Finding an Edge
Author: Jeff Platt
Date Last Modified: 10/ 01/ 2007
INTRODUCTION
I wanted
to do a write up on finding an edge. Over the past several weeks I've had a number
of lengthy phone conversations with new users taking the Demo for the first
time. Many of them have tried other computer programs in the past only to be
disappointed. Some were disappointed with JCapper2007 too. Not because
JCapper2007 couldn't produce for them. Clearly it can - if they were only
willing to give it a serious workout.
If
feedback from users abandoning the demo is accurate, the number one reason I
have heard over and over is that the program is just far more complicated than
what they expected it to be. To quote one user who gave up within the first 30
minutes: "I'm looking for a program to help me handicap. I'm not
interested in learning how to write computer language."
New users
are amazed to discover that performing Data Window research and determining
first hand which (if any) from among the hundreds of available factors (and
unlimited combinations of factors) that might lead to profitable play takes a
real time commitment on their part. They are also amazed to discover that taking
information discovered during Data Window research and modeling that
information into UDMs and testing those UDMs against a development database and
then verifying the performance of those UDMs against a validation database
before going forward with live play... wait a minute… that's not handicapping… Noooo… That sounds like work too.
Hey. I
never said learning to win was easy. The simple truth is getting to where I'm
at now was a decades long journey. Trying to impart my own knowledge and
experiences gained on that journey to those just starting out is no easy task.
But I'm
going to try anyway. You see I believe winning is a skill that can be learned.
And if it can be learned then it can be taught.
A lot of
beginning players are naive. They’ve heard and read some really good things
about the original JCapper. Some of them think of the JRating as a magic number
where all they have to do is bet the top ranked horse and rake in the dollars.
In a sense that was true. Before I let the first person other than myself see a
copy of JCapper the top ranked JRating horse nationwide was producing a flat
bet win roi of more than .97 to $1.00. Right now it looks like things have
stabilized to where the top ranked JRating horse is returning a flat bet win
roi of about .928 to $1.00... That’s just under break even if you factor in a 7
percent rebate. Imagine that.
When I
decided on the factor mix for JPR that was going into JCapper2007 I wanted to
avoid tempting someone with a large bankroll into just blindly hammering JPR
horses at the windows like had previously been the case with JRating horses.
The JPR algorithm I decided to use has historically produced a flat bet win roi
of about .91 to $1.00. I realize it's very early in the game. As I write this
JPR has only been out for about 10 weeks now but those numbers seem to be
holding up pretty solidly.
What a
losing player seeking to become a winning player needs to realize is that
winning isn't about finding some magic number. Yes. Good numbers help. They are
easier to work with than bad numbers. But winning
is really about discipline and hard work. It's about finding an edge - and
then carrying that edge into the long run.
I'll save
the subject of carrying an edge into the long run for another time. This
article is about using JCapper2007 to find an edge of your own.
THE NATURE OF THE GAME
Horse
race betting is pari-mutuel in nature. That means that the more money bet on an
individual horse the lower the odds. It has been said that the betting public
makes a pretty good opponent. No, I’m not talking about each individual bettor.
I’m talking about the collective intelligence of all bettors everywhere. As
each race is bet, tens of thousands of bettors each weigh and apply the dozen
or so (in some cases more) handicapping factors they think will shape the
outcome of the race at hand. Collectively, they reach, by my rough estimate,
somewhere between one and two million decision points as they handicap each
race. As long as there has been modern day thoroughbred racing, post time
favorites have won approximately 33 percent of all races. However, in recent
years, because average field size has been shrinking, the win percentage of
post time favorites has risen. My calendar year 2006 database shows that post
time favorites (with no attempt to break ties for post time favoritism) won
almost 35 percent of all races. 81.47 percent of all races were won by one of
the first four choices in the betting. That should tell you quite clearly that
quite a bit of collective intelligence exists in the odds.
All
right. So how does a beginning player take advantage of that?
UNDERSTANDING THE GAME
One of
the things that I was slow to realize is that one of the keys to becoming a winning
player is getting an understanding of what drives the betting in the first
place. For various reasons, the overwhelming majority of bettors consistently
factor Speed Figures into their decision making process at some point. Because
of this, most of the time, horses with high speed figures relative to the rest
of the field tend to attract most of the money in the betting. In recent years,
for various reasons, bettors have become very much aware of trainer records.
Because of this, whenever a big name trainer enters a horse into a race, that
horse is also likely to attract its share of the money in the betting. Same
thing can be said about jockeys. Quite often, the betting public overreacts to
a rider considered to be a better rider than those he (or she) is competing
against by betting a disproportionate amount of money on that rider’s horse.
What about class? Again, the betting public is very much aware that a horse
dropping in class today is likely to run an improved performance compared to
its most recent start. So class droppers tend to get bet. Same thing can be
said about finish position. Horses that finished in the money (first, second,
or third) in their most recent starts tend to attract more money than horses
that finished up the track.
It should
be obvious at this point. The key to becoming a winning player lies in
understanding how the public bets and acting accordingly. The winning player
has to know how to avoid horses that are over bet and identify and play horses
that are under bet.
I’m going
to introduce three maxims:
1.
Modern thoroughbred pari-mutuel pools are efficient by nature.
2. Horses with obvious positives in their past performance
records will almost always be over bet.
3. Horses with hidden positives in their past performance
records will sometimes be under bet.
The trick
to winning then is really being able to identify horses with hidden positives
in their past performance records and betting on them yourself only when you
know that the public is under betting them.
HIDDEN POSITIVES
In this
next section I’m going to cover some common hidden positives. These are things
that you can find in a horse’s past performance record that are frequently
overlooked by the betting public. As such, these areas can make really good
starting points for UDMs and UPR for the savvy player.
Note: All data in the charts
presented below are from my own calendar year 2006 database which was compiled
using $1.00 Bris Single Format DRF data files. TSN file users will see similar
but slightly different results.
EARLY SPEED
I first
started using a computer to perform my own research and analyze thoroughbred
racing data back in 1985. Before that I did calculations manually and kept hand
written records in notebooks. One of the very first things I noticed about the
winning percentages of different factors was that final time was more
predictive than early speed. So it seemed perfectly logical to me as a bettor
that I should give final time based speed factors more weight in my
handicapping than early speed. What I didn’t realize at the time was the way
the public bets the races. The betting public always has been very much aware
that final time produces higher win percentages than early speed or early pace.
As long as I’ve been following racing the public has always had a tendency to
overemphasize final time and late speed when they bet the majority of the
races. At the same they have always had a counter tendency: they consistently
under bet early speed. That same general betting pattern is still prevalent
today. This discovery was one of many that lead me in the right direction – and
ultimately put me on the path to becoming a winning player.
JCapper
has a number of factors that reflect a horse’s early speed ability. CPace,
AVGE1, BestE2, PctE, PaceFig, TurnTime, V1, and several others all can be used
in UDMs to identify (and exclude) horses with early speed. Here in this
document I’m only going to cover CPace, BestE2, and 2FPace Fig (Last Start.)
CPace
By: CPace Rank
(2006)
Rank Gain
Bet Roi Wins Plays Pct
Impact
1 -3887.30 50402.00 0.9229
5442 25201 .2159
1.7437
2 -8310.90 50298.00 0.8348
4250 25149 .1690
1.3646
3 -8864.10 50344.00 0.8239
3720 25172 .1478
1.1933
4 -10202.10 50298.00
0.7972 3166 25149
.1259 1.0166
5 -13914.90 49820.00
0.7207 2622 24910
.1053 0.8500
6 -13153.70 46756.00
0.7187 2222 23378
.0950 0.7675
7 -13780.10 39054.00
0.6472 1605 19527
.0822 0.6637
8 -9921.90 28840.00 0.6560
1029 14420 .0714
0.5762
9 -7619.60 19580.00 0.6108
638 9790 .0652
0.5262
10 -4685.20 12616.00 0.6286
370 6308 .0587
0.4736
11 -2751.90
6834.00 0.5973 173
3417 .0506 0.4088
12 -1681.80
3394.00 0.5045 66
1697 .0389 0.3141
13 -113.20
484.00 0.7661 12
242 .0496 0.4004
14 193.00
200.00 1.9650 6
100 .0600 0.4845
15 -2.00
2.00 0.0000 0
1 .0000 0.0000
16 -2.00
2.00 0.0000 0
1 .0000 0.0000
17 -2.00
2.00 0.0000 0
1 .0000 0.0000
18 -2.00
2.00 0.0000 0
1 .0000 0.0000
19+ -4.00
4.00 0.0000 0
2 .0000 0.0000
BestE2
By: Best E2
Rank
Rank Gain Bet
Roi Wins Plays Pct
Impact
1 -7322.80 61148.00 0.8802
6168 30574 .2017
1.6330
2 -9241.70 53152.00 0.8261
4291 26576 .1615
1.3070
3 -9237.60 50980.00 0.8188
3587 25490 .1407
1.1391
4 -11326.30 50100.00
0.7739 2990 25050
.1194 0.9662
5 -13677.50 48408.00
0.7175 2538 24204
.1049 0.8488
6 -12422.60 44136.00
0.7185 2032 22068
.0921 0.7453
7 -10254.80 35834.00
0.7138 1456 17917
.0813 0.6578
8 -9352.40 25486.00 0.6330
893 12743 .0701
0.5673
9 -5748.60 17136.00 0.6645
557 8568 .0650
0.5262
10 -3943.30 10344.00 0.6188
279 5172 .0539
0.4367
11 -1538.10
4918.00 0.6873 127
2459 .0516 0.4181
12 -674.70
2328.00 0.7102 63
1164 .0541 0.4381
13 -91.90
490.00 0.8124 9
245 .0367 0.2974
14 41.00
214.00 1.1916 7
107 .0654 0.5296
15 0.00
0.00
0.0000 0 0 .0000
0.0000
16 -2.00
2.00 0.0000 0
1 .0000 0.0000
17 -4.00
4.00 0.0000 0
2 .0000 0.0000
18 0.00
0.00
0.0000 0 0 .0000
0.0000
19+ -4.00
4.00 0.0000 0
2 .0000 0.0000
2F Pace Fig (Last Start)
By: 73 rankForPaceFig_2F_InLast
Rank Gain Bet
Roi Wins Plays Pct
Impact
1 -7205.30 56844.00 0.8732
5324 28422 .1873
1.5163
2 -10410.70 52914.00
0.8033 4096 26457
.1548 1.2532
3 -11074.70 51246.00
0.7839 3534 25623
.1379 1.1164
4 -11756.90 50578.00
0.7675 3178 25289
.1257 1.0172
5 -14332.70 48990.00
0.7074 2664 24495
.1088 0.8803
6 -11134.90 44726.00
0.7510 2293 22363
.1025 0.8300
7 -10875.10 36390.00
0.7012 1641 18195
.0902 0.7301
8 -7821.70 26544.00 0.7053
1060 13272 .0799
0.6465
9 -4975.50 17432.00 0.7146
638 8716 .0732
0.5925
10 -3029.30 10772.00 0.7188
345 5386 .0641
0.5185
11 -1570.90
5152.00 0.6951 149
2576 .0578 0.4682
12 -432.00
2354.00 0.8165 59
1177 .0501 0.4058
13 30.20
510.00 1.0592 14
255 .0549 0.4444
14 -199.80
220.00 0.0918 2
110 .0182 0.1472
15 -2.00
2.00 0.0000 0
1 .0000 0.0000
16 -4.00
4.00 0.0000 0
2 .0000 0.0000
17 0.00
0.00
0.0000 0 0 .0000
0.0000
18 -2.00
2.00 0.0000 0
1 .0000 0.0000
19+ -4.00
4.00 0.0000 0
2 .0000 0.0000
I want
you to take note of the win rates and flat win bet roi of the top ranked horses
for all three of these factors. Then scroll down below and compare them to the
win rates and flat bet win roi of the top ranked horses for their corresponding
final time based speed figure or late pace factor counterparts. Compare CPace
to CFA. Compare BestE2 to Bris Speed Figure (Best of Last 10.) Compare 2F Pace
Call (Last Start) to both Bris Speed Figure (Last Start) and Late Pace (Last
Line.)
Statistically,
final time based speed figures produce a higher win rate. But as long as I’ve
been compiling statistics on racing early speed based factors have always
produced a higher flat bet win roi than their final time based counterparts.
SPEED FIGURES
The
charts below show actual Data Window output for several different final time
based speed figure categories. There are a few things I want to point out.
First, the betting public overemphasizes the last line of a horse’s past
performance record. This can be evidenced by comparing the flat bet win roi of
top ranked Bris Speed Figure (Last Race) horses against flat bet win roi of
Bris Speed Figure (Best of Last 10) horses. In almost any pace line selection
method I have ever tested, giving the last line a lower weight than the best
line produces a higher flat bet win roi for the model being developed.
A second
thing I want to point out has to do with the value of proprietary information.
Take a look at the flat bet win roi of the top ranked JCapper Weighted Figure
and compare it to the flat bet win roi of the Bris Speed Figure. Bris makes a
more accurate (predictive) speed figure than the JCapper Weighted Figure. But
the JCapper Weighted Figure factors in a horse’s early speed ability to produce
a higher roi.
Finally,
take a look at CFA. CFA is the result of a JCapper algorithm where
representative pace lines (speed figures – both JCapper Weighted and Bris) are
selected from a horse’s past performance record and then crunched into a final
number. CFA is both more predictive and produces a higher flat bet win roi than
Bris Speed Figures do.
CFA
Rank Gain Bet
Roi Wins Plays Pct
Impact
1 -5663.70 50984.00 0.8889
6710 25492 .2632
2.1255
2 -6563.60 51300.00 0.8721
4971 25650 .1938
1.5649
3 -10556.80 51908.00
0.7966 3772 25954
.1453 1.1736
4 -12442.50 51888.00
0.7602 2951 25944
.1137 0.9185
5 -13228.30 51592.00
0.7436 2485 25796
.0963 0.7779
6 -13875.30 47038.00
0.7050 1782 23519
.0758 0.6118
7 -13860.50 38756.00
0.6424 1149 19378
.0593 0.4788
8 -9049.10 27586.00 0.6720
712 13793 .0516
0.4168
9 -5613.60 18158.00 0.6908
432 9079 .0476
0.3842
10 -3861.30 10966.00 0.6479
227 5483 .0414
0.3343
11 -2753.70
6044.00 0.5444 101
3022 .0334
0.2699
12 -1350.00
2280.00 0.4079 22
1140 .0193 0.1558
13 114.30
320.00 1.3572 5
160 .0312 0.2523
14 -92.20
100.00 0.0780 1
50 .0200 0.1615
15
-2.00 2.00 0.0000
0 1 .0000
0.0000
16 96.60
6.00 17.1000 1
3 .3333 2.6917
17 0.00
0.00
0.0000 0 0 .0000
0.0000
18 0.00
0.00
0.0000 0 0 .0000
0.0000
19 -4.00
4.00 0.0000 0
2 .0000 0.0000
WEIGHTED FIGURE
By: Weighted Figure Rank
Rank Gain Bet
Roi Wins Plays Pct
Impact
1 -6161.80 50638.00 0.8783
5768 25319 .2278
1.8441
2 -9964.50 50822.00 0.8039
4267 25411 .1679
1.3592
3 -9371.90 50794.00 0.8155
3731 25397 .1469
1.1892
4 -12091.00 51486.00
0.7652 3071 25743
.1193 0.9656
5 -11761.70 50260.00
0.7660 2725 25130
.1084 0.8778
6 -12968.10 46570.00
0.7215 2054 23285
.0882 0.7140
7 -12366.70 37952.00
0.6741 1394 18976
.0735 0.5946
8 -8925.00 27766.00 0.6786
919 13883 .0662
0.5358
9 -4856.60 18384.00 0.7358
551 9192 .0599
0.4852
10 -2795.00 11292.00 0.7525
313 5646 .0554
0.4487
11 -2310.00
5402.00 0.5724 129
2701 .0478 0.3866
12 -1107.10
2514.00 0.5596 51
1257 .0406 0.3284
13 -140.50
544.00 0.7417 18
272 .0662 0.5357
14 31.80
240.00 1.1325 5
120 .0417 0.3373
15 -6.00
6.00 0.0000 0
3 .0000 0.0000
16 0.80
6.00 1.1333 1
3 .3333 2.6982
17 -2.00
2.00 0.0000 0
1 .0000 0.0000
18 -2.00
2.00 0.0000 0
1 .0000 0.0000
19 -4.00
4.00 0.0000 0
2 .0000 0.0000
Bris Speed Figure (Last Line)
By: 62 rankForLastRaceBrisFig
Rank Gain Bet
Roi Wins Plays Pct
Impact
1 -8624.70 56186.00 0.8465
7307 28093 .2601
2.1054
2 -9399.90 53124.00 0.8231
4731 26562 .1781
1.4418
3 -10190.60 51266.00
0.8012 3540 25633
.1381 1.1179
4 -12003.20 50714.00
0.7633 2792 25357
.1101 0.8913
5 -10991.10 49048.00
0.7759 2238 24524
.0913 0.7387
6 -11389.60 44708.00
0.7452 1719 22354
.0769 0.6225
7 -11160.00 36328.00
0.6928 1160 18164
.0639 0.5169
8 -9144.10 26602.00 0.6563
688 13301 .0517
0.4187
9 -4975.10 17592.00 0.7172
438 8796 .0498
0.4031
10 -4056.10 10764.00 0.6232
228 5382 .0424
0.3429
11 -1464.90
5134.00 0.7147 108
2567 .0421 0.3406
12 -1231.60
2418.00 0.4907 33
1209 .0273 0.2209
13 64.10
540.00 1.1187 14
270 .0519 0.4197
14 -218.50
244.00 0.1045 1
122 .0082 0.0663
15 -6.00
6.00 0.0000 0
3 .0000 0.0000
16 -2.00
2.00 0.0000 0
1 .0000 0.0000
17 -4.00
4.00 0.0000 0
2 .0000 0.0000
18 0.00
0.00
0.0000 0 0 .0000
0.0000
19 -4.00
4.00 0.0000 0
2 .0000 0.0000
Bris Speed Figure (Best of Last
10)
By: Bris Speed Figure Rank (Best of
Last 10)
Rank Gain Bet
Roi Wins Plays Pct
Impact
1 -8108.60 56982.00 0.8577
6917 28491 .2428
1.9652
2 -9410.80 54264.00 0.8266
4850 27132 .1788 1.4470
3 -9244.90 52022.00 0.8223
3889 26011 .1495
1.2103
4 -10056.20 51372.00
0.8042 2988 25686
.1163 0.9416
5 -10817.10 48850.00
0.7786 2329 24425
.0954 0.7719
6 -12728.40 44176.00
0.7119 1606 22088
.0727 0.5886
7 -11541.80 35444.00
0.6744 1079 17722
.0609 0.4928
8 -8968.20 25992.00 0.6550
648 12996 .0499
0.4036
9 -7308.20 17202.00 0.5752
356 8601
.0414 0.3350
10 -3049.30 10314.00 0.7044
210 5157 .0407
0.3296
11 -2162.30
5018.00 0.5691 86
2509 .0343 0.2775
12 -1019.20
2310.00 0.5588 27
1155 .0234 0.1892
13 -209.30
508.00 0.5880 9
254 .0354 0.2868
14 -173.00
226.00 0.2345 3
113 .0265 0.2149
15 0.00
0.00
0.0000 0 0 .0000
0.0000
16 0.00
0.00 0.0000
0 0 .0000
0.0000
17 0.00
0.00
0.0000 0 0 .0000
0.0000
18 0.00
0.00
0.0000 0 0 .0000
0.0000
19 -4.00
4.00 0.0000 0
2 .0000
0.0000
FORM
Current
Form is one key area where I have discovered the betting public is lacking in
when it comes to how they bet the races. JCapper has a number of form based
factors that you can use in your models. The charts below all very clearly
demonstrate that horses rated high in numerical form based factors frequently
have hidden positives in their past performance records that the public
consistently underestimates.
AFR
By: AFR Rank
Rank Gain Bet
Roi Wins Plays Pct
Impact
1 -5110.50 50654.00 0.8991
4928 25327 .1946
1.5712
2 -8723.00 50594.00 0.8276
4205 25297 .1662
1.3423
3 -8722.70 50518.00 0.8273
3742 25259 .1481 1.1963
4 -10393.10 50246.00
0.7932 3374 25123
.1343 1.0845
5 -12176.80 49202.00
0.7525 2977 24601
.1210 0.9772
6 -12999.70 46048.00
0.7177 2412 23024
.1048 0.8459
7 -13096.10 38730.00
0.6619 1608 19365
.0830 0.6705
8 -10116.70 28730.00
0.6479 1011 14365
.0704 0.5683
9 -6951.20 19510.00 0.6437
567 9755 .0581
0.4693
10 -5400.70 12782.00 0.5775
296 6391 .0463
0.3740
11 -2786.40
6970.00 0.6002 138
3485 .0396 0.3198
12 -1615.70
3748.00 0.5689 49
1874 .0261 0.2111
13 -356.50
660.00 0.4598 9
330 .0273 0.2202
14 -161.50
412.00 0.6080 4
206 .0194 0.1568
15 -56.00
56.00 0.0000 0
28 .0000 0.0000
16 -31.10
64.00 0.5141 1
32 .0312 0.2523
17 -2.00
2.00 0.0000 0
1 .0000 0.0000
18 -2.00
2.00 0.0000 0
1 .0000 0.0000
19 -4.00
4.00 0.0000 0
2 .0000 0.0000
WOBRILL
By: Workout Brilliance Rank (2006
1st time starters removed)
Rank Gain Bet
Roi Wins Plays Pct
Impact
1 -5593.80 56182.00 0.9004
4814 28091 .1714
1.3548
2 -10511.50 59542.00
0.8235 4374 29771
.1469 1.1615
3 -15458.50 59846.00
0.7417 3935 29923
.1315 1.0396
4 -14442.20 57026.00
0.7467 3475 28513
.1219 0.9635
5 -13060.60 48320.00
0.7297 2812 24160
.1164 0.9201
6 -10400.50 36826.00
0.7176 1926 18413
.1046 0.8269
7 -7379.00 25138.00 0.7065
1160 12569 .0923
0.7296
8 -3994.10 16048.00 0.7511
689 8024 .0859
0.6788
9 -3323.80
9454.00 0.6484 369
4727 .0781 0.6171
10 -1659.00
4966.00 0.6659 196
2483 .0789 0.6240
11 -835.10
2376.00 0.6485 65
1188 .0547 0.4325
12 -536.20
924.00 0.4197 19
462 .0411 0.3251
13 -73.50
156.00 0.5288 3
78 .0385 0.3041
14 -66.00
66.00 0.0000 0
33 .0000 0.0000
15 -2.00
2.00 0.0000 0
1 .0000 0.0000
16 -2.00
2.00 0.0000 0
1 .0000 0.0000
17 -6.00
6.00 0.0000 0
3 .0000 0.0000
18 0.00
0.00
0.0000 0 0 .0000
0.0000
19 -2.00
2.00 0.0000 0
1 .0000 0.0000
By: Workout Brilliance Rank (2006 1st time starters only)
Rank Gain Bet
Roi Wins Plays Pct
Impact
1 -574.80
4232.00 0.8642 305
2116 .1441 1.7273
2 -678.30
3986.00 0.8298 223
1993 .1119 1.3409
3 -1162.60
3968.00 0.7070 179
1984 .0902 1.0812
4 -1060.60
3598.00 0.7052 137
1799 .0762 0.9126
5 -1089.70
3230.00 0.6626 99
1615 .0613 0.7346
6 -809.10
2926.00 0.7235 93
1463 .0636 0.7618
7 -528.20
2184.00 0.7582 58
1092 .0531 0.6365
8 -721.10
1516.00 0.5243 27
758 .0356 0.4269
9 -512.00
1036.00 0.5058 20
518 .0386 0.4627
10 -261.40
610.00 0.5715 11
305 .0361 0.4322
11 -282.90
346.00 0.1824 3
173 .0173 0.2078
12 37.80
142.00 1.2662 4
71 .0563 0.6751
13 193.40
22.00 9.7909 1
11 .0909 1.0894
14+ -6.00
6.00 0.0000 0
3 .0000 0.0000
FORM RATING
By: Form Rating Rank
Rank Gain
Bet Roi Wins Plays Pct
Impact
1 -6030.40 54286.00 0.8889
4579 27143 .1687
1.3656
2 -8078.30 50146.00 0.8389
3794 25073 .1513
1.2249
3 -12787.30 49662.00
0.7425 3246 24831
.1307 1.0582
4 -9664.80 49618.00 0.8052
3249 24809 .1310
1.0601
5 -11936.90 48942.00
0.7561 2935 24471
.1199 0.9709
6 -12106.10 45606.00
0.7346 2575 22803
.1129 0.9141
7 -12994.60 37842.00
0.6566 1836 18921
.0970 0.7855
8 -8843.50 28538.00 0.6901
1278 14269 .0896
0.7250
9 -5363.30 19646.00 0.7270
799 9823 .0813
0.6584
10 -3723.00 11902.00 0.6872
435 5951 .0731
0.5917
11 -2642.00
5564.00 0.5252 176
2782 .0633 0.5121
12 -722.20
2416.00 0.7011 81
1208 .0671 0.5428
13 76.50
420.00 1.1821 11 210
.0524 0.4240
14 20.60
90.00 1.2289 3
45 .0667 0.5396
15 0.00
0.00
0.0000 0 0 .0000
0.0000
16 0.00
0.00
0.0000 0 0 .0000
0.0000
17
0.00 0.00 0.0000
0 0 .0000
0.0000
18 -4.00
4.00 0.0000 0
2 .0000 0.0000
19 -2.00
2.00 0.0000 0
1 .0000 0.0000
BASIC FITNESS
By: BasicFitness
>=Min <Max Gain
Bet Roi Wins Plays Pct
Impact
-999.00 0.00
0.00
0.00
0.0000 0 0 .0000
0.0000
0.00 5.00 -6678.40 18974.00
0.6480 952 9487
.1003 0.8123
5.00 10.00 -46064.50 173194.00 0.7340 10247
86597 .1183 0.9578
10.00 15.00 -6672.80 26000.00
0.7434 1447 13000
.1113 0.9010
15.00 20.00 -15104.10 62506.00 0.7584
3798 31253 .1215
0.9837
20.00 25.00 -13131.20 62646.00 0.7904
3887 31323 .1241
1.0045
25.00 30.00 -4382.00 35548.00
0.8767 2580 17774
.1452 1.1750
30.00 35.00 -2768.30 25816.00
0.8928 2086 12908
.1616 1.3081
REC ACTIVITY DAYS LAST START
By: Recent Activity-
Days Last Start (2006 with 1st time starters removed)
>=Min <Max Gain
Bet Roi Wins Plays Pct
Impact
-999 0
0.00 0.00 0.0000
0 0 .0000
0.0000
0 20 -34136.30 138052.00 0.7527
8389 69026 .1215
0.9608
20 40 -34356.50 154838.00 0.7781 10326
77419 .1334 1.0544
40 60 -5920.00 36464.00
0.8376 2540 18232
.1393 1.1013
60 80 -3216.20 12180.00
0.7359 756 6090
.1241 0.9814
80 100 -1155.90 6012.00
0.8077 346 3006
.1151 0.9099
100 120 -1022.00 3786.00
0.7301 193 1893
.1020 0.8060
120 140
-204.90 3320.00 0.9383
177 1660 .1066
0.8429
140 160
-997.40 2808.00 0.6448
136 1404 .0969
0.7658
160 180
-376.60 2712.00 0.8611
147 1356 .1084
0.8570
180
200 -859.00 2506.00
0.6572 116 1253
.0926 0.7319
200 220 -1177.10 2310.00
0.4904 100 1155
.0866 0.6845
220 240
-318.90 2054.00 0.8447
107 1027 .1042
0.8236
240 260
-512.60 1724.00 0.7027
92 862 .1067
0.8437
260 280
-511.70 1412.00 0.6376
78 706 .1105
0.8734
280 300
-422.00 1226.00 0.6558
67 613 .1093
0.8640
300 320
-339.30 980.00 0.6538
59 490 .1204
0.9519
320 340
-240.50 844.00 0.7150
43 422 .1019
0.8055
340 360
-161.20 564.00 0.7142
31 282 .1099
0.8690
360 999999 -1417.70 3090.00
0.5412 134 1545
.0867 0.6856
I have
found that using form based factors in my UDMs and UPR helps my models
emphasize horses with hidden positives. In turn this helps give my models a
boost in roi.
TRAINER INTENT
The
public is very much in tune with well known trainers when it comes to the way
they bet the races. As of this writing trainers such as Todd Pletcher, Richard
Dutrow, Steve Asmussen, and Robert Frankel are well known for their ability to
consistently win a very high percentage of races where they enter a starter.
But the public has a tendency to over simplify. They tend to bet the trainer
name not how the horse was handled leading up to the race. In other words, the
public doesn’t really have a firm grasp of trainer intent.
JCapper
has a number of factors that will allow you to get a very firm grip on trainer
intent in your UDMs. Some of the ones I like to use are the CLEAN/DIRTY and
XFORM/YFORM preset Filters, the MED preset filters, the EQU preset filters,
XthStartForTrainer, XthStartOffOfLayoff, XthStartOnLasix, XthStartForTheMeet,
the WOSINCE preset filter, Rec Activity Days Last Start, Rec Activity Days Last
Work, Claim Races, Claim Days, and BasicFitness. You can find a complete
explanation of these and other JCapper factors on the Supported Factors page at
http://www.JCapper.com.
CLEAN Preset Filter
By: CANTRUN/XFORM Attribute
CANTRUN Gain Bet
Roi Wins Plays Pct
Impact
0 -31894.00
188688.00 0.8310 13848
94344 .1468 1.1881
Clean Horse
1 -38616.90
153584.00 0.7486 9725
76792 .1266 1.0251
XFORM
2 -24290.40 62412.00
0.6108 1424 31206
.0456 0.3694 CANTRUN
CLAIM RACES
By: Claimed Races Back
Claim Races
Gain Bet Roi
Wins Plays Pct
Impact
0 -87892.70
367010.00 0.7605 21876 183505
.1192 0.9650 Not Claimed
1 -2610.40 15762.00 0.8344
1343 7881 .1704
1.3794 Claimed 1bk
2 -1973.00 12220.00 0.8385
1036 6110 .1696
1.3725 Claimed 2bk
3 -2325.20
9692.00 0.7601 742
4846 .1531 1.2394 Claimed 3bk
XthStartForTrainer
By: Xth Start For
Trainer (With 1st Time Starters Removed)
Start # Gain Bet
Roi Wins Plays Pct
Impact
1 -7182.60 36762.00 0.8046
2481 18381 .1350
1.0670
2 -12617.70 51840.00
0.7566 3284 25920
.1267 1.0016
3 -10953.60 41694.00
0.7373 2658 20847
.1275 1.0079
4 -8554.00 34640.00 0.7531
2274 17320 .1313
1.0379
5 -6001.90 28884.00 0.7922
1971 14442 .1365
1.0789
6
-4334.60
24340.00 0.8219 1619
12170 .1330 1.0517
7 -3311.80 21126.00 0.8432
1474 10563 .1395
1.1031
8 -3635.20 18290.00 0.8012
1250 9145 .1367
1.0806
9 -3538.40 16564.00 0.7864
1081 8282 .1305
1.0318
10 -3733.30 16064.00 0.7676
1039 8032 .1294
1.0226
11 -23482.70 86678.00
0.7291 4706 43339
.1086 0.8584
XthStartOffOfLayoff
By: Xth
Start Off Of Layoff (with 1st Time Starters Removed)
Starts Gain Bet
Roi Wins Plays Pct
Impact
0 -26415.80
102582.00 0.7425 6248
51291 .1218 0.9630
1 -16363.40 67360.00
0.7571 3893 33680
.1156 0.9138
2 -15021.40 51132.00
0.7062 3096 25566
.1211 0.9573
3 -8628.50 38936.00 0.7784
2590 19468 .1330
1.0517
4 -4631.40 29740.00 0.8443
2127 14870 .1430
1.1308
5 -3105.00 22956.00 0.8647
1627 11478 .1417
1.1206
6 -3764.60 17864.00 0.7893
1196 8932 .1339
1.0585
7 -2157.40 13976.00 0.8456
986 6988 .1411
1.1154
8 -2192.10 10858.00 0.7981
740 5429 .1363
1.0775
9 -1930.10
8578.00 0.7750 557
4289 .1299 1.0267
10 -1448.00
7080.00 0.7955 429
3540 .1212 0.9580
11 -1688.10
5820.00 0.7099 348
2910 .1196 0.9454
MEDICATION
By: Medication
Medication Gain Bet
Roi Wins Plays Pct
Impact
0 -8146.50 20228.00 0.5973
684 10114 .0676
0.5474 None
1 -77743.80
341604.00 0.7724 21898 170802
.1282 1.0378 Lasix
2 -292.00
1124.00 0.7402 57
562 .1014 0.8210
3 -3378.00 14462.00 0.7664
957 7231 .1323
1.0713 Lasix & Bute
4 -5260.80 26504.00 0.8015 1354
13252 .1022 0.8271
1st Lasix
5 19.80
762.00 1.0260 47
381 .1234 0.9986
6 0.00
0.00
0.0000 0 0 .0000
0.0000
7 0.00
0.00
0.0000 0 0 .0000
0.0000
8 0.00
0.00
0.0000 0 0 .0000
0.0000
9 0.00
0.00
0.0000 0 0 .0000
0.0000 Unknown
EQUIPMENT CHANGES
By:
Equipment Change
Change Gain Bet
Roi Wins Plays Pct
Impact
0 -89431.00
381352.00 0.7655 23677 190676
.1242 1.0051 No Change
1 -3920.50 13750.00 0.7149
709 6875 .1031
0.8348 Blinkers On
2 -556.80 6050.00
0.9080 383 3025
.1266 1.0249 Blinkers Off
3 0.00
0.00
0.0000 0 0 .0000
0.0000
4 0.00
0.00
0.0000 0 0 .0000
0.0000
5 0.00
0.00 0.0000
0 0 .0000
0.0000
6 0.00
0.00
0.0000 0 0 .0000
0.0000
7 0.00
0.00
0.0000 0 0 .0000
0.0000
8 0.00
0.00
0.0000 0 0 .0000
0.0000
9 -893.00
3532.00 0.7472 228
1766 .1291 1.0451 Unknown
REC ACTIVITY DAYS LAST START
By: Recent Activity-
Days Last Start (2006 with 1st time starters removed)
>=Min <Max Gain
Bet Roi
Wins Plays Pct
Impact
-999 0
0.00 0.00 0.0000
0 0 .0000
0.0000
0 20 -34136.30 138052.00 0.7527
8389 69026 .1215
0.9608
20 40 -34356.50 154838.00 0.7781 10326
77419 .1334 1.0544
40 60 -5920.00 36464.00
0.8376 2540 18232
.1393 1.1013
60 80 -3216.20 12180.00
0.7359 756 6090
.1241 0.9814
80 100 -1155.90 6012.00
0.8077 346 3006
.1151 0.9099
100 120 -1022.00 3786.00
0.7301 193 1893
.1020 0.8060
120 140
-204.90 3320.00 0.9383
177 1660 .1066
0.8429
140 160
-997.40 2808.00 0.6448 136
1404 .0969 0.7658
160 180
-376.60 2712.00 0.8611
147 1356 .1084
0.8570
180 200
-859.00 2506.00 0.6572
116 1253 .0926
0.7319
200 220 -1177.10 2310.00
0.4904 100 1155
.0866 0.6845
220 240
-318.90 2054.00 0.8447
107 1027 .1042
0.8236
240 260
-512.60 1724.00 0.7027
92 862 .1067
0.8437
260 280
-511.70 1412.00 0.6376
78 706 .1105
0.8734
280 300
-422.00 1226.00 0.6558
67 613 .1093
0.8640
300 320
-339.30 980.00 0.6538
59 490 .1204
0.9519
320 340
-240.50 844.00 0.7150
43 422 .1019
0.8055
340 360
-161.20 564.00 0.7142
31 282 .1099
0.8690
360 999999 -1417.70 3090.00
0.5412 134 1545
.0867 0.6856
Using
trainer intent factors intelligently in your UDMs (and UPR) is quite often a
good way to get your models to select horses with hidden positives in their
records and will often point the way towards boosting the roi strength of your
overall results.
CONNECTIONS
JCapper
has a number of ratings that allow you to evaluate the effectiveness of a
horse’s human connections. I personally like to use the CXNRating (or CScore)
which represents the combined ability of both the rider and trainer.
CXN Rating
By: CXN Rating (2006)
>=Min <Max Gain
Bet Roi Wins Plays Pct
Impact
-999.00 0.00
0.00
0.00
0.0000 0 0 .0000
0.0000
0.00 5.00
0.00 0.00 0.0000
0 0 .0000
0.0000
5.00 10.00
0.00 0.00 0.0000
0 0 .0000
0.0000
10.00 15.00
0.00 0.00 0.0000
0 0 .0000
0.0000
15.00 20.00
0.00 0.00 0.0000
0 0 .0000
0.0000
20.00 25.00
0.00 0.00 0.0000
0 0
.0000
0.0000
25.00 30.00
0.00 0.00 0.0000
0 0 .0000
0.0000
30.00 35.00 -1448.90 3174.00
0.5435 40 1587
.0252 0.2040
35.00 40.00 -4081.30 11064.00
0.6311 191 5532
.0345 0.2795
40.00 45.00 -6006.00 17620.00
0.6591 405 8810
.0460 0.3721
45.00 50.00 -6863.70 26166.00
0.7377 759 13083
.0580 0.4696
50.00 55.00 -10257.00 34074.00 0.6990
1130 17037 .0663
0.5369
55.00 60.00 -10346.70 40990.00 0.7476
1636 20495 .0798
0.6462
60.00 65.00 -11897.20 44100.00 0.7302
2037 22050 .0924
0.7478
65.00 70.00 -10262.30 48434.00 0.7881
2781 24217 .1148
0.9296
70.00
75.00 -9966.20 46838.00
0.7872 3092 23419
.1320 1.0687
75.00 80.00 -8052.20 42522.00
0.8106 3370 21261
.1585 1.2831
80.00 85.00 -6151.40 34940.00
0.8239 3100 17470
.1774 1.4364
85.00 90.00 -5073.30 26950.00
0.8118 2821 13475
.2094 1.6946
90.00 999999.00 -4395.10
27812.00 0.8420 3635
13906 .2614 2.1159
By: CXN Rating Rank
Rank Gain Bet
Roi Wins Plays Pct
Impact
1 -8714.40 50528.00 0.8275
5855 25264 .2318
1.8760
2 -8335.00 49780.00 0.8326
4597 24890 .1847
1.4950
3 -9385.00 49630.00 0.8109
3731 24815 .1504
1.2171
4 -11057.60 49600.00
0.7771 3146 24800
.1269 1.0268
5 -11914.30 49056.00
0.7571 2556 24528
.1042 0.8435
6 -13002.10 46032.00
0.7175 1890 23016
.0821 0.6647
7 -11698.80 38412.00
0.6954 1353 19206 .0704
0.5702
8 -7099.80 28668.00 0.7523
883 14334 .0616
0.4986
9 -6162.60 19860.00 0.6897
491 9930 .0494
0.4002
10 -4459.60 12756.00 0.6504
310 6378 .0486
0.3934
11 -1344.80
6570.00 0.7953 131
3285 .0399 0.3228
12 -1513.40
3064.00 0.5061 43
1532 .0281 0.2272
13 -151.10
564.00 0.7321 6
282 .0213 0.1722
14 35.00
152.00 1.2303 4
76 .0526 0.4260
15 -2.00
2.00 0.0000 0
1 .0000 0.0000
16 -2.00
2.00 0.0000 0
1 .0000 0.0000
17 12.20
2.00 7.1000 1
1 1.0000
8.0947
18 -6.00
6.00 0.0000 0
3 .0000 0.0000
19 0.00
0.00
0.0000 0 0 .0000
0.0000
The above
charts clearly show that horses being handled by competent human connections
outperform (think hidden positives) horses handled by less competent human
connections. Quite often adding cutoffs for CXN Rating (CScore) will boost roi
performance of a model’s overall results.
AGE of HORSE
Horses
have shorter lifespans than humans. A horse is at age four is considered fully
developed. In my own head I equate a four year old horse to a human adult in
his or her own mid twenties. Just like human beings, horses lose speed and
stamina as they get older. The chart below shows all starters in my calendar
year 2006 database broken out by age of horse gap. Where the age gap is one or
more years (an older horse facing one or more younger horses today) both win
rate and flat bet win roi are decidedly lower. Conversely, horses facing off
against competition their own age or older today have higher (think hidden
positive) win rates and flat win bet roi than the set of all starters taken as
a whole. This particular hidden positive may be a subtle thing, but my own
experience has shown over and over that bringing Age of Horse into a UDM can
often boost performance of the model.
By: Age Of Horse Gap
>=Min <Max Gain
Bet Roi Wins Plays Pct
Impact
-999 -10
0.00 0.00 0.0000
0 0 .0000
0.0000
-10 -9
0.00 0.00 0.0000
0 0 .0000
0.0000
-9 -8
-30.20 58.00 0.4793
3 29 .1034
0.8374
-8 -7
21.80 256.00 1.0852
15 128 .1172
0.9486
-7 -6
-46.10 770.00 0.9401
53 385 .1377
1.1143
-6 -5
-672.10 2002.00 0.6643
116 1001 .1159
0.9380
-5 -4 -2189.80 6416.00
0.6587 355 3208
.1107 0.8958
-4 -3 -8981.70 39724.00
0.7739 2492 19862
.1255 1.0156
-3 -2 -22322.30 98868.00 0.7742
6302 49434 .1275
1.0319
-2 -1 -17887.90 85940.00 0.7919
5554 42970 .1293
1.0463
-1 0 -17521.30 72042.00 0.7568
4336 36021 .1204
0.9744
0 1 -19526.40 82846.00 0.7643
4932 41423 .1191
0.9638
1 2 -3565.30 11216.00
0.6821 604 5608
.1077 0.8718
2 3 -1444.40 3298.00
0.5620 178 1649
.1079 0.8738
3 4
-404.90 888.00 0.5440
44 444 .0991
0.8022
4 5
-190.70 278.00 0.3140
8 139 .0576
0.4659
5 6 -21.00
58.00 0.6379 4
29 .1379 1.1165
6 7
-16.00 16.00 0.0000
0 8 .0000
0.0000
7 8
-3.00 8.00 0.6250
1 4 .2500
2.0237
8 999999
0.00 0.00 0.0000
0 0 .0000
0.0000
By:
Age Of Horse Years
Age Yrs Gain Bet
Roi Wins Plays Pct
Impact
0 0.00
0.00
0.0000 0 0 .0000
0.0000
1 0.00
0.00
0.0000 0 0 .0000
0.0000
2 -10134.50 42086.00
0.7592 2517 21043
.1196 0.9682
3 -31169.80
138752.00 0.7754 8687
69376 .1252 1.0136
4 -24326.30
103422.00 0.7648
6581 51711 .1273
1.0302
5 -13302.50 64588.00
0.7940 3988 32294
.1235 0.9996
6 -9126.40 30038.00 0.6962
1746 15019 .1163
0.9410
7 -3246.60 14730.00 0.7796
850 7365 .1154
0.9342
8 -2081.30
6852.00 0.6962 403
3426 .1176 0.9522
9 -614.20
2790.00 0.7799 168
1395 .1204 0.9748
10 -574.10
1008.00 0.4305 38
504 .0754 0.6103
11 -126.60
198.00 0.3606 6
99 .0606 0.4906
12 -78.60
190.00 0.5863 11
95 .1158 0.9373
13 -20.40
30.00 0.3200 2
15 .1333 1.0793
14 0.00
0.00
0.0000 0 0 .0000
0.0000
15 0.00
0.00
0.0000 0 0 .0000
0.0000
16 0.00
0.00
0.0000 0 0 .0000
0.0000
17 0.00
0.00
0.0000 0 0 .0000
0.0000
18 0.00
0.00
0.0000 0 0 .0000
0.0000
My
coverage of hidden positives in this document is limited to only a handful of
factors. It should be obvious to you that hidden positives exist to some degree
of usefulness within the Data Window output numbers for almost every factor. It’s up to you to do your
own research, discover those hidden positives, and bring them into your own
models.
OBVIOUS POSITIVES
In this next
section I’m going to cover a few obvious positives. These are attributes found
in a horse’s past performance record that the betting public frequently
overreacts to. As such, when horses with these (and other) overly obvious
positives in their past performance records are excluded from UDMs, or
downgraded in UPR, the overall profitability of the model can be improved.
THE MORNING LINE
The
morning line odds are set by a track employee known as the morning line
oddsmaker. The morning line is not necessarily designed as an attempt to pick
the winner of each race. Instead, the job of the morning line oddsmaker is
often twofold: First, the morning line can be a prediction of how the public
will bet the race. And second, it can be an attempt by the morning line
oddsmaker to maximize betting on the race for the track. Nonetheless, many
morning line oddsmakers are bright talented handicappers. At a lot of tracks
the crowd reacts to the morning line in such a way as to cause it to drive the
betting. My calendar year 2006 database shows that there is quite a bit of
inherent predictability in the morning line odds. Nationally, the morning line
favorite won approximately 31 percent of all races.
By: Morning Line Rank
Rank Gain Bet
Roi Wins Plays Pct
Impact
1 -10705.00 51096.00 0.7905
7963 25548 .3117
2.5230
2 -9762.10 51474.00 0.8103
5140 25737 .1997
1.6166
3 -10119.20 54976.00
0.8159 4082 27488
.1485 1.2021
4 -13038.50 54920.00
0.7626 2933 27460
.1068 0.8646
5 -10115.10 50654.00
0.8003 2045 25327
.0807 0.6536
6 -11783.40 44698.00
0.7364 1247 22349
.0558 0.4517
7 -8563.30 37392.00 0.7710
828 18696 .0443
0.3585
8 -7812.10 26848.00 0.7090
435 13424 .0324
0.2623
9 -5815.80 16902.00 0.6559
203 8451 .0240
0.1944
10 -3507.60
9406.00 0.6271 84
4703 .0179 0.1446
11 -2728.60
4278.00 0.3622 25
2139 .0117 0.0946
12 -468.60
1658.00 0.7174 12
829 .0145 0.1172
13 -284.00
284.00 0.0000 0
142 .0000 0.0000
14
-86.00 86.00 0.0000
0 43 .0000
0.0000
15 -12.00
12.00 0.0000 0
6 .0000 0.0000
16 0.00
0.00
0.0000 0 0 .0000
0.0000
17 0.00
0.00
0.0000 0 0 .0000
0.0000
18 0.00
0.00
0.0000 0 0 .0000
0.0000
19 0.00
0.00
0.0000 0 0 .0000
0.0000
Take a
close look at the above chart. I want you to take specific notice of something.
In 2006 morning line favorites had a lower flat bet win roi than both the
second and third morning line choices. That’s an indication that morning line
favorites contain a lot of obvious positives in their past performance records
- enough to attract a disproportionate amount of money from the betting public
relative to their true chances of winning. At the same time, statistically,
this phenomenon creates opportunity elsewhere in the pools. I will demonstrate
later how you can use this to your advantage and actually strengthen the flat
win bet roi of your UDMs.
FORMCYCLE
Another
more subtle obvious positive can be found using JCapper FormCycle. The chart
below shows horses with a FormCycle of one (have run a good race in each of
their three most recent starts) actually have a lower flat bet win roi than
horses with FormCycles of two or three (one bad race from among their three
most recent starts.) I will demonstrate later how you can use this to your
advantage and actually strengthen the flat win bet roi of your UDMs.
FormCycle
By: Form Cycle
Gain Bet
Roi Wins Plays Pct
Impact
0 0.00 0.00 0.0000
0 0 .0000
0.0000 Missing
Info
1
-17148.50 78624.00
0.7819 6945 39312
.1767 1.4300 GOOD GOOD GOOD
2 -9048.20 45972.00 0.8032
3769 22986 .1640
1.3273 GOOD GOOD BAD
3 -7363.50 35378.00 0.7919
2595 17689 .1467
1.1875 GOOD BAD
GOOD
4 -12497.50 45996.00
0.7283 3005 22998
.1307 1.0577 GOOD
BAD BAD
5 -11093.70 40754.00
0.7278 2338 20377
.1147 0.9288 BAD
GOOD GOOD
6 -6644.40 31320.00 0.7879
1589 15660 .1015
0.8214 BAD GOOD
BAD
7 -5889.40 31300.00 0.8118
1428 15650 .0912
0.7386 BAD BAD GOOD
8 -17660.60 67538.00
0.7385 2168 33769
.0642 0.5197 BAD BAD BAD
9 -7455.50 27802.00 0.7318
1160 13901 .0834
0.6755 FIRST TIME STARTER
LATE SPEED
Very early
on my research into handicapping factors revealed that the betting public pays
more attention to late speed than early speed. In the overwhelming majority of
races blindly betting the top last line 2f pace call horse (with no further
handicapping whatsoever) will yield a higher flat bet win roi than the same
blind bet on the top last line late pace horse. This realization was the
catalyst for me to think in terms of models or UDMs rather than in terms of
individual handicapping factors. As soon as I began using factors in my UDMs
that pointed to horses with hidden positives rather than obvious positives, I
noticed an immediate improvement in my own results at the windows.
By: Late Pace Rank (Last Start)
Rank Gain Bet
Roi Wins Plays Pct
Impact
1 -9842.70 54818.00 0.8204
5094 27409 .1859
1.5044
2 -11574.60 52322.00
0.7788 4116 26161
.1573 1.2736
3 -11369.90 51398.00
0.7788 3603 25699
.1402 1.1349
4 -11835.00 50860.00
0.7673 3237 25430
.1273 1.0304
5 -12380.70 49268.00
0.7487 2732 24634
.1109 0.8977
6 -11370.60 45096.00
0.7479 2303 22548
.1021 0.8268
7 -10340.30 36798.00
0.7190 1660 18399
.0902 0.7303
8 -7274.30 26998.00 0.7306
1040 13499 .0770
0.6236
9 -4719.90 17698.00 0.7333
611 8849 .0690
0.5589
10 -1669.00 10958.00 0.8477
363 5479 .0663
0.5363
11 -1629.20
5164.00 0.6845 147
2582 .0569 0.4608
12 -355.60
2472.00 0.8561 75
1236 .0607 0.4912
13 -253.50
554.00 0.5424 14
277 .0505 0.4091
14
-156.80 244.00 0.3574
1 122 .0082
0.0663
15 -16.00
16.00 0.0000 0
8 .0000 0.0000
16 -7.20
14.00 0.4857 1
7 .1429 1.1564
17 0.00
0.00 0.0000 0
0
.0000 0.0000
18 -4.00
4.00 0.0000 0
2 .0000 0.0000
19 -2.00
2.00 0.0000 0
1 .0000 0.0000
By: 73
rankForPaceFig_2F_InLast
Rank Gain
Bet Roi
Wins Plays Pct
Impact
1 -7205.30 56844.00 0.8732
5324 28422 .1873
1.5163
2 -10410.70 52914.00
0.8033 4096 26457
.1548 1.2532
3 -11074.70 51246.00
0.7839 3534 25623 .1379
1.1164
4 -11756.90 50578.00
0.7675 3178 25289
.1257 1.0172
5 -14332.70 48990.00
0.7074 2664 24495
.1088 0.8803
6 -11134.90 44726.00
0.7510 2293 22363
.1025 0.8300
7 -10875.10 36390.00
0.7012 1641 18195
.0902 0.7301
8 -7821.70 26544.00 0.7053
1060 13272 .0799
0.6465
9 -4975.50 17432.00 0.7146
638 8716 .0732
0.5925
10 -3029.30 10772.00 0.7188
345 5386 .0641
0.5185
11 -1570.90
5152.00 0.6951 149
2576 .0578 0.4682
12 -432.00
2354.00 0.8165 59
1177 .0501 0.4058
13 30.20
510.00 1.0592 14
255 .0549 0.4444
14 -199.80
220.00 0.0918 2
110 .0182 0.1472
15 -2.00
2.00 0.0000 0
1 .0000 0.0000
16 -4.00
4.00 0.0000 0
2 .0000 0.0000
17 0.00
0.00
0.0000 0 0 .0000
0.0000
18 -2.00
2.00 0.0000 0
1 .0000 0.0000
19+ -4.00
4.00 0.0000 0
2 .0000 0.0000
CONNECTIONS
Another
subtle obvious positive exists when horses are ranked first in CXN Rating or
CScore. Take a look at the chart presented below. Horses ranked second in CXN
Rating actually have a higher flat bet win roi than horses ranked first. During
UDM development, once the point has been reached where the UDM is doing a good
job of selecting horses with hidden positives, adding a requirement that CXN
Rating or CScore min rank be greater than or equal to two (using the
CSCORERKMIN02 Preset Filter) can often provide a surprising boost in win bet
roi.
By: CXN Rating Rank
Rank Gain Bet
Roi Wins Plays Pct
Impact
1 -8714.40 50528.00 0.8275
5855 25264 .2318
1.8760
2 -8335.00 49780.00 0.8326
4597 24890 .1847
1.4950
3 -9385.00 49630.00 0.8109
3731 24815 .1504
1.2171
4 -11057.60 49600.00
0.7771 3146 24800
.1269 1.0268
5 -11914.30 49056.00
0.7571 2556 24528
.1042 0.8435
6 -13002.10 46032.00
0.7175 1890 23016
.0821 0.6647
7 -11698.80 38412.00
0.6954 1353 19206
.0704 0.5702
8 -7099.80 28668.00 0.7523
883 14334 .0616
0.4986
9 -6162.60 19860.00 0.6897
491 9930 .0494
0.4002
10 -4459.60 12756.00 0.6504
310 6378 .0486
0.3934
11 -1344.80
6570.00 0.7953 131
3285 .0399 0.3228
12 -1513.40
3064.00 0.5061 43
1532 .0281 0.2272
13
-151.10 564.00 0.7321
6 282 .0213
0.1722
14 35.00
152.00 1.2303 4
76 .0526 0.4260
15 -2.00
2.00 0.0000 0
1 .0000 0.0000
16 -2.00
2.00 0.0000 0
1 .0000 0.0000
17 12.20
2.00 7.1000 1
1 1.0000 8.0947
18 -6.00
6.00 0.0000 0
3 .0000 0.0000
19 0.00
0.00
0.0000 0 0 .0000
0.0000
My
coverage of specific factors containing obvious positives (which are almost
always over bet) ends here. Understand that I have only covered a very limited
number of factors. Without question there are many other factors not covered in
this document that contain obvious positives when looked at in the Data Window.
Again, your job as a JCapper user/horseplayer is to do your own research and
bring those factors into your own models in such a way as to strengthen your
overall results.
VALUE
JCapper
contains many factors that will enable you to reach out and “grab” the concept
of value and bring it into your UDMs. Later, after we develop a UDM that
contains Hidden Positives and avoids common Obvious Positives, I’ll show you
actual Data Window output where results for that UDM are broken out by some of
the value based factors in JCapper such as PScore, Race Volatility, Field Size,
JPROfMLFAV1, and Odds Line Ratios such as OR3, MLOR3, and MLOR4. Once you’ve
developed a model that contains a good mix of hidden positives while at the
same time avoids obvious positives – quite often all that is required to move
your UDM from negative into positive territory is the addition of one or more
value based factors.
COMPREHENSIVE POWER RATINGS
When I developed
the original JRating I discovered that I could create a rating with a flat bet
win roi that was superior to other commercially available power ratings. I got
a lot of emails from horseplayers (and other software developers too) wondering
how I was able to accomplish this. Looking back it wasn’t hard to do this at
all. I simply used factors that involve hidden positives. I discovered I could
strengthen the rating even further by lowering the weightings of certain
factors involving obvious positives. JCapper contains a number of comprehensive
power ratings: Alchemy, JPR, JPRClass, JRating (in the original program only),
Prime Power, QRating, PRating, and UPR. All of these are examples of
comprehensive power ratings. Depending on the specific situation being
addressed, each can be used to good effect when creating UDMs. But in the
interest of keeping things simple, in this document, I will specifically focus
on using the hidden positives built into JPR to illustrate the process of
creating a Positive Expectation UDM as the basis of finding an edge.
JPR
By:
JPR Rank
Rank Gain Bet
Roi Wins Plays Pct
Impact
1 -4201.90 49656.00 0.9154
7401 24828 .2981
2.4129
2 -7015.80 49658.00 0.8587
4982 24829 .2007
1.6242
3 -8186.00 49658.00 0.8352
3676 24829 .1481
1.1984
4 -9533.10 49604.00 0.8078
2781 24802 .1121
0.9076
5 -12419.20 48986.00
0.7465 2150 24493
.0878 0.7105
6
-13247.30 45970.00
0.7118 1569 22985
.0683 0.5526
7 -12498.10 38410.00
0.6746 1078 19205
.0561 0.4544
8 -11488.90 28770.00
0.6007 641 14385
.0446 0.3607
9 -6677.70 19828.00 0.6632
372 9914 .0375
0.3037
10 -4872.70 12810.00 0.6196
212 6405 .0331
0.2679
11 -2885.10
6718.00 0.5705 86
3359 .0256 0.2072
12 -1497.70
3646.00 0.5892 41
1823 .0225 0.1821
13 9.80 642.00
1.0153 7 321
.0218 0.1765
14 -275.60
316.00 0.1278 1
158 .0063 0.0512
15 -2.00
2.00 0.0000 0
1 .0000 0.0000
16 -2.00
2.00 0.0000 0
1 .0000 0.0000
17 -2.00
2.00 0.0000 0
1 .0000 0.0000
18 -2.00
2.00 0.0000 0
1 .0000 0.0000
19 -4.00 4.00
0.0000 0 2
.0000 0.0000
Track Surfaces Update (
This section (shown in red font) should serve to both enlighten and
warn the user-player about important differences between track surfaces from
one track to another and changes in the nature of the game starting in 2007.
Back in 2001, when I first started wrapping my head around taking
concepts like track weight and speed and shaping models into the JRating
algorithm, PolyTrack didn't exist. All dirt tracks (taken as a whole) seemed to
share one common trait: Horses with early speed enjoyed a distinct advantage
over horses without it.
As a result I was able to create one JRating algorithm for dirt and
another for turf. I then applied these two algorithms to all tracks everywhere.
For the past several years that approach has let me (and many other JCapper
users) enjoy almost universal success simply by playing JRating rank = 1 horses
under a wide variety of circumstances.
When I set about developing JPR in early 2005 I used very similar
thinking to what I did when I created the JRating. I created one JPR algorithm
for dirt and a second algorithm for turf. And it worked pretty much everywhere
I wanted to play.
The advent of PolyTrack changes that. It plays differently than
plain old dirt. Horses running on or near the lead on PolyTrack no longer have
an advantage over their rivals. Instead, horses with early speed on PolyTrack,
unless they are in a race with little or no pace pressure, are at a
disadvantage. I am convinced that poly is different enough that it deserves
treatment as a separate surface.
As I write this (
We can argue about reasons both for and against PolyTrack until we
are blue in the face. But the simple truth is this: It is now part of the game.
There is a belief among horsemen that artificial surfaces are safer for horses
to race on. Wait. It gets better. I have seen track management at some tracks
have their maintenance crews change dirt surfaces that have historically
favored early speed into deeper more tiring surfaces that favor closers.
In my opinion the game
is currently undergoing a landscape change.
Universal data samples of the type used as a basis for creating the
My-Edge UDM in the next section of this document need to be examined more
closely. I have come to believe that UDMs and UPR should really be modeled on a
track by track basis.
Now, here is my word of
warning: The UDM I am showing you in the next section of this document is
based on a wide data sample across all distances and surfaces. That data sample
was taken at the end of 2006. That sample may not be representative of track
surfaces as they exist when you download race card files in 2007 and beyond. My
real world experience shows that the My-Edge UDM outlined in this document
works very well on speed favoring surfaces. My experience also tells me that
the UDM does not work on speed tiring surfaces. It is not my intention to lead
you in any way to think the specific universal UDM outlined in this document
and based on 2006 data will somehow magically work across any and all track
surfaces found in years 2007 and beyond. Rather it is my intention to always get
you in the habit of thinking on your own – and testing your UDMs and UPR using data
that is track specific.
For that reason I must warn you that any UDMs or UPR you develop as a result of this document should be
tested using data that is track specific for any tracks you actually plan on
playing.
But don’t let that deter you. At the very least the following
section will show you the exact concepts, thought processes, and procedures I
use myself when developing UDMs and UPR for successful live play.
CREATING POSITIVE EXPECTATION UDMs TO YIELD AN EDGE
In this
next section I am going to present a step by step tutorial where I create a
Positive Expectation UDM to yield an edge. While the example UDM I am about to
present provides the beginning player an edge – please don’t limit yourself to
using only what I present here. If you do you will really be missing out. Understand going in that what I am about to
present – even though it provides the beginning player an edge – really only
scratches the surface of what is possible. Feel free to use your own
ingenuity and creativity when creating UDMs and UPR. JCapper2007 is a very
robust program. The number of conditionals you can add to a UDM like this one
(or one of your own) is limited only by your own imagination.
GENERAL FRAMEWORK for MODEL
DEVELOPMENT
Whenever
I create a new UDM (or UPR) I like to start out by focusing on getting my model
to include horses with hidden positives. Once the model is doing a pretty good
job of selecting horses with hidden positives I’ll then shift my focus to
having the model avoid horses with obvious positives. Once I feel the model is
doing a good job of selecting horses with hidden positives while at the same
time avoiding horses with obvious positives, I’ll then shift my focus towards
bringing value based factors into my model. Once I have a model that performs
well after bringing in one or more value based factors, then I’ll shift my
focus to testing that model against a validation database that contains races
not included in the database used to develop the model. If the model performs
well against the validation database then I’ll consider integrating the model
into my own live play.
I know
from experience that not every factor (or combination of factors) can be
modeled successfully to perform well going forward at the windows. In fact very
few of them can. But my own experience tells me that when I apply this general
framework of model development around good factors – more often than not I get
a model that performs well going forward – a model that I can use in my own
live play if I want to. Further, my experience also tells me that models
developed in a haphazard manner outside of this general framework - those are
the models that frequently fall apart going forward.
My
advice, if you want it, would be to give this general framework a try whenever
you develop models of your own. Stick to this general framework until your own
experience grows to the point to where you start to develop a general framework
for model development of your own.
The MY-EDGE UDM
Before
attempting this tutorial you should already have a solid understanding of how
to use the UDM Wizard to create simple basic UDMs. If you haven’t already done
so, listen to the Simple Basic UDMs Podcast while working with the UDM
Wizard to create a handful of basic UDMs. You should also become familiar with
both the Preset and Dynamic Filter Tools. Once you are comfortable adding and
editing UDM factor constraints to UDMs then
and only then are you probably ready for something a little more advanced
like this tutorial.
STEP1
Let’s
start out by focusing on hidden positives. In the UDM Wizard, create a new UDM
named MY-EDGE. Set JPR min and max rank equal to 1. Then run the UDM through the
Data Window broken out by the CANTRUN/XFORM Attribute. Actual Data Window
results using my own calendar year 2006 database are shown below:
UDM Definition: MY-EDGE
Divisor: # UDM Def Divisor: 999
Surface Req: *ANY Surface*
Distance Req: *ANY Distance*
JPR: MinRank= 1 MaxRank= 1
MinVal= -999
MaxVal= 999
MinGap= -999
MaxGap= 999
Running Style: ALL
Data Window Settings:
Divisor = 999
Filters Applied: -
Surface: (ALL*) Distance: (All*)
(From Index File: D:\2007\Q1_2007\pl_JPR_1_06.txt)
Data
Mutuel Totals 45454.10 45973.40 44528.60
Bet -49656.00 -49656.00 -49656.00
Gain -4201.90 -3682.60 -5127.40
Wins 7401 12313
15355
Plays 24828 24828 24828
PCT .2981 .4959
.6185
ROI 0.9154 0.9258
0.8967
Avg Mut 6.14 3.73
2.90
By: CANTRUN/XFORM
Attribute
CANTRUN Gain Bet
Roi Wins Plays Pct
Impact
0 -1982.60 32502.00 0.9390
4851 16251 .2985
1.0014 Clean Horse
1 -2212.90 17136.00 0.8709
2548 8568 .2974
0.9976 XFORM
2 -6.40
18.00 0.6444 2
9 .2222 0.7455 CANTRUN
The above
results indicate that CLEAN horses contain the trainer intent hidden positives
we seek. By eliminating XFORM and CANTRUN horses we will improve our overall
results.
STEP 2
Using the
Preset Filter Tool, add the CLEAN Preset Filter to the UDM Definition. The run
the UDM through the Data Window broken out by BESTE2 Gap:
UDM Definition: MY-EDGE
Divisor: # UDM Def Divisor: 999
Surface Req: *ANY Surface*
Distance Req: *ANY Distance*
JPR: MinRank= 1 MaxRank= 1
MinVal= -999
MaxVal= 999
MinGap= -999
MaxGap= 999
Running Style: ALL
Data Window Settings:
Divisor = 999
Filters Applied:
CLEAN-
Surface: (ALL*) Distance: (All*)
(From Index File: D:\2007\Q1_2007\pl_JPR_1_06.txt)
Data
Mutuel Totals 30519.40 30633.20 29418.70
Bet -32502.00 -32502.00 -32502.00
Gain -1982.60 -1868.80 -3083.30
Wins 4851 8001
9984
Plays 16251 16251 16251
PCT .2985 .4923
.6144
ROI 0.9390 0.9425
0.9051
Avg Mut 6.29 3.83
2.95
By: Best E2 Gap
>=Min <Max Gain
Bet Roi Wins Plays Pct
Impact
-999.00 -15.00
158.80 2176.00 1.0730
320 1088 .2941
0.9853
-15.00 -14.00
-109.90 372.00 0.7046
47 186 .2527
0.8465
-14.00 -13.00
-56.00 426.00 0.8685
62 213 .2911
0.9751
-13.00 -12.00
-86.40 560.00 0.8457
77 280 .2750
0.9213
-12.00 -11.00
10.20 500.00 1.0204
69 250 .2760
0.9246
-11.00 -10.00
-40.60 684.00 0.9406
92 342 .2690
0.9012
-10.00 -9.00
-101.40 838.00 0.8790
109 419 .2601
0.8715
-9.00 -8.00
-183.60 928.00 0.8022
122 464 .2629
0.8808
-8.00 -7.00
-74.30 1064.00 0.9302
156 532 .2932
0.9823
-7.00 -6.00
-67.60 1206.00 0.9439
181 603 .3002
1.0056
-6.00 -5.00
-60.00 1414.00 0.9576
194 707 .2744
0.9192
-5.00
-4.00 -81.10 1390.00
0.9417 205 695
.2950 0.9881
-4.00 -3.00
-106.30 1662.00 0.9360
231 831 .2780
0.9312
-3.00 -2.00
-252.80 1760.00 0.8564
245 880 .2784
0.9327
-2.00 -1.00
-310.40 1732.00 0.8208
239 866 .2760
0.9245
-1.00 0.00
17.00 1664.00 1.0102
266 832 .3197
1.0710
0.00 1.00
-39.70 2360.00 0.9832
366 1180 .3102
1.0391
1.00 2.00
-194.60 1712.00 0.8863
251 856 .2932
0.9823
2.00 3.00
-42.10 1584.00 0.9734
251 792 .3169
1.0617
3.00 999999.00 -361.80
8470.00 0.9573 1368
4235 .3230 1.0821
The above
results show that horses within 8 points of the BestE2 in the race contain the
early speed hidden positives we seek. Eliminating horses not within 8 points of
the race top for BestE2 will improve overall results.
STEP 3
Set min
gap for BestE2 to -8 and run the UDM through the Data Window broken out by
Recent Activity Days Last Start:
UDM
Definition: MY-EDGE
Divisor: # UDM Def Divisor: 999
Surface Req: *ANY Surface*
Distance Req: *ANY Distance*
BestE2: MinRank= -999 MaxRank= 999
MinVal= -999
MaxVal= 999
MinGap= -8
MaxGap= 999
JPR: MinRank= 1 MaxRank= 1
MinVal= -999
MaxVal= 999
MinGap=
-999 MaxGap=
999
Running Style: ALL
Data Window Settings:
Divisor = 999
Filters Applied:
CLEAN-
Surface: (ALL*) Distance: (All*)
(From Index File: D:\2007\Q1_2007\pl_JPR_1_06.txt)
Data
Mutuel Totals 24444.30 24196.90 23601.10
Bet -26018.00 -26018.00 -26018.00
Gain -1573.70 -1821.10 -2416.90
Wins 3953 6485
8048
Plays 13009 13009 13009
PCT .3039 .4985
.6186
ROI 0.9395 0.9300
0.9071
Avg Mut 6.18 3.73
2.93
By: Recent Activity-
Days Last Start
>=Min <Max Gain
Bet Roi
Wins Plays Pct
Impact
-999 0
0.00 0.00 0.0000
0 0 .0000
0.0000
0 5
-6.00 12.00 0.5000
1 6 .1667
0.5485
5 10
-159.60 612.00 0.7392
84 306 .2745
0.9034
10 15
-48.10 1936.00 0.9752
279 968 .2882
0.9485
15 20
-29.00 2466.00 0.9882
367 1233 .2976
0.9795
20 25
-386.70 4422.00 0.9126
656 2211 .2967
0.9764
25 30
-339.40 4216.00 0.9195
645 2108 .3060
1.0069
30 35
-297.60 2678.00 0.8889
405 1339 .3025
0.9954
35 40
-85.80 2338.00 0.9633
388 1169 .3319
1.0923
40 45
88.80 1762.00 1.0504
312 881 .3541
1.1655
45 50
100.40 1016.00 1.0988
160 508 .3150
1.0365
50 55
-10.30 592.00 0.9826
87 296 .2939
0.9673
55 60
108.00 614.00 1.1759
104 307 .3388
1.1148
60 65
-71.10 416.00 0.8291
59 208 .2837
0.9335
65 70
-16.10 256.00 0.9371
40 128 .3125
1.0284
70 75
7.20 258.00 1.0279
41 129 .3178
1.0460
75 80
-94.00 260.00 0.6385
28 130 .2154
0.7088
80 85
52.30 184.00 1.2842
36 92 .3913
1.2878
85
90 -35.10 110.00
0.6809 11 55
.2000 0.6582
90 999999
-351.60 1870.00 0.8120
250 935 .2674
0.8799
The above
results show that horses with a most recent start between 10 and 59 days of
today’s race date contain desired form based hidden positives. Eliminating the
others will have the effect of improving overall results.
At this
point the UDM itself is already doing a pretty good job of selecting horses
with hidden positives. For illustrative purposes, in the next step I will shift
focus away from finding hidden positives and move on to avoiding horses with
obvious positives. However, feel free on
your own to tune this (or any other) UDM as you see fit by testing and using
any and all factors that might point out horses with hidden positives.
STEP 4
Set Rec
Activity Days Last Start min val=10 and max val=59 and
then run the UDM through the Data Window broken out by some of the factors that
commonly point out obvious positives:
.
UDM Definition: MY-EDGE
Divisor: # UDM Def Divisor: 999
Surface Req: *ANY Surface*
Distance Req: *ANY Distance*
BestE2: MinRank= -999 MaxRank= 999
MinVal= -999
MaxVal= 999
MinGap= -8 MaxGap= 999
Days Last Start: MinVal= 10 MaxVal= 59
JPR: MinRank= 1 MaxRank= 1
MinVal= -999
MaxVal= 999
MinGap= -999
MaxGap= 999
Running Style: ALL
Data Window Settings:
Divisor = 999
Filters Applied:
CLEAN-
Surface: (ALL*) Distance: (All*)
(From Index File: D:\2007\Q1_2007\pl_JPR_1_06.txt)
Data
Mutuel Totals 21140.30 20782.60 20185.40
Bet -22040.00 -22040.00 -22040.00
Gain -899.70 -1257.40 -1854.60
Wins 3403 5586
6908
Plays 11020 11020 11020
PCT .3088 .5069
.6269
ROI 0.9592 0.9429
0.9159
Avg Mut 6.21 3.72
2.92
By: Form Cycle
FC Gain
Bet Roi Wins Plays Pct
Impact
0 0.00
0.00
0.0000 0 0 .0000 0.0000
Missing Info
1 -780.30 7218.00
0.8919 1161 3609
.3217 1.0418 GOOD GOOD GOOD
2 -166.40 3684.00
0.9548 621 1842
.3371 1.0917 GOOD GOOD BAD
3 52.60
2462.00 1.0214 392
1231 .3184 1.0312
GOOD BAD GOOD
4 -236.10 3044.00
0.9224 467 1522
.3068 0.9936 GOOD BAD
BAD
5 94.10
2006.00 1.0469 308
1003 .3071 0.9944
BAD GOOD GOOD
6 20.20
1274.00 1.0159 173
637 .2716 0.8795
BAD GOOD BAD
7 235.00 1080.00
1.2176 144 540
.2667 0.8636 BAD BAD GOOD
8 -118.80 1272.00
0.9066 137 636
.2154 0.6976 BAD BAD BAD
9 0.00
0.00
0.0000 0 0 .0000 0.0000
FIRST TIME STARTER
By: Morning Line Rank
Rank Gain Bet
Roi Wins Plays Pct
Impact
1 -1442.20 10612.00 0.8641
2040 5306 .3845
1.2450
2 -202.00
4696.00 0.9570 680
2348 .2896
0.9378
3 96.50
2722.00 1.0355 329
1361 .2417 0.7828
4 130.10
1688.00 1.0771 164
844 .1943 0.6292
5 229.00
1080.00 1.2120 105
540 .1944 0.6297
6 245.80
594.00 1.4138 49
297 .1650 0.5343
7 -13.40
354.00 0.9621 18
177 .1017 0.3293
8 110.70
180.00 1.6150 14
90 .1556 0.5037
9 -61.80
70.00 0.1171 1
35 .0286 0.0925
10 3.60
30.00 1.1200 2
15 .1333 0.4318
11 4.00
14.00 1.2857 1
7 .1429 0.4626
By:
CXN Rating Rank
Rank Gain Bet
Roi Wins Plays Pct
Impact
1 -542.00 5190.00
0.8956 992 2595
.3823 1.2379
2 -216.00
4102.00 0.9473 702
2051 .3423 1.1084
3 219.00
3580.00 1.0612 563
1790 .3145 1.0185
4 -242.70
2846.00 0.9147 408
1423 .2867 0.9285
5 47.90
2252.00 1.0213 308
1126 .2735 0.8858
6 -265.80 1622.00
0.8361 184 811
.2269 0.7347
7 11.90
1036.00 1.0115 112
518 .2162 0.7002
8 42.80
708.00 1.0605 65
354 .1836 0.5946
9 12.00
370.00 1.0324 39
185 .2108 0.6827
10 -24.60 188.00
0.8691 14 94
.1489 0.4823
11 39.00
94.00 1.4149 10
47 .2128 0.6890
12 30.80
40.00 1.7700 6
20 .3000 0.9715
13 -8.00
8.00 0.0000 0
4 .0000 0.0000
14 -4.00
4.00 0.0000 0
2 .0000 0.0000
15 0.00
0.00
0.0000 0 0 .0000
0.0000
16 0.00
0.00
0.0000 0 0 .0000
0.0000
17 0.00
0.00 0.0000
0 0 .0000
0.0000
18 0.00
0.00
0.0000 0 0 .0000
0.0000
19 0.00
0.00
0.0000 0 0 .0000
0.0000
STEP 5
Eliminate
horses with obvious positives in their records. Use the Preset Filter Tool to
set min rank for FormCycle to 2. Use the Dynamic Filter Tool to set Morning
Line min rank to 2 (and possibly max rank to 8.)
For
illustrative purposes the only factor constraints I set related to Obvious Positives
were FormCycle Min Val (FCYCLEMIN02) and Morning Line Rank ([ALL[MIN2.) I purposely didn’t set a
constraint for CXN Rating (CScore) because overlap exists between Morning Line
Rank=1 and CXN Rating (CScore) Rank=1 horses. However, feel free to experiment
on your own. Set constraints for any and all factors where your own Data Window
output indicates that obvious positives exist and then run your UDM through the
Data Window and look at your own results.
After
setting constraints for Obvious Positives then run the UDM through the Data
Window broken out by several factors that represent value:
UDM Definition: MY-EDGE
Divisor: # UDM Def Divisor: 999
Surface Req: *ANY Surface*
Distance Req: *ANY Distance*
BestE2: MinRank= -999 MaxRank= 999
MinVal= -999 MaxVal= 999
MinGap= -8 MaxGap= 999
Days Last Start: MinVal= 10 MaxVal= 59
JPR: MinRank= 1 MaxRank= 1
MinVal= -999 MaxVal= 999
MinGap= -999 MaxGap= 999
Running Style: ALL
Data Window Settings:
Divisor = 999
Filters Applied: CLEAN-FCYCLEMIN02-[ALL[MIN2-
Surface: (ALL*) Distance: (All*) (From Index File:
D:\2007\Q1_2007\pl_JPR_1_06.txt)
Data
Mutuel Totals 8778.30
8052.30 7717.30
Bet -8196.00 -8196.00 -8196.00
Gain 582.30 -143.70
-478.70
Wins 955 1664
2192
Plays 4098 4098 4098
PCT .2330 .4061
.5349
ROI 1.0710 0.9825
0.9416
Avg Mut 9.19 4.84
3.52
By:
JPROfMLFav1
>=Min <Max Gain
Bet Roi Wins Plays Pct
Impact
-999.00 15.00 0.00
0.00
0.0000 0 0 .0000
0.0000
15.00 20.00
0.00 0.00 0.0000
0 0 .0000
0.0000
20.00 25.00
0.00 0.00 0.0000
0 0 .0000
0.0000
25.00 30.00
0.00 0.00 0.0000
0 0 .0000
0.0000
30.00 35.00
0.00 0.00 0.0000
0 0 .0000
0.0000
35.00 40.00
-2.00 46.00 0.9565
3 23 .1304
0.5597
40.00 45.00
-26.80 138.00 0.8058
14 69 .2029
0.8707
45.00 50.00 -14.80
274.00 0.9460 24
137 .1752 0.7517
50.00 55.00 74.90
678.00 1.1105 78
339 .2301 0.9873
55.00 60.00
109.10 1388.00 1.0786
150 694 .2161
0.9275
60.00 65.00
270.10 2072.00 1.1304
262 1036 .2529
1.0852
65.00
70.00 235.00 2072.00
1.1134 242 1036
.2336 1.0024
70.00 75.00
-34.80 1232.00 0.9718
146 616 .2370
1.0170
75.00 80.00
-26.20 288.00 0.9090
35 144 .2431
1.0430
80.00 85.00
-2.20 8.00 0.7250
1 4 .2500
1.0728
85.00 90.00
0.00 0.00 0.0000
0 0 .0000
0.0000
90.00 95.00
0.00 0.00 0.0000
0 0 .0000
0.0000
95.00 100.00
0.00 0.00 0.0000
0 0 .0000
0.0000
100.00 105.00
0.00 0.00 0.0000
0 0 .0000
0.0000
105.00 999999.00 0.00
0.00
0.0000 0 0 .0000
0.0000
By: PScore
>=Min
<Max Gain Bet
Roi Wins Plays Pct
Impact
-999.00 0.00
0.00
0.00
0.0000 0 0 .0000
0.0000
0.00
10.00 0.00 0.00 0.0000
0 0 .0000
0.0000
10.00 20.00
0.00 0.00 0.0000
0 0 .0000
0.0000
20.00
30.00 26.40 36.00
1.7333 7 18
.3889 1.6688
30.00
40.00 -43.00 130.00
0.6692 11 65
.1692 0.7262
40.00
50.00 -79.00 352.00
0.7756 37 176
.2102 0.9021
50.00 60.00
36.10 524.00 1.0689
74 262 .2824
1.2120
60.00
70.00 135.20 790.00
1.1711 101 395
.2557 1.0972
70.00
80.00 79.90
888.00 1.0900 108
444 .2432 1.0438
80.00
90.00 112.40 814.00
1.1381 103 407
.2531 1.0860
90.00
100.00 117.20 808.00
1.1450 97 404
.2401 1.0303
100.00
110.00 -82.20 714.00
0.8849 68 357
.1905 0.8174
110.00
120.00 27.20 598.00
1.0455 71 299
.2375 1.0190
120.00
130.00 -36.60 478.00
0.9234 52 239
.2176 0.9336
130.00
140.00 107.70 392.00
1.2747 49 196
.2500 1.0728
140.00
150.00 25.30 344.00
1.0735 40 172
.2326 0.9979
150.00
160.00 152.60 302.00
1.5053 36 151
.2384 1.0230
160.00
170.00 17.90 206.00
1.0869 20 103
.1942 0.8332
170.00
180.00 -36.40 152.00
0.7605 12 76
.1579 0.6775
180.00 999999.00 21.60
668.00 1.0323 69
334 .2066 0.8865
By: Field Size
Field Size Gain
Bet Roi
Wins Plays Pct
Impact
1 0.00
0.00
0.0000 0 0 .0000
0.0000
2 0.00
0.00
0.0000 0 0 .0000
0.0000
3 1.60
8.00 1.2000 2
4 .5000 2.1455
4 5.60
50.00 1.1120 11
25 .4400 1.8881
5 5.20
448.00 1.0116 71
224 .3170 1.3601
6 -139.00
1182.00 0.8824 144
591 .2437 1.0455
7 49.70 1482.00
1.0335 187 741
.2524 1.0829
8 200.20
1494.00 1.1340 176
747 .2356 1.0110
9 86.70
1178.00 1.0736 119
589 .2020 0.8670
10 205.10
1090.00 1.1882 119
545 .2183 0.9370
11 -0.70
520.00 0.9987 50
260 .1923 0.8252
12 106.90
638.00 1.1676 66
319 .2069 0.8878
13 4.40
48.00 1.0917 4
24 .1667 0.7152
14 58.60
56.00 2.0464 6
28 .2143 0.9195
15 0.00
0.00
0.0000 0 0 .0000
0.0000
16 0.00
0.00
0.0000 0 0 .0000
0.0000
17 0.00
0.00
0.0000 0 0 .0000
0.0000
18 0.00
0.00
0.0000 0 0 .0000
0.0000
19+
-2.00 2.00 0.0000
0 1 .0000
0.0000
By: Race Volatility
>=Min <Max Gain
Bet Roi Wins Plays Pct
Impact
-999.00 0.00
-20.20 50.00 0.5960
5 25 .2000
0.8582
0.00
10.00 -1.20 8.00
0.8500 1 4
.2500 1.0728
10.00
20.00 -6.40 10.00
0.3600 1 5
.2000 0.8582
20.00
30.00 -4.40 22.00
0.8000 4 11
.3636 1.5604
30.00
40.00 38.80 80.00
1.4850 16 40
.4000 1.7164
40.00
50.00 -33.10 132.00
0.7492 19 66
.2879 1.2353
50.00 60.00
62.50 298.00 1.2097
55 149 .3691
1.5840
60.00
70.00 -11.00 430.00
0.9744 63 215
.2930 1.2574
70.00
80.00 73.40 564.00
1.1301 78 282
.2766 1.1869
80.00
90.00 -9.60 712.00
0.9865 94 356
.2640 1.1330
90.00
100.00 -43.40 820.00
0.9471 89 410
.2171 0.9315
100.00
110.00 50.70 844.00
1.0601 106 422
.2512 1.0779
110.00
120.00 -59.40 760.00
0.9218 73 380
.1921 0.8243
120.00
130.00 148.20
690.00 1.2148 81
345 .2348 1.0075
130.00
140.00 50.90 506.00
1.1006 55 253
.2174 0.9328
140.00
150.00 226.50 534.00
1.4242 61 267
.2285 0.9804
150.00
160.00 36.30 480.00
1.0756 42 240
.1750 0.7509
160.00
170.00 -68.70 344.00
0.8003 29 172
.1686 0.7235
170.00
180.00 95.20 216.00
1.4407 26 108
.2407 1.0330
180.00 999999.00 57.20
696.00 1.0822 57
348 .1638 0.7029
By:
MLOR3
>=Min <Max Gain
Bet Roi Wins Plays Pct
Impact
-999.00 0.50 1.80
2.00 1.9000 1
1 1.0000 4.2911
0.50 0.55
-2.00 2.00 0.0000
0 1 .0000
0.0000
0.55 0.60
-3.80 10.00 0.6200
1 5 .2000
0.8582
0.60 0.65
-2.60 18.00 0.8556
3 9 .3333
1.4304
0.65 0.70
-14.00 14.00 0.0000
0 7 .0000
0.0000
0.70 0.75
5.40 32.00 1.1688
6 16 .3750
1.6092
0.75 0.80
-11.60 64.00 0.8188
8 32 .2500
1.0728
0.80 0.85
-8.40 60.00 0.8600
7 30 .2333
1.0013
0.85 0.90
-11.70 92.00 0.8728
12 46 .2609
1.1194
0.90 0.95
16.30 98.00 1.1663
13 49 .2653
1.1385
0.95 1.00
-15.70 102.00 0.8461
13 51 .2549
1.0938
1.00 1.05
31.00 136.00 1.2279
21 68 .3088
1.3252
1.05 1.10
15.50 136.00 1.1140
19 68 .2794
1.1990
1.10 1.15
-24.80 154.00 0.8390
18 77 .2338
1.0031
1.15 1.20
-48.60 160.00 0.6962
17 80 .2125
0.9119
1.20 1.25
-17.70 160.00 0.8894
20 80 .2500
1.0728
1.25 1.30
-7.80 170.00 0.9541
22 85 .2588
1.1106
1.30 1.35 0.20
174.00 1.0011 26
87 .2989 1.2824
1.35 1.40
84.50 192.00 1.4401
31 96 .3229
1.3857
1.40 999999.00 596.30
6420.00 1.0929 717
3210 .2234 0.9585
By: MLOR4
>=Min <Max Gain
Bet Roi Wins Plays Pct
Impact
-999.00 0.50
-10.40 228.00 0.9544
56 114 .4912
2.1079
0.50 0.55
-19.00 106.00 0.8208
19 53 .3585
1.5383
0.55 0.60
12.60 158.00 1.0797
34 79 .4304
1.8468
0.60 0.65
24.60 170.00 1.1447
39 85 .4588
1.9689
0.65 0.70
8.80 138.00 1.0638
28 69 .4058
1.7413
0.70 0.75
-29.40 190.00 0.8453
29 95 .3053
1.3099
0.75 0.80
-48.80 196.00 0.7510
26 98 .2653
1.1385
0.80 0.85
-6.20 222.00 0.9721
36 111 .3243
1.3917
0.85 0.90
-22.30 212.00 0.8948
29 106 .2736
1.1740
0.90 0.95
-8.00 202.00 0.9604
31 101 .3069
1.3171
0.95 1.00
12.00 222.00 1.0541
38 111 .3423
1.4690
1.00 1.05
-19.80 200.00 0.9010
26 100 .2600
1.1157
1.05 1.10
-27.20 206.00 0.8680
25 103 .2427
1.0415
1.10 1.15 32.80
210.00 1.1562 35
105 .3333 1.4304
1.15 1.20
-12.80 186.00 0.9312
23 93 .2473
1.0612
1.20 1.25
41.20 212.00 1.1943
34 106 .3208
1.3764
1.25 1.30
16.80 202.00 1.0832
28 101 .2772
1.1896
1.30 1.35
45.00 190.00 1.2368
29 95 .3053
1.3099
1.35 1.40
14.10 152.00 1.0928
21 76 .2763
1.1857
1.40 999999.00 578.30
4594.00 1.1259 369
2297 .1606 0.6893
By: OR3
>=Min <Max Gain
Bet Roi Wins Plays Pct
Impact
-999.00 0.50
-8.60 46.00 0.8130
11 23 .4783
2.0523
0.50
0.55 -18.60 34.00
0.4529 4 17
.2353 1.0097
0.55
0.60 -20.20 70.00
0.7114 11 35
.3143 1.3486
0.60
0.65 -4.90 106.00
0.9538 19 53
.3585 1.5383
0.65
0.70 -35.50 220.00
0.8386 36 110
.3273 1.4044
0.70 0.75
69.10 340.00 1.2032
69 170 .4059
1.7417
0.75
0.80 -20.10 466.00
0.9569 73 233
.3133 1.3444
0.80
0.85 13.00 466.00
1.0279 69 233
.2961 1.2708
0.85
0.90 16.30 516.00
1.0316 75 258
.2907 1.2474
0.90
0.95 -44.70 576.00
0.9224 71 288
.2465 1.0579
0.95
1.00 30.90 556.00
1.0556 78 278
.2806 1.2040
1.00
1.05 92.30 642.00
1.1438 84 321
.2617 1.1229
1.05
1.10 -14.30 584.00
0.9755 65 292
.2226 0.9552
1.10
1.15 48.90 558.00
1.0876 58 279
.2079 0.8921
1.15
1.20 87.60
442.00 1.1982 47
221 .2127 0.9126
1.20
1.25 57.10 408.00
1.1400 36 204
.1765 0.7573
1.25
1.30 118.70 332.00
1.3575 38 166
.2289 0.9823
1.30
1.35 56.30 288.00
1.1955 30 144
.2083 0.8940
1.35
1.40 4.00 208.00
1.0192 13 104
.1250 0.5364
1.40 999999.00 155.00
1338.00 1.1158 68
669 .1016 0.4362
STEP 6
Pick
cutoffs for one or more value based factors. Use the UDM Wizard where possible
to set cutoffs for factors that are available before the odds are known such as
JPROfMLFav1, MLOR3, PScore, Field Size, and Race Volatility. Use the
BettingInstructions Field in the UDM Wizard to write yourself short notes for
factor cutoffs that are only available after the odds are known such as MLOR4
and OR3.
STEP 7
Validate
the UDM by running it through the Data Window using a Validation Sample. There are
no hard and fast rules here. The main idea is to confront the UDM with fresh
races that were not present in the database used during UDM development and
verify that the UDM does in fact appear to perform well going forward. I use
quarterly databases. I like to develop a UDM using older data from a prior
quarter or year and validate against a quarter that came later time-wise such
as the current quarter. Move on to STEP 8 only after you’ve seen the UDM
perform well going forward against a validation database.
STEP 8
Pick a
strategy for live play. Use the System Definitions Screen to dial in Overlay
Highlighting for MLOR3 and OR3. Consider using the Live Play Module as a tool
to aid you in play or pass decision making once your own Overlay Highlighting
settings have been dialed in. Continue to monitor the performance of the UDM –
and tune it accordingly – as you move forward with live play.
SUMMARY
In this
document I’ve presented some of my own observations about why hidden and
obvious positives exist despite the overall inherent efficiency of the
pari-mutuel pools. Hopefully my strategies for exploiting them during UDM (and
UPR) construction make sense to you after reading this document.
Four
maxims for successful positive expectation UDMs and UPR:
Depending
on the number of JCapper2007 program copies sold, there may well come a time
when a simple JPR Rank=1 UDM like the one presented here will no longer be
effective. When that day comes I guarantee you that the concepts presented
within this document will still be
valid. As long as you are willing to do the work you will be able to use these
same concepts to create successful models using factors other than JPR.
It is my
sincerest hope that after reading this document you will be well on your way to
using JCapper2007 to find an edge of your own.
Jeff Platt