|
Ratings Update
Since i have not looked at updating my ratings ideas in a few years and getting some mixed results i have decided to try a few changes and thought i would post my ideas for others to look at and maybe adjust for their own use.
Below you will see each category i use along with a brief explanation, i am using either a 10,8,6,4,2,0 or a 5,4,3,2,1,0 points ranking system, this is just a basic idea and would need further improvement. API Rank each runner based on the API value and assign the 10 point system. Win% Rank each runner from highest to lowest using 5 point system. Place% Rank each runner from highest to lowest using 5 point system. Distance/Track I was going to use percentage but found this to hard for example a runner with 1 start 1 win as opposed to a runner with 7 starts 5 wins so i decided to to use the number of wins and places so a runner with more wins and places would get a higher rating than one with less. I decided to give a weighting to those runners that had won over those that had placed so i allocated points like this. Double the number of wins so a runner that had 10:4-0-2 would get a score of 10 and a runner with 10:2-0-4 would get 8 i did the same for the track once i added all the figures i then ranked them in order highest to lowest and use the 5 point system. Track Condition If the track is rated slow or heavy i used the above method with a 5 point system. Fitness I calculated the days between the last run and second last run then added this together and ranked them lowest to highest then used the 10 point system. Class Class is always a difficult assignment with lots of different classes in each state so i decided to use the race prize money value as a guide, this not ideal but will do the job for now. I looked at the last race prize value and compared it with today's value if the value is more today than the class is higher and if it is less then it is lower. I then ranked them in order of the runner with the highest drop in class first using the 10 point system. Form I used the last 3 runs beaten margins and added them together to give me i final lengths beaten then i ranked them lowest to highest using the 10 point system. Qtab Rating I decided to use this as a final test just because i like them, these are ranked highest to lowest using the 10 point system. That's it the max score is 70 if track conditions are soft otherwise 65 is the highest possible score. I then used a calculation to get a price for each runner but because of the limited scores and the likely hood runners could be close in score i can't say the price will be accurate i will still need to work on that. Here are a couple races for today that i ran this over. Code:
|
Hi Shaun,
Very interesting approach - good luck with your ratings. One aspect that always "worries me" is horses that have had no starts at track or distance. You can not really penalise them as they have not had any starts at the track or distance, so they have no history. Compare this with a horse that has had 5 starts at distance for zero placings - do they score that same rating? Your comments? |
Shaun,
re your pricing thoughts, if you can get your head round this it could be of interest for your style of ratings. http://www.racebase.co.nz/market.htm |
Yes i agree, seeing as this is a mathematical and automated approach i would need to come up with a better option, maybe i could look at the starts and work something out using that.
|
Yes have seen that and although it never suited my old ratings it may suit this type, will check it out and compare the prices.
|
Quote:
Looks pretty good, but I've always found that using the median field career prizemoney, a better guide than race prizemoney. |
Quote:
Agree, this is where a lot fall down. Tend to rate horses with good track or distance too highly, because 0 starts also means 0 failures. A horse should be penalised for failures, rather than rated because of success when it comes to track and distance. In fact I find more success, doing reverse ratings. i.e. all horses start with a rating and deduct points for various failures. Also gets around the pricing issue, as bolters truely become bolters and favoured horses come up shorter than the other way. ...imho ;) |
Quote:
Can you give me an example of this please. |
I agree that horses with no form at the track or distance should not have zero rating. It would be better to use say a 5 point scale and give these horses 3 points and have 4 points for horses that have performed reasonably well and 5 points for those that score extremely at the distance. Give 1 point for a horse that has many failures and 2 points for at least one failure.
I assume you would have had no bets on winners as your prices were above those on offer. |
Prices were not accurate today as i was still working on that part.
|
I thought i would try this again, nothing changed on the ratings as yet but an adjustment to the pricing.
Code:
|
I always been interested in doing my own ratings to see how they compare. I see all the things you mention as good to rate on.
How how do you work this or is it a secret for now? |
Not sure what you mean by a secret i have shown what i do but if you have specific question ask away will give what ever answer i can.
|
Shaun, I don't know if you are into exotics or check them with your selections, but today they snared the quinellas and trifectas in the three races for POT of more than 100%.
|
Yeah nice result that.
|
I thought i would see how these go with better quality races.
Flem 3 only has 4 selections because they stuffed up the form in that race on Cyberhorse. Code:
|
Quote:
Shaun you might like to redownload the form, I have nine runners. |
I ment i only had 4 selections because it selected the same horse twice if you look at http://formguide.cyberhorse.com.au/...k=Flem&raceno=3
you will see they had Module listed twice with different stats. |
Quote:
Shaun, that's not all that unusual for them. Most times it happens you'll find different trainers are listed for each entry. I dunno if they're aware of it or not. I just use a SS column to alert me if the one name comes up twice. Keeps me on me toes.....:) |
Yeah i might add that in mine as i have seen it a few times.
|
Quote:
It's deliberate I think, I use the paid for zip files and never a problem! |
Code:
|
I've obviously missed soemthing then Shaun as I cannot replicate what you are doing with the info in this thread, but then I am still a little new to all this, especially ratings as I have always avoided it till now.
|
I will give you an example of 1 runner from a race today, all results are after scratchings removed
Muswellbrook Race 5 2 FASHION STATEMENT Highest API position 1 = 10 Points 5th Rated in win% = 1 Point 3rd Rated in place% = 3 Points Dist form 2:0-0-1 track form 0:0-0-0 gives him 1 point and makes it equal 5th rated = 2 Points As track is rated Dead nothing added for going. Days since L/S 12 2ndL/S 33 add these together and rank 3 = 6 Points Class last race was $27000 today is $20000 so down $7000 rank 1 = 10 Points Last 3 beaten margins it rated lower than the top 5 = 0 Points (This could be an interesting stat as the last 3 runs were poor) Qtab Rating 92 2nd rated = 8 Points Total = 40 here are the ratings for that race. No2 40.00 No3 48.00 No4 43.00 No5 35.00 No6 11.00 No7 21.00 No8 8.00 To get the prices i use the method from this site http://www.racebase.co.nz/market.htm Then i convert this to a 100% market but still working on the price part. Even with the prices done the rating order does not change so if you can get that part right then you would get the same runners. If you are doing it with a manual process i would just do the top races of the day, i have found i get better results from these and will be doing the same. The race i have done is the highest prize money on offer for today. |
Wow! How long does it take to do all this?
btw, what is "API"? |
Because i import everything i need in to excel takes me about 30 seconds a race.
API is the Average Prizemoney Index this is the average amount of money won at each start, the higher the average the more the horse has won, it is one way of working out the class runner of the race in some sense. |
Quote:
Quote:
|
Dennis, you might be able to help me out with this problem i have.
http://forums.ozmium.com.au/showthread.php?t=22134 |
Shaun,
I had the same difficulties a few months back - after googling til I went google, I never actually did solve the problem. It seems to only affect columns A & B. I managed to work around it as the data that was scrambled was data I didn't need at that time. I might go to MrExcel to see if I can find a solution. |
Yeah i am after the Dist/Trak/Going info sometimes it is fine and other times no good, seams when the starts is over 10 it gets scrambled.
|
Quote:
However, if cell format is "Text" then it remains as it is! Did you try formatting first 2 columns to text?" and now buried on page 12 :( |
Yep tried to format as text the whole sheet but as soon as i import the next page it just changes it back, i am thinking there is a setting i can use to not import the formats but i have tried everything i know.
|
Quote:
I don't use Excel as such for this, but did have a look at the page source on the Cyber site, and as expected the formatting is the same for every race, so it's definitely an Excel peculiarity. I will look at it and let you know. Cheers |
Quote:
I actually saved the page and edited the number of runs and still comes out as date for some reason. Will work on it later. Cheers |
Quote:
No answer on the net either yet, but I'm sure there is a solution, I just haven't got it. Drives you mad, glad I'm using a database and not Excel. Good luck with it, let us know if you find anything. Cheers |
Quote:
|
If you have disabled date recognition , then surely it won't display as a date - or does it still?
|
Still does
|
Quote:
Unfortunately I only have the latest. Just wondering? |
2010 Professional
|
| All times are GMT +10. The time now is 02:50 PM. |
Powered by: vBulletin Version 3.0.3
Copyright ©2000 - 2025, Jelsoft Enterprises Ltd.