Composite Speed Rating - CSR

[Ted Craven]

I just wanted to comment that CSR is not, in my view, some 'magical' number or the 'new' recommended approach to things. There are so many tools, so many consistently applicable approaches possible with RDSS' tools that do not stop working because of some peoples' focus one one particularly interesting method.

As has been discussed (perhaps not enough), CSR is particularly sensitive to the impact of the last race Adjusted Speed Rating. If the last race final time (or worse, the last 2 races) was poor for whatever reason, the resulting CSR will be lower ranked than if you thought that race (or races) could be excused.

In this discussion race, the Winner #9 had 2 last races during which it finished beaten double lengths. It is a lightly raced horse, who arguably ran against higher condition races those last 2, then took a break for freshening. It was the highest APV and CR horse. It was clearly 2nd in the tote odds. If you believe either of those last 2 races can be excused, it's 'effective' CSR will be higher than shown.

Further, consider the #8 horse, ranked 4th on CSR. Given its recent performance, if you do not consider it a Win contender today (as I would not), then the #9 horse's 6th CSR rank again will be better (if indeed, you consider the #9 a win contender on any consistent set of principles).



CSR does not take into account the above issues (judgements about primary contention, surface suitability, trouble or other excuses, timeframe). You can do that yourself with some hands-on analysis.

CSR is as much useful in a 'longitudinal' sense, over a group of races, as it is in one specific race (as with ANY factor). The stats published by rmath and kahunab or others tell a larger picture of results gained by CSR and VDC and perhaps other combined factors, and one interesting thing is that they can be applied without too much (or even NO) subjective assessment. BUT - those published stats do not include ROI measurements (you may get a lot of no-bet races, or no overlay situations if you can't separate the final contenders).

I presume in many of such cases, applying other tools at our disposal - including a matchup analysis, proper contender identification, judicious forgiveness of recent poor races, etc - you can start out with a contender set indicated by CSR Top 5, validate those further by VDC and come up with a solid set of 3 win contenders from which to choose 1 or 2 to see if bettable net odds are offered.

As it stood in this race, 'out of the box' Best of Last 3 comparable PSS ranks the #9 winner 3rd on BL/BL, and with top APV, CR, 2nd Early NewPace, 2-1 odds along with 4/5 on the top ranked BL/BL #4 horse results in a potential pass for Win. A 3 horse Exacta box on the top 3 (which nets $49.40 for the 9-3-4 finish) looks like a possibility due to the 7-1 odds on the #3.



All in all, just saying that either you take your race by race lumps by using CSR uncritically and trust the longer term stats, or you use it as a further 'guideline' to the subset of horses which are likely true contenders and apply your other tools in the current race.

FWIW.

Ted

[Bill V.]

Thank you Richard and Jim and Ted

Now I am clear with my first question

Horse #9 Is Not a top 5 CSR In a race that fits Rmath's guidelines

Now may I move on to a series of questions

Ted's post is perfect timing as I now move into question #2
which in have I broken into a 2 part question.

A. Are you stats based on "checking" races after the fact.?

Here is what I wonder,

Once again lets look at how you work a race .

You open race 8 on Monday from Parx
do you note the top 5 CSR and then handicap the race ?


B. Do you only look for pace lines of those 5 CSR horses ?

I wait patiently for your answers to question 2
part A. and B.

[Because I Can Jim]

Well put Ted.

I can see your points and can't agree with you enough.

Jim

[Segwin]

Originally Posted by rmath

Bill, the answer to both of your ?s is yes.
1) the 9 horse would not be a contender based on CSR ratings.
2) I have posted my guidelines before,but will post them again so you will not have to search for them.
GUIDELINES:
Only races with 8 or more betting interests are considered, 3yr olds & up.
No maidens or MSW races,
Only races from 5.5f to 8.5f were used in my testing. The reason for this is because these are the most often run distances at most tracks.

I am not saying 5f races and races over 8.5f are not playable (because they are) but that I did not include them in my stats because I do not see that many of them.
Races for 2yr. olds are excluded because most of the horses have run less than 4 times and therefor their CSR numbers are questionable.
The same holds true for maiden races.

I personally like races where all the entries have at least 4 races, because then the CSR numbers hold up better.
One thing I have noticed is that in races that I have been beaten it is usually ( about 3-4% of the time ) by a 6th or 7th ranked CSR horse.

I hope this helps answer your questions.
If not feel free to ask again.
I am always willing to help anyone who asks.
Rmath

Thank you Richard. One last question, do you make any changes to the standard configuration?

[Mark]

You can see with your eyes what this rating tells you. It is worthless and horribly flawed in its construction. Just take a look at the two horses' PPs. Why would you even consider two 5f turf sprints as germane to the subject. Look at the two horses maiden victories. Look at the early fractions in those races. Look at Trackmaster's variants for those two races. Whose the faster horse?

[Bill V.]

Ted wrote this

I just wanted to comment that CSR is not, in my view, some 'magical' number or the 'new' recommended approach to things. There are so many tools, so many consistently applicable approaches possible with RDSS' tools that do not stop working because of some peoples' focus one one particularly interesting method.

I find this point made by Ted to be important. The CSR is easy to rate and test and chart, because the calculations are based on a group of pace lines.
No decision making or analysis is needed. You just look at the rankings.

I suppose Ted used the Adjusted Speed Rating because its the staple of Doc's guidelines on pace line selection.

The speed rating has always been part of the methodology.
"The best of the last 3 speed adjusted speed ratings at a comparable distance and surface" Even back in the Energy /KGEN days
the instructions were to use a paceline with a speed rating over 80
or within 2 of the mean of the speed ratings of the contenders

Anyway What would have happened if Ted use another staple of the methodology Total Energy

Would study's and test of races be just as conclusive as the CSR test ?
Is the CSR any different than what a Composite total energy rating would be?

Here is race 8 using the top 4 Total Energy figures for each horses last 4 usable lines _ I just used basic math but my point is that Ted could have use Total Energy or TPR and probably got similar results as a speed rating calculation,


Bill's Composite Total Energy Rating

[Mark]

Bill, I am in complete agreement with the implications of your questions. The CSR is highly flawed and to handicap a race by using it as your contender selection method as we did in the old days with APV is a very questionable approach. What it rewards is high adjusted speed ratings from the last 4 "usable" races. That is not handicapping and what you will get is heavily bet horses. At least with APV you have the influence of historical purse values that the horse competed in and won purse money.
Here's the perfect example: today is a 8.2 mile route. Horse A has two races since returning from a 250 day layoff. His last two races are sprints in which he ran close to the lead before fading beaten 12 lengths and another in which he showed a 5 length stretch gain but finished beaten 8 lengths. His two races prior to the layoff where poor route races, indicating the horse was hurting and needed time away to recover. This horse has been classically prepped for a top effort yet his CSR is in the dumpster. He wins easily at 10/1.
We can't check our brain at the door when we open the handicapping program or racing form. Prices on obvious horses today are hammered down. In the early 90s I played the entire Santa Anita winter meet and can count on my fingers and toes the number of horses at 6/5. I remember Farma Way shipping out for the Santa Anita Handicap when he was at the top of his game and he only went to post at 6/5.
Look at the Beulah and Finger Lakes cards yesterday in some of the worst races carded in North America, absolute plow horses. 3/5, 4/5, even money and I am sure all these horses had top CSR #s. These type of standout horses in cheap races so skew the statistics that it makes them unusable in my opinion.
Each horse has a set of past performances, examine them top to bottom and bottom to top. Find the good races, see how the horse likes to run, what early pace can he succeed against? Look at his pattern of activity and the efforts of his trainer to coax an improved performance previously and today. No esoteric factor can do what your eyes and brain can.
Don't use questionable factors because they are there. You are competing against big money syndicates that spare no software programming expense. Most of us don't get the 7% - 9% rebates on every dollar they wager. You have to find horses that computer algorithms can't.
One Man's Opinion

[rmath]

A. no my stats are based on races that I ran before they actually ran.
My question to you would be: What is to be gained by running the races after the results are in?
?B: No, the program is set to pick the Best of last 3 Perceptor. When I check the computers line choices I may adjust a line here and there if I see a reason to change a line.
I go to BLBL screen and eliminate horses according to Docs guidelines using Primary Line scores.

Ted did a great job in his post.
My goal was not to change how anyone uses RDSS: BUT I was looking for a way to simply get the most likely winner in the final 4 or 5 contenders.
If others think they have a better way than I use then PLEASE share it with all of us.
I knew from the start of this project that I would not get the winner in my final contenders 100% of the time, but I have shown how I do get a very high percentage of winners in my final contenders.
Yes I get a lot of low priced winners in my contenders and so will everyone else that ends up with 85-90% of the winners in their top 4 or 5.
Just because you have the winner in your final contenders doesNOT mean you have to play that race.

[rmath]

I am almost sorry that I ever posted my results. My intentions were to only share my findings so others could profit from them.
I never thought that anyone would ridicule me or the results that I posted.

Yes an exception can be found to contradict my findings. But I never said that this is the only way that should be used.
There are races in my study like the one Bill posted , I lost those types of races. But most races are not like that one.
God gave each of us a brain to use , so when a race like this comes up look a little deeper. No one ever said not to think and only believe only what the numbers say.

[Lt1]

Well said Rich. Thank you for sharing you work with us. As you stated nobody is forced to use your selection process. Folks should be grateful to each person who shares their insights with the rest of the group.