PDA

View Full Version : calculating odds


Freightliner
04-21-2011, 11:44 PM
I have been away for a while, but even when I was very active, I never calculated my own odds line. Can anyone suggest a method or a book on this subject?

pondman
04-22-2011, 11:42 AM
I have been away for a while, but even when I was very active, I never calculated my own odds line. Can anyone suggest a method or a book on this subject?

Observe at least two variables for awhile. If they occur together at 40%, then your break even, in concept, would be 1.5 - 1 (plus your returned $2.) And then you need to leave yourself a cushion for the longterm health of you bankroll.

It's really up to you as to the variables. What track do you play? Class? Surface? Distance?

headhawg
04-22-2011, 12:06 PM
These should give you a good start: Four Quarters of Horse Investing (Steve Fierro); Winning Thoroughbred Strategies or Commonsense Betting (both by Dick Mitchell); Value Handicapping (Mark Cramer).

Ray2000
04-22-2011, 12:41 PM
If you looking for a math expression that can be used to make a line based on your ratings for each entry, then the attached spreadsheet can be used to set a line. Since rating numbers can be on a lot of different scales, you may want to do some adjustment to best fit your numbers. Also, adjusting constant "C" to 1 assumes no Takeout and the probability sum is unity. Set to 1 + Takeout for a line closer to ToteBoard Odds.

Let me know if anyone finds any errors, I put it together on the fly.

Jay Trotter
04-22-2011, 02:32 PM
Ray,

What exactly is "Set B to make longs, longer, shorts shorter"?

Please explain what the chart is telling me?

Thanks:ThmbUp:

Ray2000
04-22-2011, 03:13 PM
Jay


The constant "B" defines the curvature of the line. As an example, as downloaded, the 1 horse is the fave at 6/5 and the 6 is at 11/1. By adjusting "B" you can make Horse #1 ...4/5 Odds (shorter) and the 6 goes to 12/1 (slightly longer) I've found that in my rating system the favorite's odds were always too high so I put that adjustment in to try and match real time odds more closely.

Someone else's rating system may need a different setting.

Jay Trotter
04-22-2011, 03:43 PM
Thanks Ray.

On a slightly different problem, I'm "Importing Odds (that refresh) into Excel" as outlined by CBedo into a seperate page in your document and then trying to copy them over to your sheet. I'm importing okay but when the odds show up on the tote as fractions (ie. 5/2) they are showing up as "0". Any thoughts on how to format the cell so that the odds come over exactly as shown on the tote?

Much appreciated:ThmbUp:

TrifectaMike
04-22-2011, 03:48 PM
If you looking for a math expression that can be used to make a line based on your ratings for each entry, then the attached spreadsheet can be used to set a line. Since rating numbers can be on a lot of different scales, you may want to do some adjustment to best fit your numbers. Also, adjusting constant "C" to 1 assumes no Takeout and the probability sum is unity. Set to 1 + Takeout for a line closer to ToteBoard Odds.

Let me know if anyone finds any errors, I put it together on the fly.

Very nice. Z-scores and an approximation to a guassian. I'm impressed. Just one question. How do you determine that the ratings are distributed according to a guassian. Especially since you are suggesting others to use the algorithm.

Mike (Dr Beav)

Freightliner
04-22-2011, 04:16 PM
So much information. It's awesome. This is my plan, so you can lampoon me if it sounds stupid:
1-determine the correlation between ML odds and actual winning %;
2-determine the correlation between ML odds (rank order posion) and actual finish position;
3-determine the correlation between various data items in the PP with actual winning %;
4-determine the correlation between calculated numbers (i.e internal times) and the actual winning %;
5-graph all that crap
6-using some trigonometry, determine the slope of each line.
Some lines will have slopes that are condusive to winning wagers over the long haul. I just have to figure out which lines by careful reverse bankroll analysis.

My database is currently kind of small (25,000 individual PPs from approximately 6,000 races). I should have a fairly large dB by the end of the summer season for the midwestern tracks.

Robert Goren
04-22-2011, 04:38 PM
A method I use for making a M/L some times.
1)Take odds in the last race and convert to a %
2)adjust for finish, class change , days away, jockey change etc(this the tricky part, but you quickly learn whats. A favorite who runs 2nd in the class stays the same. A 20/1 shot doing the same gets about 10% added to it . Ramon in NY gets 10-15% when they switch to him. ect)
3)add them up and adjust till the total = 125%
4)convert back to odds.
This gives you idea of what the pool odds will be, necessarily his chance of winning.

Ray2000
04-22-2011, 05:17 PM
Jay Trotter
Not sure which sheet you're referencing but
If the imported data is showing up in the import range as a 0 when a "/" is involved then right click any cell in the import range and select "Data Range Properties" and make sure "Preserve Cell Formatting" is checked. Save the Workbook.

Any reference to a cell in the import range should show as imported but the "fractions" 5/2, 6/5 etc need to be converted to a number to use in a formula. I use a a Vlookup table similar to the one in the spreadsheet I attached but in reverse, to result in a number.

does this help?

Trifecta Mike
I use the formula simply because it "best fits" the empirical data for my rating system. I present it as an aide to help with a problem Oonga Boonga ran into when he painted the first Morning Line on the cave wall. :) How do you get from a rating, expressed in point totals, to a probability of winning? Setting the lowest ranked horse's chance to 0 doesn't work and constructing tables is too much work. So I settled on a natural log approach and let others see if it works for them.

FreightLiner

Good Luck, All I can say is I play almost all Harness and ML correlation is my biggest gripe. There may be some relation of Morning line to the phases of the moon but not to winning percentages at the track....:)

Freightliner
04-22-2011, 06:45 PM
FreightLiner

Good Luck, All I can say is I play almost all Harness and ML correlation is my biggest gripe. There may be some relation of Morning line to the phases of the moon but not to winning percentages at the track....:)
I am not sure if there is a correlation or not.

My problem is I do not have the time to watch tote boards. I have been using ML as a stunt double for the tote board. I am currently running at 100.48% (for every $100 I bet I get back $100.48) over a stretch of 70 races. I believe there will be a weak correlation to the winning percentage. Knowing that correlation will allow me to test other factors and develop my own odds.

My goal is to stop using the ML because it usually bears no resmeblance to the tote board.