O.K. the premise was that backing the overlays only should return profit and the underlays should return loss.

Going by actual TAB price the reverse transpired.

The shorter the price, the more likely to win and the loss less, the greater the price, the less likely to win and the greater the loss.

__________________
RaceCensus - powerful system testing software.
Now with over 422,000 Metropolitan, Provincial and Country races!
http://www.propun.com.au/horse_raci...ng_systems.html
*RaceCensus now updated to 31/07/2025
Video overview of RaceCensus here:
http://www.youtube.com/watch?v=W821YP_b0Pg

So I do the exercise again, based on 100 raters having an impact value of 2.38 and priced to field size.

Looking at overlays to rated price, I come up with:
3,290 selections
440 winners
13.37% strike rate
$2,692.80 return
$597.20 loss
18.15% loss on turnover.

And this is rated to 100% market.

Clearly the market price is outperforming the ratings by a substantial amount, even though a 100 rated horse is 2.38 times more likely to win.

The reverse scenario is to look at horses under the rated odds to prove the price guide is better than the 2.38 times greater rated price.

2,897 selections
973 winners
33.59% strike rate
$2,538.10 return
$358.90 loss
12.39% loss on turnover

So a horse under the rated price is far more likely to win, and more likely to return less loss than backing overlays.

So the ratings still can't beat the market price despite being 2.38 times better than the raw data.

__________________
RaceCensus - powerful system testing software.
Now with over 422,000 Metropolitan, Provincial and Country races!
http://www.propun.com.au/horse_raci...ng_systems.html
*RaceCensus now updated to 31/07/2025
Video overview of RaceCensus here:
http://www.youtube.com/watch?v=W821YP_b0Pg

So taking this a step further, one would imagine if you could lay a horse at half the rated price and back it at twice the rated price in a 100% market, one should be able to make money.

should

1266 selections obtaining twice the rated price
105 winners
8.29% strike rate
$1,026.80 return
$239.20 loss
18.89% loss

Half the rated price or less

6,185 selections
1,412 winners
22.83% strike rate
$5,230.90 return
$954.10 loss
15.43% loss on turnover

One is better taking lower priced 100 raters, one loses less percentage and nearly three times the strike rate of "overlays".

Market Price is better value than the rating, but the rating is a better guide to winners.

__________________
RaceCensus - powerful system testing software.
Now with over 422,000 Metropolitan, Provincial and Country races!
http://www.propun.com.au/horse_raci...ng_systems.html
*RaceCensus now updated to 31/07/2025
Video overview of RaceCensus here:
http://www.youtube.com/watch?v=W821YP_b0Pg

It is an interesting conumdrum and a clear demonstration of the inadequacy of ratings by comparison to market prices.

Of course, there are ratings that outperform the market, this was never the argument, the argument was the accuracy and value of ratings by direct comparison to market influence.

__________________
RaceCensus - powerful system testing software.
Now with over 422,000 Metropolitan, Provincial and Country races!
http://www.propun.com.au/horse_raci...ng_systems.html
*RaceCensus now updated to 31/07/2025
Video overview of RaceCensus here:
http://www.youtube.com/watch?v=W821YP_b0Pg

Chrome,

The problem is that most ratings are setup to imprive the strike rate not the profit. Using the impact values for profit instead of strike rate will allow a more realistic rating system that can then be used for overs/unders analysis.

Choosing which values to put into the ratings and how to weight them is the key though and this is also the hardest part.

Good Luck

To be honest I'm glad I started this thread because I hadn't even considered your suggestion and it's definitely food for thought.
Thanks wesmip1, that's another idea to test.

Cheers.

__________________
RaceCensus - powerful system testing software.
Now with over 422,000 Metropolitan, Provincial and Country races!
http://www.propun.com.au/horse_raci...ng_systems.html
*RaceCensus now updated to 31/07/2025
Video overview of RaceCensus here:
http://www.youtube.com/watch?v=W821YP_b0Pg

While I haven't tested laying the horses starting under the price for a profit, the simple answer is Yes. I have ratings which provide a profit for every top rater straight out, and a better profit for every top rater that is an overlay.

They do exist.

__________________
Permanence is an illusion

I think that is really the benchmark. If the ratings provide a profit on overlays, and that profit is in direct proportion to the amount of overlay (20% overlay equals 20% profit rated to 100% market), then you can pretty much brand them value over a decent number of races as a sample.

__________________
RaceCensus - powerful system testing software.
Now with over 422,000 Metropolitan, Provincial and Country races!
http://www.propun.com.au/horse_raci...ng_systems.html
*RaceCensus now updated to 31/07/2025
Video overview of RaceCensus here:
http://www.youtube.com/watch?v=W821YP_b0Pg

Its excellent that you have delved into your past figures and are trying to make sense of them.
But this is where you are falling down.
Its your paradigm that is mistaken, not your Ratings.
It has been shown many times that if you backed every single horse that runs, you will lose around 15 % of your investment.
This is not some remarkable figure.
It is not just how it is.
It is the exact same figure that is the deduction that those who run the pools deduct.

Usually if you back those that are under , say $10, the actual loss will be closer to 10%, and if over the $10, then your losses will be closer to 20%.
The reason for that is because those under $10 are actually starting at a price that is pretty much their actual Chance of winning.
Those over $10 , actually have a much less chance of winning than their Markey Price would indicate.

This percentage loss will occur whether you only back the topweights, or if you only back Barrier 1, or if you only back the Favourite.
OR IF YOU ONLY BACK THE TOP RATER.

UNLESS.

The TOP Rater is the top Rater because it really is.
After the race, the Winner should have been the top Rater, because it won.
If it wasn't, then the Race wasn't analysed correctly, and the method used to Rate the horses needs improvement.

I'm sure that the only way to analyse a race 'correctly' to decide the genuine top rater in a race, is after it has been run. If it was so easy, we would all have recognised and backed the winner of the C/Cup which was the correct top rater and at massive overlay odds.

Personally I've never seen any publicly available ratings that produce a long term profit and correctly recognising 'overlays' is a subjective guess, not a science that only requires correct computing.