Hi Clarence,
Setting the weights is relatively easy. To set the weights for the handicapping factor, go to the Main Menu and select ‘Handicapping Custom Weights 599 Yards…. and then select which Handicapping factor you want to change.
Then enter the weight you wish to assign and click okay. You can also set separate weights for races 600 yards and longer.
But, the harder part of setting the weights is determining what the weights should be. I will try and describe how I go about it, and will also break it into level 1, 2, and 3, as you requested.
Level 1: Run a simple backtest:1)Since you are interested in Win, Quin and 1/23/234 betting, then 2,000 races should be sufficient for a baseline. If you were looking at Superfectas or even straight Trifectas, then you would need more races to feel good about the results since the Win % is so low. Open up the optimization screen by clicking on Tools Optimization. Statistically, the lower your Win %, then the higher the standard deviation, which means you want to run more races which will give you more confidence that your Win % is okay over the long run.
2)Select in the RaceID boxes the last 2,000 races. In my case, the database has 5,301 total races so I would set the Minimum RaceID to 3,301 and the Maximum race ID to 5,301 (note this is actually 2,001 races, but who’s counting?). In the Custom bet combo box (bottom right), select 1/23/234. Then click on ‘Start to Optimize Backtest.’
NOTE: I would NEVER set the Minimum Race ID to less than 500. When evaluating Race 500, GM will only look at races 1-499. If you, for example set this number to 20, then GM would only look at races 1-19 in the database, which would not provide enough history for good handicapping.
3)This backtest will run using the default handicapping factor weights, expressed in the order of Speed, Break, Close, Class, Trouble and Kennel, of 35,10,7,5,28,10 for 599 and shorter races and 35,10,10,3,28,10 for 600 yard and longer races (or whatever you have set for your default weights).
4)This run took 19 minutes on my computer. The more races in the database, then the longer an optimization run will take.
5)This resulted in:
a.Win = 29.0% Win percentage; ROI = 8.0%
b.Quin = 12.6% ROI = 6.6%
c.1/23/234 = 5.2% ROI = -16.2%
Doesn’t look too promising for 1/23/234, but at least Win and Quin are giving me a positive ROI, but both are low and don’t really excite me. Also, looking across the backtest grid, I noticed Perfectas/Exactas gave me a 24.3% ROI and Place bets were a 17.8% ROI. For me, anything above a 15% ROI deserves further investigation. Maybe instead of looking at Win, Quin and 1/23/234, I should look at Perfecta and Place bets here? Or I could continue to Level 2 and Level 3 below and try to optimize for my preferred style of betting.
Level 2: Run an optimization backtest for SPEED only, look at only Grade C and D races and only races 5xx yards long and look at situations only where the #1 picked dog has odds off of less than 2.5.1)Click on the checkbox at the top left next to Speed in the Optimization form. This will cause all the check boxes to be checked. When these are checked, GM will now use the min/max/steps detailed in the Optimization Parameters section of the Optimization form, and not the default weights like it did in the “Level 1” run above.
2)Select the following (expressed as min/max/step) :
a.Speed 20/50/5
b.Break 10/10/1
c.Close 7/7/1
d.Class 5/5/1
e.Trouble 28/28/1
f.Kennel 10/10/1
So, now we will be running multiple iterations and speed will be 20, 25, 30, 35, 40, 45, and 50 for each iteration.
3)In Length, at the top of the form, select 5xx and make sure ALL is not checked. I am only going to look at races that are 5xx yards long. To speed it up, I will also only look at Grades C and D also, so go to Grades in the top and check those two and make sure ALL is not checked.
4)I now want to know when the dog with the highest PR (as determined by my handicap system) has odds off of less than 2.5, what effect does that have? So, I check the box next to Low Odds, put 2.5 in the box next to it and then hit ‘Copy’ to copy it to all blocks. (Note: after looking at low odds, maybe the next run I look at odds off of greater than 2.5 by checking High Odds and putting 2.5 in all the boxes across, or maybe I look at odds between 2 and 5 by checking both high odds and lows odds and putting 2 in the low odds boxes and 5 in the high odds boxes…)
So, now I am varying only the Speed Weights for seven iterations, looking at only 5xx yard races that are Grade C or D and only at races where the dog with the highest PR went off at odds of 2.5 or better. I am still testing the last 2,000 races.
This run took me 1 hour and 45 minutes to complete.
I won’t list the details, but will say that in one of the runs, I had a 27.8% ROI for 1/23/234! A promising result that would require more investigation.
Now, once you narrow down speed, then try varying another variable like Break while keep all the others the same. See what effect it has.
Level 3: Change multiple Optimization Parameters and see what effect the two changes have.1) Select the following (expressed as min/max/step) :
a.Speed 20/50/5
b.Break 10/10/1
c.Close 7/7/1
d.Class 5/5/1
e.Trouble 5/30/5
f.Kennel 10/10/1
Now I am changing both Speed (20, 25, 30, 35, 40, 45, 50) and Trouble (5, 10, 15, 20, 25, 30) over multiple iterations. During this run, the boxes in the top right will keep track of which factor weighting produces the best WPS Win % and ROI %. Changing two or more factors results in more iterations and will obviously take longer.
Just to demonstrate some more ideas:
2)I want to look at situations where my #1 ranked dog has a significant advantage, according to my handicapping, over the rest of the field. So, I check the box next to low PR and set the boxes across all to 80. This means the #2 ranked dog and the rest of the dogs have a PR of less than 80. Or maybe I also want to look at situations where my handicapping thinks the #1 dog has an advantage, but the crowd thinks they don’t. So, I also check next to High Odds and put a 3 ( or 4 or 5) in the boxes across.
3)I only want to look at A or B Grade races, so I select A and B at the top of the form.
Running this run took me 3 hours and 45 minutes. The results were interesting and I had some ROIs over 30%....again that deserves further investigation. Any time I bet $1 and, on average, get $1.30 back is good.
Hopefully, that helps a bit. Sadly, there are no magic formulas to this. You have to backtest, examine the results, and then backtest some more.
Good handicapping!
Scott
www.GreyhoundMaster.com
