Version 3.5 of WagerMate has been released. It has one major new feature: it uses a new track-to-track variant.
That one improvement gets us more than a 1% point increase in both the Win Rate and the ROI. Here are the back testing results on my 2009-2013 data, with the default filter:
This indicates that if you blindly followed WagerMate’s picks you’d lose 7%. If you only followed the picks for turf races, you’d win 1%.
By now, you should know that’s not how I really think you should use the WagerMate results — I want you to combine the WagerMate results with your own handicapping work. If you use the WagerMate results, instead of starting with an ROI of -20% (due to the takeout), you’re starting at -7%, and that’s a big difference.
The latest version of WagerMate, build 338, has a couple of minor improvements in it:
- WagerMate knows that Del Mar is switching back to dirt tomorrow.
- A bug related to saving user-scratch information is fixed.
I’m working on improving the track-to-track variant. It may take a couple of months to do all of the calculations, and insert the new work into WagerMate.
The latest version of WagerMate has a new feature…
You can now use the default WagerMate speed figure, or use the TrackMaster speed figure instead. You can even use the average of the two figures.
Here are the results when I use each of the three settings on my 2009-2013 data to look at Win bets (in this example, I’m not using any WMST file to filter out bad bets, so the ROIs aren’t very good). The “Average” setting has some nice attributes:
- More bets are made
- More bets are won
- Highest ROI
If we use the new “Average” setting and the default filters on my 2009-2013 data for back testing…
…WagerMate’s picks get an overall ROI of -8%. It’s interesting to note that the Turf Sprint bets would have paid a 3% profit over this span..
So, we don’t get guaranteed profits, but I think it’s nice to have an advantage in our handicapping — without applying a bit of our wisdom or experience, we’re way ahead of the game.
The latest version of WagerMate, build 322, has a couple of minor improvements in it:
- WagerMate ignores horses with morning lines odds over 20/1.
- WagerMate ignores horses that have raced 15 times and never won.
That leads to a slightly improved ROI…
Build 322 ROIs
I use the “Rqrd Odds” field of the WagerMate reports to make conditional bets with my ADWs. In essence, the Required Odds determine the lowest odds I’m willing to take.
Here’s how WagerMate calculates that number.
1) For each dollar bet at post time odds of X/Y, the payout will be (1 + X/Y) dollars. A bet at 6/1 odds will return about $7 for each $1 bet (1 + 6/1), or $14 for a $2 bet.
2) WagerMate calculates a probability of winning for each horse, the “Win %” field in the reports. If the probability of winning is 19%, then we’d expect our horse to win 19 out of 100 such races.
Imagine a $2 bet on each of those 100 (hypothetical) races — for a total of $200 bet. We expect to win $14 in each of 19 of those races, for a total return of $266.
That’s an ROI of 33% (a profit of $66 divided by an investment of $200).
3) The expected payout, in dollars for each dollar bet, is
Expected Payout per Dollar = (1 + X/Y) * (Win %)
In our example:
Expected Payout per Dollar = (1 + 6/1) * (0.19) = 1.33
4) WagerMate assumes we want an ROI of at least 20%. That’s an Expected Payout per Dollar of 1.20.
So 1.20 = (1 + X/Y) * (Win %)
Rearrange that to get the Required Odds: X/Y = [ 1.20 / (Win %)] – 1
For our example: [1.20 / 0.19] – 1 = 5.32
WagerMate reports that as 5/1.
I uploaded a new version of WagerMate today. The most noticeable enhancements are:
- WagerMate is using a better source for the “Race Conditions” (under “Race Details”) on the View Card tab. The old source was limited to 400 characters, and that meant the text was sometimes truncated.
- WagerMate is smarter about distinguishing graded stakes races from non-graded stakes.
- The Preference menu lets you pick your preferred format for ROI.
Here’s an example where the Race Conditions are complex, and are now shown in their full glory…
As for the ROI change: the menu looks like this…
Up until now, WagerMate has said that a profit of 5% was an ROI of 105%. Most people call that an ROI of 5%, not 105% (although one of my ADWs calls that a “$1 ROI” of “+0.05″).
The default format is now the “5%” format, so we’ve joined the mainstream. I’ll try to use that format from now on in this blog. I’m not going to go back and modify old entries though — I think that would violate the blogger’s code of ethics.
Here’s what my back testing (on about 4 year’s worth of data) looks like in the new format:
Build 321 Back Test Report
Typically, I’ll use WagerMate the night before a race to start my handicapping. At that point, the “also eligible” horses are still included, so if I click “Scratch Horses” on the Handicapping tab…
…I’ll see every horse on the card…
When I do the same thing a little before post time, WagerMate will download the latest scratch data, so that I see near-real-time information about the horses running…
If my pick for the race has been scratched, I can simply re-run the WagerMate simulations — WagerMate will automatically exclude the scratched horses.
You can also use this window to manually “scratch” any horses in the race, if you have a reason to exclude them (e.g. post position). This screen shot shows that I’ve told WagerMate to skip horses #1 and #2 when running the simulations. When I re-run the simulation, they’ll be treated as if they were scratched, and WagerMate will pick other horses to bet on.
Two horses manually scratched
Part of the process of releasing a new version of WagerMate is back testing. This figure shows how the new Build 318 of WagerMate performs, using the default strategy. This is based on about four year’s worth of data.
318 Back Test Report
So, if you blindly followed all of WagerMate’s suggestions, you would have won about 22% of your bets and got back 96% of the money you bet (i.e. a 4% loss).
In one sense, that’s terrible — we’d lose money! But in another sense, this is incredibly encouraging — before we’ve even applied a bit of our handicapping skill, we’ve changed the rules of the game. Everyone else is playing a game with a 20% takeout, but WagerMate users are playing a game with a 4% takeout.
If you can use your own experience and wisdom to sort through WagerMate’s selections, you should be able to improve on these results. If you have enough data to do your own back testing, you should be able to develop a better strategy than the default strategy built into WagerMate.
Of course, Your Mileage May Vary. Making money in back testing is not identical to making money with real bets.
I uploaded a new version of Wagermate today. Most of the enhancements are minor.
Here’s something convenient: when you’re working on the back testing tab to perfect a strategy, you can make it apply to just dirt races or just sprints (for example) by using the new checkboxes.
Back test tab with new checkboxes
I’ve taken to saving my strategy files with names that indicate what surface and distance they pertain to, like “DS long shots.wmst”.