grade 4. That also suggests he will have a LARGER ATBW average against grade 3 competition because he will likely finish farther back against those faster dogs. And it suggest he will have a SMALLER ATBW average against lesser grade 5 competition. So if I am using ATBW to predict some race TODAY, I would get far better results from ATBW numbers from in-grade past formlines lines,

Dogs win more races in the top couple of Grades,when they drop down in grade.(Up graders win even less in the top grades)
Dogs win more races in the Lower Grades when they rise.(down graders win less in the bottom grades ie. 4 to 5th grade)

Hope this helps in how you construct your Grade tables for each track an compared to Track to Track.

Another quality post Woof. Just to digress to the topic of overbet box numbers, if for eample in a typical dog race box 1 starts favourite on the tote at around the $ 2.50 mark, what price do thinkwould be a more accurate measure of the dogs real chances given that some of the money placed on him was purely by systems punters. In other words what % of money do you think is placed on box 1 purely because it is box 1, or how much is box 1 typically overbet by?

The overal result on the above dog is an ROI -18.8%, in an earlier post I advised that when you enter a race in your database you identify the race by the Fav's strongest Factor/Variable for later cluster analysis.

Say we have a $2.50 Fav in Box 1 who is the Best Speed Dog, but a slow beginner this type of dog is now a ROI -7.01% and the inverse of this type of dog is horrible -36.%.

Knowing the above overal result one could change from a vertical wager and move over to wagering horizontally in the exotics, a wager in the trifectas of 1/all/2 will show a ROI +11% for every race.

hope this helps somewhat

I know you are very rich Woof43 and you are a scientist and all that,but can you post some tips,as in "back this tonight",so us losers,well me, can win once.