Predictive Factors for Contenders
 

I'm building a new database using Template 4 and hope it will help me identify contenders. My dirt sprint and route tabs contain enough races for a meaningful sample and I thought I would post a summary. My cut off was 80% or more winners in the top four. (BTW I have no maiden races in my database and I'm using the default pacelines with only a few exceptions.) This reinforces the predictive power of Rx1, Rx2 and Rx3 that others have already noted.

Good
 

stuff Mick. My question for you is once you have it whittled down to your top 4 what is your process for getting it down to your top 2?

bob

Great work Mick.


I see you took my advice to record factors in RDSS to get the winner in the top 4 85% of the time. Not only that once you know this it greatly reduces the amount of time to get the proper contenders which can be spent on analysis.


Accepting 64% or less only reduces the power to win. This is a tough game and every edge possible must be taken to succeed.
Again Mick terrific work.


Mitch44

Bob, I look for bets that don't require me to whittle it down to a top 2, like rolling doubles and pick 3s. I'll bet more on the more likely combinations but spread and cover with smaller amounts on the less likely combinations.

I spend a lot of time on the TPR and Primary screens trying to find separation between horses. Regarding paceline selection, I use the default, best of last three comparable, except with cheap claimers. For them, I'll change the paceline to the last race and then try to find some reason to go back to a previous line. Recency is crucial for those plodders. (Bill Lyster did us all a favor when he compiled FTL's guidelines.)

I tried matching but then a Sartin veteran who had worked with Jim Bradshaw told me how "The Hat" had done it and I realized that I just wasn't smart enough. That man didn't make a mark on his Daily Racing Form. He kept it all in his head! How is that possible?

Anyway, that's sort of how I do it. I'm sure I haven't told you anything you didn't know already and probably haven't done many times.

Very interesting stats Mick. Thanks for your hard work and for posting this info.
Tim

Bob:
I would say to separate by use of advance concepts such as : NTL, suspicious dropdowns, too high % Med, too low % Med, one run horses, false early horses etc.


Mitch44

Predictive Factors for Contenders - Updated
 

After a busy weekend of racing and beta-testing, I’ve added at least another hundred races to my “MasterTemplate4” database. Below are the updated statistics, which further support the predictive power of the Rx Factors.

I’ve added two columns: PL# and PP#. Those are the Profit Line ranking from Twinspires and the Prime Power ranking from BRIS. The Profit Line rankings are on the Rx screen. The Prime Power rankings were manually extracted from BRIS PPs and added to the database by hand. (Yes, it’s a pain.)

Regarding turf sprints and routes, I don’t have enough races yet to comfortably publish those stats. With turf sprints, the Rx factors look strong but it is very small sample. With turf routes, the best statistical factors are the positions at the stretch call and finish of the selected paceline. But again, it’s a small sample.

contenders
 

I have an Aqueduct only database of approx 100 races (much too small for definitive answers), but my figures are tending to reflect yours. RX2 is slightly better than RX3.

Thanks for the update Mick. Charlie thanks to you also for your Aqu update.
Tim

Hard work is rewarded. Maintain Mick & Charlie as there is a yellow brick road.
Nice work.
Mitch44

Thanks
 

Thank you Mick and Charlie

Mick. As I mentioned in another thread
In my big spreadsheet with 1774 races, I get my RX1,RX2 and RX3
just about the same as yours 82 82 and 83 percent

In my "FTL" spreadsheet Which is only 90 races
This sheet has no maiden races and only races from
GP, TAM, PARX, PENN, OP, FG, SA, and GG,

I bet everyone of these races,

My top 4 RX1, 2, and 3 are doing real well
The RX3 is 93%

Bill

Attachment 45915

The King of Picking Pacelines
 

Bill, that is an astonishing stat! In Memphis, Elvis is the King. At Sartin Downs, you're the "King of Picking Pacelines". :D

Wow 93% that's outstanding Bill.
Tim

Nice work Bill, great stats.


Mitch44

Great work Bill
 

Someday I hope to be as good as you at picking pacelines.

Topic for discussion: when collecting data in a parimutuel game, I would suggest that tote-board public betting choices need to be tabulated as well. Like they say in poker, it's not how many pots you win, it's how big the pots are when you win. Similarly, when betting horses, we are betting AGAINST the computer robots and the syndicates, both of whom receive larger rebates than any individual, and have access to making multiple wagers at the bell. Remember, if someone in the betting pool is receiving a larger rebate than you are, you are effectively betting at a higher take-out.

Because of this structure, the closing tote odds are very efficient estimates of a horse's probability of winning. In fact, many of the large syndicates include the tote odds in their formulas to estimate win probabilities. Remember, we only want to make +EV bets (positive expected value), so we need a horse's odds to exceed its probability of winning. In the case of 2-horse betting, the net odds after adjusting for the losing bet and the dutch weightings must exceed the cumulative win probability of the 2 horse "group".

These days, favorites are winning at a close to 40%, and the top 2 favorites at closes to 60%. Contender selection needs to include price as well as percentages.

Bottom line, calculating win percentage without including the tote odds gives results that are not statistically significant. Sorry to pour cold water on all the hard work and analysis, but the analysis needs to be expanded. Thanks. Thoughts?

no cold water
 

hello Rdiam (name please):)

You bring up a good subject, We need to look into the future and
make the tote board a big piece of our battles against the whales.

I'm not 100% sure of what you would like to see,

Would you be interested in the mutual prices of all the winners who were in my top 4 RX3 ? in my analysis

Thanks
Bill

Bill:

We met last summer at Saratoga.

A few thoughts:

1. For your sample, show what percentage the tote favorite won, the second favorite, third favorite, and fourth, with a cumulative percentage. This can help show whether your Rx3/paceline selection is any improvement on what the "public" is able to do on the races you bet.

2. Calculate the average odds of your sample of winners selected. Here is why this is important: to simplify, let's say you can pick a winner 50% of the time, but the average odds on your winners is only 4/5. That is -EV. Conversely, let's say, my selection only wins 25% of the time, but my average odds are 4-1. That is +EV. But on the spreadsheet right now, by not including average odds, your 50% looks better than my 25%.

Hope this helps.

Richard

Hi
 

Hey Richard :)

Yes of course, now I remember, Hey I finally took your advice and updated my profile pic,

I will get going on those calculations.

Bill

Hi, Richard.

I'm driving to Hot Springs this morning and will think about your suggestion on the way. It may require exporting the data to Access but ... well, we'll see. You've made a valid point. Thank you.

Bill, you look mighty sporty in that trilby.

Hey guys,

I am interested in getting opinions on what I think is an interesting predictive factor from my record keeping. I am always looking for ways to reduce the work load and spend more time matching the contenders that matter.

I have a database for two years of all the tracks I play. It includes only non-maiden races of 7 or more horses running on non-off surfaces. So for 2016/2017 a total of 10,000 races breaks down to about 4500 dirt sprints, 2500 dirt routes, 2500 turf routes, and 500 turf sprints.

Of the factors to the left on the entries screen that aren’t affected by paceline selection Composite Speed Rating or CSR averages out with the winner coming from the top 4, 70% of the time. Is this not fairly impressive? More specifically DS=73%, DR=72%, TR=65%, and TS=68%. Betting on the top CSR horse alone returns -13% ROI.

Now when you read the Matchup section you frequently come across The Hat telling Rich that he likes fast horses. He only bets on fast horses. Makes sense that the top 4 CSR horses are fast horses. APV and CR or CR+ don’t match this level of performance. I didn’t have PL at the time so I have no stats for that. Keep in mind that this excludes short fields. With fields of 6 or less included the number jumps to 75%.

The races were done for the purpose of the database with auto paceline best of last three comparable. With that the RX1 factor comes in at 60% top 4. Didn’t have RX2 and RX3. Nothing close to 85% that some of you have reported.

I just find it interesting that without doing any handicapping or paceline selection at all there is a way to be working with the winner 70% of the time when looking at top 4 CSR. Any of your large databases should reflect the same thing. I didn’t do anything special except keep the database.

Is there something here?

Pook

Pook :


No it's not impressive. CSR is a product of final tine or final speed. It doesn't tell how it was achieved. The match up is about how it was earned or achieved.


Strive for your winner in your top 4 Above 80% of the time . Have to run for a DR apt, later.


Hopes this helps.
Mitch4,

based on a comment by bill v a couple of weeks back, i rechecked my database. I found that 88% of my winners have a bl/bl score greater than 10, regardless of distance. My database consists only of AQU races and numbers approximately 110 races

Very nice work Charlie. Great info.


Mitch44

Follow up to Richard's posts
 

I've been mulling over Richard's thoughtful posts on this thread and decided to do a little work in Access combining Rx win percentages and win payouts. This chart deals solely with Rx3 and dirt sprints. While he wrote in terms of Expected Value (EV), I'm more accustomed with the $1 ROI. The lesson here is that you shouldn't blindly bet one of the top three Rx3 horses. Handicapping is still required. The fourth ranked Rx3 horse is a puzzle though. I scanned the payouts and with the exception of two big numbers, most of the payouts are healthy double digits.

Nice stats Mick. The prices are in line with what should be expected to happen when the winner performs to its tier level. You are correct that tier levels must not be blindly accepted. I, myself, routinely move horses up or down the tier levels based on handicapping factors not part of the Methodology. And remember that Doc always stated that your win bets should be your top 2 horses not the programs' top 2.
Tim

That's a lot of work Mick, eventually hard work pays dividends. Excellent point in that handicapping is still required.


Mitch44

Richard and Mick
 

Hi Richard and Mick

I'm sorry for the delay in getting this report done
I have been having some more health issues. Feeling better,

Here is my report

My model is up to 111 races ( actually bet)
Races from 2/24/18 to 3/28/18

These results are from.
All race distances and surfaces
Non-maiden only
Field of 6 or more
Tracks Aqu, Parx, Lrl, OP, GP, Tam, FG, SA, GG
Contender and paceline selection method -FTL guidelines.

In all 3 categories, RX 3 outperforms the public
in top 4 win% and odds.

Attachment 45942

Bill,

First, I want to commend you for the amount of work you put into your statistics. And second I want to tell you how impressed I am at your ability to pick pacelines. Those odds in the third box are amazing. Wow. Betting a single horse, you only have to hit about one in four to make a profit. In my dreams ...

Glad that you are feeling better Bill:)

I believe that I asked this question to RMath years ago....

From the RX3 rankings....Can someone check how many of the #3rd rank and #4th rank RX3 horses had the top ranking on Early Pace (First Fraction Only)

Great stats guys....

Mick:


In doing an analysis of your stats, specifically posts # 1,7, & 26 of this thread I completely concur with your findings. They meet my findings of research that the top 4 of Profit contain the winner about 77% of the time.


I undertook that well over 10 years ago in reference to the top 4 of Profit. Its been so long I don't really recall when that was undertaken. Mine was conducted without regard to class of the race etc. Later refinements proved that the % of top 4 Profit can be improved upon with selected types of races.


Your study also validates Sartin's finding that routes are more predictive than sprints. The reason for this is that a horse has time to overcome trouble in a route which it can't in a sprint unless really superior on class. A good rounded number for a sprint is about 6 lengths per second. (Actually about 5.80 lengths). Therefore if it runs up on the heels and has to take up or any other trouble it'll lose about 6 lengths minimum, more dependent on the amount of time and also where the interference takes place.


Additionally your study used the automated pace line selection as in the program. If a member were to verify the pace lines using Sartin's latest guidelines and Preceptor the percentage in the top 4 could also be raised. It reasonable for a member to reach an 85% in the top 4 RX 3. And if he or she were a two horse bettor it'll serve them well.


Great stuff Mick, excellent work. Your work is like steering control which is much better than cruise control which will steer you in your future endeavors.


Terrific posts,



Mitch44

Bill:


You posted;" -FTL guidelines." Is that a minus meaning without his guidelines or with his guidelines?
Thanks,
Mitch44

Updated Rx Factor Chart
 

I've added more sprints and included routes.

Edit
 

hi Mitch

That was kind a misprint. what I was hoping to point out was, I am doing very well using FTL's guidelines, I wish I could be doing better,
Right now my major weakness are using horses with "0" races and moving to
plus races, when the 0 line has no excuses.
Another weakness, so far is reading the conditions carefully,
Richie P's most recent Match Up race from Laurel is a good example of this
weakness in my game.

Blessings
Bill

Hi Bill

You are doing just fine Bill
One question Using Horses in the"0" races and moving to
plus races, when the 0 line has no excuses. Would you explain further

Thanks for the clarification Bill. One has to start with the conditions to pick proper pace lines noting a minimum of surface
& distance but I also note the other conditions which aide in separating contenders. I just skip over a bad line and use what represents the horse by using the best of the last 3 year olds are notorious for throwing in a clinker for absolutely no reason and run back with a big one.


Hope your doing well and better.


Mitch44

I took a look at my last 100 races to see how how RX3 performed. Contender selection was the top 5 PL horses. Paceline selection was the best of the last 3 perceptor. I did change the program line when I did not agree with the line selected. All races were for 3 &up dirt and turf. The tracks involved were Parx,Tampa Bay, OP, and MVR. Here are the results


#1 31 wins 31% AWP $5.80
#2 17 wins 17% AWP $10.49
#3 12 wins %12 AWP $10.25
#4 15 wins 15% AWP $14.00
#5 10 wins 10% Awp $25.94
For those who might feel that 100 races maybe too small a sample I will state that the 85% winners in the PL top5 has been constant over the last 5 yrs that I've tracked it. No large data base was involved but I did conducted several random surveys a year. For what it's worth, since nothing wins 100%, I'm confident that using the top 5 PL horses is a very solid contender selection process.
Tim

hello Lonespeed

Here are some stats on my top 4 F1 horses. I used 3 criteria
Pace of Race velocity F1, Pace of Horse velocity F1
and CBL F1 column BK on the template.
I have 58 sprints and 43 routes

Attachment 45946
Attachment 45945

Jimmy Bradshaw said;" You always have a bad horse in the mix." So even when getting it down to 3 there is a non winner in there. This is where experience comes in and
also is a reason why a two horse bettor has an edge.

Getting the winner in the top 4 a high percentage of the time puts the player in the game and on the field and not as a spectator in the stands. They should demand a top 4 of 80 to 85 % to be a major player.


Enjoy the crowd and take their money.
Mitch44