Database Handicapping Software- JCapper CANRUN/CANTRUN Filters
June, 2005
June, 2005

Necessity- the Mother of Invention
The other day, Wed 06-08-2005 to be exact, I bet a couple of J1 horses in turf races at Belmont Park. Both ran so poorly that it kind of left me scratching my head afterwards.

Below is a link to the HTML Report for the 06-08-05 Belmont card so that you can follow along:

The first horse was MIKES MISSILE in R4. The other was CANDY ADVENTURE in R9.

MIKES MISSILE was a J1 horse at odds of 11-1 and was selected by a handful of my own UDMs. Each of my own UDMs employs a slightly different factor mix. My strategy in using multiple UDMs is aimed at identifying logical contenders likely to be let go in the betting using different but hopefully valid perspectives. But MIKES MISSILE put in a dismal performance. He was 8th entering the far turn against fractions of 23.1 47.1 and 111.2, and even though Luzzi made a vigorous effort to rouse him at the top of the stretch, MIKES MISSILE actually lost ground from that point and finished well up the track - beaten almost 13 lengths. This is a considerable margin of defeat for any turf race. This also left quite an impression on me especially considering the way he was asked by the rider to run. IMHO, or at least based on my HTML Report, MIKES MISSILE should have given a much better account of himself.

CANDY ADVENTURE was a J1 horse at odds of 27-1. While not selected by any of my UDMs, this horse was picked up on the HTML Report by JCapper's Overlay Highlighting feature. Since its inception, Overlay Highlighting has been very very good to me - and hopefully - other players as well. At 27-1 I liked the horse enough to take a shot. CANDY ADVENTURE trailed the field to the top of the lane and came up empty when asked to run. Her performance was far worse than what I expected. In fact it looked eerily similar to the race run by MIKES MISSLE just a few hours earlier.

Every now and then I like to go back and review my plays both good and bad. I sometimes find that a little time spent doing research yields dividends later.

Moving past the fact that MIKES MISSILE was a J1 horse and was selected by 3 different UDMs and was 11-1, are there any clues on the HTML Report that might have hinted at his upcoming dismal performance?

MIKES MISSILE was 8th in CPace and 7th in PMI. He was 5th in TPace, 7th in PAL, 5th in L3. To his credit he was 3rd in Form and WoBrill. Obviously, MIKES MISSILE doesn't stand out in any one category.

How about the other horse? CANDY ADVENTURE was 8th in CPace and 7th in PMI. She had 0 Optimization Points and was 7th in TPace and 9th in Sustained Pace. To her credit she was 3rd in PAL and 2nd in L3. The numbers said that she had no early speed but did have some ability to run late. The Pace Index of the race was on the low side at 44.79. When she turned for home trailing the field against slow fractions of 24.2 and 48.1 she had no chance whatsoever. In hindsight, her dismal performance should have come as no surprise.

How did both these horses end up as J1? The JRating algorithm combines the rankings and gaps for Speed and Pace, Fig history, Form, and Connections together into a single number. I created the algorithm to do this based on large numbers of historical occurances. Admittedly, it's not a perfect algorithm. It will, in isolated cases, point to horses like MIKES MISSILE and CANDY ADVENTURE. But this same algorithm also points out lots of overlay winners. And it does do what a human brain can not do. It analyzes race data and performs a lengthy calculation in less time than it takes you to blink your eyes. Overall it does a pretty good job. Historically, J1 horses win about 1 out of every 4 races. The value of the JRating algorithm lies, not in the fact that it wins 1 out of every 4 races, but that the winners it points to tend to be horses that the public is somewhat afraid to bet.

If you think about it, not all J1 horses are created equal. Sprinkled in among those J1 horses that manage to win 25 percent of the time (or lose 75 percent of the time) there will always be many identifiable sets of plays that stand a greater than 25 percent chance of winning. Conversely, if you spend some time at the Data Window, you will find sets of J1 horses that have a much greater than 75 percent chance of losing. The real crux of successful wagering lies in understanding the race in front of you - or at least in understanding it better than the general public does.

A Simple Belief System
Back in the days when I used to handicap with pen and paper, I had a simple belief system. At times that belief system served me quite well. The essence of my belief system was this: Try and understand ways contending horses might run and win the race in front of me. In cases where I had that understanding, those were my plays. Cases where I had no understanding? No bet. Simple as that.

Some horses run on or near the lead. They are early threats. Some horses consistently close. They are late threats. Some horses run evenly. I'm not sure if anyone has ever coined the phrase "an even threat." But horses that run evenly, because they never lose contact with the field, and are seldom pushed to the point of being tired before turning for home, are often great threats to win the race.

I used to have a saying:

This saying was valid for me 25 years ago. And I think it is just as valid today.

Revisiting the HTML Report for MIKES MISSILE and CANDY ADVENTURE in the context of that saying, I'm a little abashed that I hadn't created any Filters in JCapper to take advantage of that belief system.

Well, that just changed.

JCapper has a number of factors that measure a horse's ability to run early. CPace, PMI, and AvgE1 all do a very good job of this.

JCapper also has a number of factors that measure a horse's ability to run late. PAL, Late Pace (Best of Last 3), and Late Pace (Last Start) all do a very good job of that.

Sustained Pace and TPace, depending on the strength of either the E2 or Late pace figure(s) used, can indicate both types of horses. A dominant E2 can result in a top ranking for either Sustained Pace or TPace. Dominant E2 numbers in either situation can point to a horse with the ability to win by running early. Likewise, a dominant Late Pace Figure can result in top rankings for Sustained Pace and/or TPace. Dominant Late Pace figures in either situation can point to a horse with the ability to win by making a late move.

Even runners, at first glance, can be a little more difficult to conceptualize. However, the very same factors used to identify both early and late runners can also be used to identify horses that run evenly. For example: Doesn't a horse that ranks 4th in both CPace and Late Pace fit the mold of an even runner? Sure he does.

Introducing the CANRUN and CANTRUN Filters
Because both MIKES MISSLE and CANDY ADVENTURE ran so poorly, and because in hindsight both performances could have been predicted from a simple human analysis of the HTML Report, I recently spent several hours doing some Data Window research. As a result of that research I came up with two new Preset Filters: CANRUN and CANTRUN.

The CANRUN filter identifies horses that can run early, can run late, or can run evenly. In essence, it identifies horses that can find a way to run against today's field. These are horses that you might want to keep around as contenders. The CANRUN filter might be a good candidate for improving the effectiveness of many Positive Expectation UDMs.

The CANTRUN filter identifies horses that can't run early, can't run late, and can't run evenly. In essence, it identifies horses that can't run period. These are horses I'd typically want to avoid playing. The CANTRUN filter might be a very good candidate for use in Negative Expectation UDMs.

2004 Starter Database - All Horses
Let's take a look at some data. Below are several samples taken from the 2004 Starter Database.

2004 Starter Database
ALL Starters
Data Summary Win Place Show
Mutuel Totals 116415.10 114292.80 113476.90
Bet -151564.00 -151564.00 -151564.00
Gain -35148.90 -37271.20 -38087.10
Wins 9311 18539 27460
Plays 75782 75782 75782
PCT .1229 .2446 .3624
ROI 0.7681 0.7541 0.7487
Avg Mut 12.50 6.16 4.13

2004 Starter Database Filters:
Data Summary Win Place Show
Mutuel Totals 81769.20 80120.40 79132.00
Bet -97816.00 -97816.00 -97816.00
Gain -16046.80 -17695.60 -18684.00
Wins 7613 14708 21078
Plays 48908 48908 48908
PCT .1557 .3007 .4310
ROI 0.8359 0.8191 0.8090
Avg Mut 10.74 5.45 3.75

2004 Starter Database Filters:
Data Summary Win Place Show
Mutuel Totals 34645.90 34172.40 34344.90
Bet -53748.00 -53748.00 -53748.00
Gain -19102.10 -19575.60 -19403.10
Wins 1698 3831 6382
Plays 26874 26874 26874
PCT .0632 .1426 .2375
ROI 0.6446 0.6358 0.6390
Avg Mut 20.40 8.92 5.38

The above three samples speak for themselves. The CANRUN filter improves flat bet win roi by almost 8 percent. The CANTRUN filter reduces win percent and flat bet win roi significantly. Approximately 1 horse in 3 fits the CANTRUN profile. Approximately 2 horses in 3 fit the CANRUN profile.

Post Time Favorites
One of the most talked about subjects in thoroughbred handicapping is that of the post time favorite. Many sophisticated players understand the importance of identifying and separating vulnerable post time favorites from legitimate post time favorites. The belief here is that the player is far better off focusing on and playing those races where the favorite is identifiably beatable than simply playing all races. The next logical question becomes: What happens if we restrict our analysis of the CANRUN / CANTRUN filters to post time favorites only?

2004 Starter Database
All Post Time Favorites
Data Summary Win Place Show
Mutuel Totals 16186.70 17145.80 17172.60
Bet -19326.00 -19326.00 -19326.00
Gain -3139.30 -2180.20 -2153.40
Wins 3377 5447 6633
Plays 9663 9663 9663
PCT .3495 .5637 .6864
ROI 0.8376 0.8872 0.8886
Avg Mut 4.79 3.15 2.59

2004 Starter Database
Post Time Favorites
Data Summary Win Place Show
Mutuel Totals 14662.90 15389.10 15367.90
Bet -17106.00 -17106.00 -17106.00
Gain -2443.10 -1716.90 -1738.10
Wins 3099 4946 5987
Plays 8553 8553 8553
PCT .3623 .5783 .7000
ROI 0.8572 0.8996 0.8984
Avg Mut 4.73 3.11 2.57

2004 Starter Database
Post Time Favorites
Data Summary Win Place Show
Mutuel Totals 1523.80 1756.70 1804.70
Bet -2220.00 -2220.00 -2220.00
Gain -696.20 -463.30 -415.30
Wins 278 501 646
Plays 1110 1110 1110
PCT .2505 .4514 .5820
ROI 0.6864 0.7913 0.8129
Avg Mut 5.48 3.51 2.79

The above three samples are limited to post time favorites only. The CANRUN filter "validates" approximately 8 out of 9 post time favorites as being legitimate. It boosts the win percent and win roi of post time favorites by about 2 percent.

The CANTRUN filter, on the other hand, is a bit more drastic. It identifies approximately 1 of 9 post time favorites as being vulnerable. The win percent and flat bet win roi for post time favorites identified by the CANTRUN filter is significantly lower than the population of post time favorites as a whole. Think for a second what this means and how significant this can be. In races where the post time favorite fits the CANRUN profile, the post time favorite will be in the exacta more than 57 percent of the time. In such races your chance of cashing an exacta combination that doesn't involve the post time favorite is reduced. Compare that to races where the post time favorite fits the CANTRUN profile. In these races, the post time favorite will only be in the exacta 45 percent of the time. This means that your chances of cashing an exacta combination not involving the post time favorite suddenly increases. This is exactly the situation you should be looking for as a player.

Quite obviously the CANRUN / CANTRUN filters can be used in a UDM Definition any number of different ways by the sophisticated player. Give them a try. I'm guessing you just might like what happens once you do.

Until Next Time,

Happy Hunting,


Copyright 2005 JCapper Software