Introduction and a rookie question

[Stillriledup]

Originally Posted by Poindexter

While I understand your point of trying to stay away form connections that will be bet no matter what horse is running.......you cannot ignore the fact that who the trainer is and who the jockey is both are paramount to determining how much chance each horse has. When I assign a line to a horse, if I see Prat or Bejerano (same goes for your top guns at other circuits) up I might make a horse 15-1 or even 12-1 while I might make the same horse 30-1 if I see guys like Quinonez or David Flores up. For one they ride that much better, also why are they riding to begin with? Same with trainers, you cannot give the same respect to a 4% trainer as you do a 20% trainer. Your goal as a player is to estimate how much chance a horse has(whether making an odds line or using whatever grading system you use) and not giving extra credit to top connections makes this job impossible. If you underestimate the connections you are going to underestimate how much chance their horse has and thus overestimate how much chance every other horse has.

By the way every factor is factored into the price (some more than others), as a player you have to do a better job than the betting public of weighing those handicapping factors and finding value.

Good thoughts. Your Flores vs bejarano example, to me, when I see Flores I automatically question if the horse is even live. Flores might be a bad example for what I was saying because he's SO bad that he's hard to bet even if you figure out that the horse is somehow live, I was generally talking about someone fully competent like the 6th or 7th leading jock vs the top jock and how you can get a much better price with them.

Another factor for me is that certain jocks 'fit' certain styles of horses so I may upgrade a horse on that factor.

I was just giving general thoughts to try n avoid betting on top humans since there are plenty of live horses to bet on that aren't trained by top connections.

[whodoyoulike]

Originally Posted by cnollfan

Average earnings per start is not as effective a tool as it used to be for two reasons. There are more state-bred races than before. These races have inflated purses relative to the horses' talents. And there are more racinos than before, also offering inflated purses.

The responses I've seen here makes some sense but, the reason I switched from average to evaluating total earnings is because I'm a big believer of the cycles of things. There are seasonal cycles (winter, summer etc.), business cycles (financial sectors go up and down) and in horse racing the owners, jockeys, trainers and horses will also go thru cycles.

An example I can think of is Bob Baffert. He has been winning for as long as I can remember (since the 90's). He just may be on a 20 year winning streak of a 30 year winning cycle.

Getting back to the average earnings per start (e/p/s) of a horse becomes misleading when the horse is on a downward or improving cycle which will report a higher or lower trending amount. So, switching to total earnings I can make a better evaluation of this trend.

[Tom]

Originally Posted by thaskalos

Yes...but I didn't see earnings-per-start to be an effective class barometer. Did I miss something?

I don't know - I remember it was tested so I suggested it.
But do you want it to be a class barometer or a win barometer.

At NYRA tracks for the last few months, all races, #1 EPS is in 41% of all exactas, and by itself, it shows 25% winners and and ,91 roi, out-performing the top speed figure.

I'll run a larger study tomorrow, but time for my ice cream and bed.

[Mulerider]

Originally Posted by Tom

At NYRA tracks for the last few months, all races, #1 EPS is in 41% of all exactas, and by itself, it shows 25% winners and and ,91 roi, out-performing the top speed figure.

Tom, is that NYRA e/p/s stat lifetime or current year? I've been using current year when given a decent number of races, thinking that it might give a better picture of current class. As a part of my process I also consider Brisnet's Current Class numbers, though not as a specific spreadsheet category in my system. I have considered doing that, but I have no idea how Brisnet comes up with their class ratings. If EPS is part of their formula, then I'd be dealing with a redundancy since I'm using it already.

[Tom]

Lifetime.

[Red Knave]

Originally Posted by thaskalos

Yes...but I didn't see earnings-per-start to be an effective class barometer. Did I miss something?

What is your 'barometer' for Class?
The highest ranked APV horse does have an Impact Value over twice as high as the horses in the Rear Half of the field. Does that show Class?
P&P has a couple of paragraphs about what Dave calls Earnings Box Class Rank. It is a composite of EPS, APV and Quirin's MRCT. This factor has higher highs and lower lows than APV alone but seems to say the same thing.
Of course, neither factor has a positive ROI but I think you already knew that.

[thaskalos]

Originally Posted by Red Knave

What is your 'barometer' for Class?
The highest ranked APV horse does have an Impact Value over twice as high as the horses in the Rear Half of the field. Does that show Class?
P&P has a couple of paragraphs about what Dave calls Earnings Box Class Rank. It is a composite of EPS, APV and Quirin's MRCT. This factor has higher highs and lower lows than APV alone but seems to say the same thing.
Of course, neither factor has a positive ROI but I think you already knew that.

To me...class is the level at which the horse figures to perform best TODAY. Yes...people talk about the concept of "back class", and point to horses who are able to ascend to their former glory after recovering from their physical ailments...but these are the rare exceptions, not the rule. A big race for a fat purse can distort a horse's "average earnings-per-start" for the entirety of the horse's racing career...and that's an imprecision that I can't tolerate.

You look at the bottom of his past performances, and you see that the horse had been beating allowance horses a year ago. Now, he seems slower than the bottom claimers that he is asked to face today...but his average earnings-per-start TOWER over those of the rest of the field. Is this the "class of the race"? I think not...

[raybo]

Whew! Here again we're in a discussion about a single factor, EPS. Like the single factor "SR", alone it means very little regarding an individual race. It's in the "combining" with other factors and ratings that it has merit, and in how you filter past performances regarding that EPS calculation.

And finally, it is about, or should be about, long term results, not individual race results. Any factor or rating under the sun can be terribly wrong in individual races.

[thaskalos]

Originally Posted by raybo

Whew! Here again we're in a discussion about a single factor, EPS. Like the single factor "SR", alone it means very little regarding an individual race. It's in the "combining" with other factors and ratings that it has merit, and in how you filter past performances regarding that EPS calculation.

And finally, it is about, or should be about, long term results, not individual race results. Any factor or rating under the sun can be terribly wrong in individual races.

I am not telling you not to use the EPS in your own calculations, Raybo...you are free to do whatever you want. Red Knave asked me a question, and I am telling him why I don't use that metric.

I use speed ratings because they provide a clue to the puzzle that I can't get by another means. Whatever the EPS can provide, I can find in a more accurate form someplace else. But, as I said, this is strictly my OWN opinion...and I wouldn't presume to enforce it on someone else.

[raybo]

Originally Posted by thaskalos

I am not telling you not to use the EPS in your own calculations, Raybo...you are free to do whatever you want. Red Knave asked me a question, and I am telling him why I don't use that metric.

I use speed ratings because they provide a clue to the puzzle that I can't get by another means. Whatever the EPS can provide, I can find in a more accurate form someplace else. But, as I said, this is strictly my OWN opinion...and I wouldn't presume to enforce it on someone else.

I never said anything about you telling anyone not to use EPS. That would be very presumptive indeed. My post was meant for a general audience. You, of all people, know that single factors, in isolation, mean almost nothing in the long term. EPS is a single factor calculation, and as was stated, could be very misleading if calculated generically, like lifetime earnings divided by lifetime starts, or even recent earnings divided by recent starts. For example, the big hit that was mentioned, that race type/distance/surface/date, etc., could very well not apply to today's race at all, and would certainly skew the final rating. So, some degree of filtering would be a good start, in obtaining something legitimately significant. I won't go further, than to say that, the filtering and occasional adjustments to the mean of past race results data, just like with outliers in the calculation of ROI, is extremely important if you are looking for long term positive results.

[thaskalos]

Originally Posted by raybo

I never said anything about you telling anyone not to use EPS. That would be very presumptive indeed. My post was meant for a general audience. You, of all people, know that single factors, in isolation, mean almost nothing in the long term. EPS is a single factor calculation, and as was stated, could be very misleading if calculated generically, like lifetime earnings divided by lifetime starts, or even recent earnings divided by recent starts. For example, the big hit that was mentioned, that race type/distance/surface/date, etc., could very well not apply to today's race at all, and would certainly skew the final rating. So, some degree of filtering would be a good start, in obtaining something legitimately significant. I won't go further, than to say that, the filtering and occasional adjustments to the mean of past race results data, just like with outliers in the calculation of ROI, is extremely important if you are looking for long term positive results.

"Earnings-per-start" is a well-known handicapping metric which has been defined in a variety of handicapping publications. It's what we end up with when we divide the horse's earnings by the number of races that the horse has run, either lifetime...or more recently. When you add in the "filtering" and the "occasional adjustments" that you are talking about here, then you are indeed refining the numbers...but you are no longer talking about "earnings-per-start".

When Mulerider, Cnollfan, Tom, Overlay, Red Knave, and I, talk about "earnings-per-start"...we mean the metric in its more widely known definition. You are talking about something else entirely. Yours is a better metric, no doubt...but it needs a different name. The name "earnings-per-start" is already taken.

[raybo]

Originally Posted by thaskalos

"Earnings-per-start" is a well-known handicapping metric which has been defined in a variety of handicapping publications. It's what we end up with when we divide the horse's earnings by the number of races that the horse has run, either lifetime...or more recently. When you add in the "filtering" and the "occasional adjustments" that you are talking about here, then you are indeed refining the numbers...but you are no longer talking about "earnings-per-start".

When Mulerider, Cnollfan, Tom, Overlay, Red Knave, and I, talk about "earnings-per-start"...we mean the metric in its more widely known definition. You are talking about something else entirely. Yours is a better metric, no doubt...but it needs a different name. The name "earnings-per-start" is already taken.

No, I'm not talking about typical earnings per start. I'm talking about filtered and adjusted earnings per start. Generic earning per start are no better than using raw times to categorize/rank horses' performances.

[whodoyoulike]

Originally Posted by raybo

No, I'm not talking about typical earnings per start. I'm talking about filtered and adjusted earnings per start. Generic earning per start are no better than using raw times to categorize/rank horses' performances.

What does filtered and adjusted eps mean?

[raybo]

Originally Posted by whodoyoulike

What does filtered and adjusted eps mean?

I filter by route, sprint, dirt, and turf (dependent on what today's race is), and I only go back 10 races or 1 year, whichever comes first (I don't use the raw earnings/starts boxes at all). I also occasionally reduce large individual earnings when it is apparent that they will adversely skew the calculations to the point that they are obviously inaccurate.

[whodoyoulike]

Originally Posted by raybo

I filter by route, sprint, dirt, and turf (dependent on what today's race is), and I only go back 10 races or 1 year, whichever comes first (I don't use the raw earnings/starts boxes at all). I also occasionally reduce large individual earnings when it is apparent that they will adversely skew the calculations to the point that they are obviously inaccurate.

Thanks, I guess I was using the BRIS earnings summary of the breakdown of earnings (fst, off etc.) without realizing it until you just pointed it out.