Top Horse Analytics: 08/15/2016 Weekly Statistics

[Augenj]

Welcome to the public subset of the weekly statistics produced by Top Horse Analytics.

The "OCC PC" column means occurs percent.

In the "BET TYPE" column,
"TH" means Top Horse.
"TH 2nd" means the second pick in a race, "TH 3rd" means third pick, etc.
"TH/Fav" means Top Horse when it's also the favorite.
"TH/MLFav" means Top Horse when it's also the morning line favorite.
"TH/Fav/MLF means Top Horse when it's also the favorite and morning line favorite.
"Any X" means any horse in a race that has that flag including the Top Horse.

Horizontal bets use the Top Horse as a single in each race. (Double, Pick 3, etc)
Vertical bets use TH, TH 2nd, TH 3rd, etc, as a single straight bet. (Exacta, Trifecta, etc)
"Any X" flagged horse is a single bet.

The following graphs simulate wagers from the start of the current logical year to the present.

Yellow cells indicate 0.9001 ROI to 1.0000 ROI. (close but no cigar)
Green cells indicate 1.0001 ROI and above. (profitable)

Questions are welcome. :)

[Augenj]

Del Mar

[Augenj]

Gulfstream Park

[Augenj]

Saratoga

[VigorsTheGrey]

Augenj,
I was wondering if it would be possible to run a special line that shows the win % and ROI for the longest shot in the field category...since their are 2 to 14 horses typically in each race, I don't know how it could select the longest in each race...but I am not a programmer and know very little about computers...
Regards, Vigors

[Augenj]

Augenj,
I was wondering if it would be possible to run a special line that shows the win % and ROI for the longest shot in the field category...since their are 2 to 14 horses typically in each race, I don't know how it could select the longest in each race...but I am not a programmer and know very little about computers...
Regards, Vigors
The CSV results files that I download contain the odds of all horses in a race when they left the gate. I believe what you want can be done but I have to find the time to do it. I'll announce it in the first statistics topic if or when it's done. ;)

[VigorsTheGrey]

The CSV results files that I download contain the odds of all horses in a race when they left the gate. I believe what you want can be done but I have to find the time to do it. I'll announce it in the first statistics topic if or when it's done. ;)

That's great! Just kind of curious that's all....not expecting to be surprised here but one never knows....there are a lot of short fields and it may happen more than one thinks....I just look up and shake my head fairly regularly and say "geez, longest shot on the board wins!"

[Augenj]

That's great! Just kind of curious that's all....not expecting to be surprised here but one never knows....there are a lot of short fields and it may happen more than one thinks....I just look up and shake my head fairly regularly and say "geez, longest shot on the board wins!"
Couldn't agree more and strangely enough (or not), the favorite is out of the money. ;)

[formula_2002]

Welcome to the public subset of the weekly statistics produced by Top Horse Analytics.


Not quite the information needed, rather

for all horses in the race, indicate incremental odds and win%. (odds ranges can be established later)
Break out system plays by odds , quantity and number of winners.
To start I would suggest using just your top horse system
If your system is not track specific combine all track totals.
This is the information required to calculate the standard deviation and confidence level.

For example, looking at all horses in the odds range >=2 and <2.5
Say 100 horses are in your sample, winning 25% of the time
40 are top horse system plays the won 40% of the time

Then sqrt(40 x .25 x (1-.25)= 3 = standard deviation. To get to the 1% confidence level, you need to win
(2.5 x 3) + (40 x .25) = 18
The number of actual winners is only (40 x .40)= 16.. not good enough for the 1% level
Testing for the 5% level
(2 x 3) + (40 x .25) =16 good enough for the 5% level
.
If I interpreted his work correctly, this analysis is in line with Butron P. Fabrican’s book “Horse Sense”
Chapter 8.

From my experience, if this holds up for 100 more systems plays, the results would be more reliable

[Augenj]

Not quite the information needed, rather

for all horses in the race, indicate incremental odds and win%. (odds ranges can be established later)
Break out system plays by odds , quantity and number of winners.
To start I would suggest using just your top horse system
If your system is not track specific combine all track totals.
This is the information required to calculate the standard deviation and confidence level.

For example, looking at all horses in the odds range >=2 and <2.5
Say 100 horses are in your sample, winning 25% of the time
40 are top horse system plays the won 40% of the time

Then sqrt(40 x .25 x (1-.25)= 3 = standard deviation. To get to the 1% confidence level, you need to win
(2.5 x 3) + (40 x .25) = 18
The number of actual winners is only (40 x .40)= 16.. not good enough for the 1% level
Testing for the 5% level
(2 x 3) + (40 x .25) =16 good enough for the 5% level
.
If I interpreted his work correctly, this analysis is in line with Butron P. Fabrican’s book “Horse Sense”
Chapter 8.

From my experience, if this holds up for 100 more systems plays, the results would be more reliable



Yeh, this doesn't do that.... and by the way, you're way over my head. :D

[formula_2002]

Yeh, this doesn't do that.... and by the way, you're way over my head. :D
No I'm not

just read my comments again. You probably have all the information required somewhere in your data

how many races did you review?
how many won.
how many of your system plays were that review
how many did they win?
all the above broken down the final odds
like this;

2/5 - 4/5
1/1 - 7/5
8/5 - 9/5
2/1
5/2
3/1 and up

[Augenj]

No I'm not

just read my comments again. You probably have all the information required somewhere in your data

how many races did you review?
how many won.
how many of your system plays were that review
how many did they win?
all the above broken down the final odds
like this;

2/5 - 4/5
1/1 - 7/5
8/5 - 9/5
2/1
5/2
3/1 and up
Well, all of this is listed in the grids I posted above except the odds breakdowns. My fifty years of experience in computers and horse racing has shown me that no matter how I filter the odds, the win percent is proportional to the odds and the ROI is the same (negative ROI) except for one category - odds on. That sometimes produces a break-even ROI, at least it did in the past.

I know what square root is because I use it to smooth earnings extremes from high purse races. I tried to use standard deviation at one time to exclude bad races from past performances for speed and finish. Over thousands of races I found that bad races should be included. They must be there for a reason. My stats are "boots on the ground" stats. You stick your cash into this one condition and out the other in comes your ROI for the current year.

This is my public version and doesn't show the other 800 or so compound factors that I have for my personal use. Your stats sound very professional but we do things differently. ;)

[bobphilo]

Well, all of this is listed in the grids I posted above except the odds breakdowns. My fifty years of experience in computers and horse racing has shown me that no matter how I filter the odds, the win percent is proportional to the odds and the ROI is the same (negative ROI) except for one category - odds on. That sometimes produces a break-even ROI, at least it did in the past.


Correct. Other studies have shown that horses generally win according to their odds. The exception, not known to many longshot players, is that short price favorites win at a higher rate than their odds predict and thus yield higher ROI, while longshots win at a lower rate than their odds and have a lower ROI.
An often overlooked factor in handicapping is psychological factors. Many who are chasing longshots often exaggerate high price horse's abilities. A kind of unconscious wishful thing.

[Augenj]

Correct. Other studies have shown that horses generally win according to their odds. The exception, not known to many longshot players, is that short price favorites win at a higher rate than their odds predict and thus yield higher ROI, while longshots win at a lower rate than their odds and have a lower ROI.
An often overlooked factor in handicapping is psychological factors. Many who are chasing longshots often exaggerate high price horse's abilities. A kind of unconscious wishful thing.
What? Somebody agrees with me? :D

[VigorsTheGrey]

Agreed, but over the years of scanning results charts I have noticed that higher odds horses do indeed finish 2nd spot, the place hole. And significant long shot often finish 2nd as well...so in my opinion, the same cannot be said for the place spot...