Tampa Bay takeout rated an overall B...Why?

[InsideThePylons-MW]

Let's try this again.....maybe somebody from HANA will respond.

Tampa's takeout on trifectas and superfectas is a larcenous 25.9%. According to the HANA rating system rules, the tracks in the bottom "quintile" range of all tracks should receive an F grade. Tampa is definitely in the bottom 20% on tris and supers yet they received a "C" grade.......WHY?

Tampa's takeout on WPS is 17.5% which is easily worse than the industry average and rates them in the bottom half of all tracks.

Quickly looking at the charts from their last race day (3/31), almost two thirds (2/3) of their total handle was bet at larcenous and/or higher than avg. takeout rates.

1. How can a track which has almost 2/3 of it's handle being bet at larcenous and/or higher than avg takeout rates be given an overall takeout grade of B?

2. Why was a grade of C given to Tampa's tri and super takeout rates when per your rules they obviously should have been graded an F?

3. Why does HANA keep having a love affair with a track that robs it bettors on 25.9% takeout rates and charges the highest independent track signal fee in the industry?

[Robert Goren]

They grade on the curve.;)

[chickenhead]

they are not actually graded on the curve -- ITP is asking why they're not.

The "Quintiles" are obviously not set up to get an even distribution of tracks, the Quintiles are based on takeout rate, i.e. the spread in takeout rates are broken up into quintiles, and the track distribution within them falls where it may.

The Superfecta takeout spread, divided into quintiles, roughly, is:

19% - 20% - 4
21% - 23% - 3
24% - 26% - 2
27% - 28% - 1
29% - 30% - 0

The downside of this is what ITP is pointing out, you obviously don't get an equal distribution of tracks inside those buckets. You can end up with a majority of tracks getting better scores because of a few really bad tracks. This is what happened with Tris and Supers it looks to me, the distribution is skewed upwards because of that. Something non-linear would probably work better.

I think the difficulty in grading on the curve is that takeouts rates are bunched, for Tris it looks like almost a majority have a 25% takeout. You can't really curve that, or get a quintile worth of tracks into each bucket. You would just have to decide what grades you want to hand out.

For the weighting different pools according to their handle, to reflect effective takeout, I know that's been talked about because I've suggested it, I think its a good idea.

[chickenhead]

I should add, that even with the above:

19% - 20% - 4
21% - 23% - 3
24% - 26% - 2
27% - 28% - 1
29% - 30% - 0

The windows are biggest for B and C, so obviously its going to be skewed towards B's and C's. Better probably to divide it equally into true quintiles and have 11/5 = 2.2% per, which would have given:

19% - 21.19% - 4
21.2% - 23.39% - 3
23.4% - 25.59% - 2
25.6% - 27.79% - 1
27.8% - 30% - 0

But for pools with outliers, say some track with a 12% bet and some track with a 30% bet, and everyone else grouped somewhere in the middle, that method wouldn't work very well, either.

Like I said in some earlier post -- I see plenty of things non-perfect about how these grades are done...but its also not a small thing to say how it should be done to make it better. Ideally you want the same method you are applying to all the pools, that works equally well for all of them. But the distributions and groupings are totally different in each of the pools.

If I knew what that perfect method was I'd suggest it, but I don't. If anyone else does, please do.

[chickenhead]

Here is my suggestion for improving the grading of takeouts, it is essentially a hybrid approach.

I do not like the way it is done currently, I do not like having different ranges depending on bet type. But I also do not think it is appropriate to give every track an F based on optimal rates, even tho they pretty much all legitamately deserve an F. I'm looking for a hybrid of the 2.

We roughly have a takeout spread of 12% to 30% across all bet types. 12% is probably about as close as we can expect for a bottom to takeout rates anytime soon, so even though it is only out their sparingly, I think its an ok bottom to use for the first part of the calculation.

I want it to be harder to get an A than a B, and harder to get a B than a C, and harder to get a D than an F. So I believe the windows for each of those grades should be narrower at the top than at the bottom.

12% to 30% gives an 18 point spread. If we divided it equally, there would be (5) 3.6% wide windows for each grades. I suggest instead the windows be of gradually increasing size, such as:

A: 1.8
B: 2.7
C: 3.6
D: 4.5
F: 5.4

(this relates to 10% of the spread for A, 15% for B, 20% for C, etc etc)

We don't really need upper and lower limits for a grading system, so the final result of this would be:

A 0 13.79
B 13.8 16.49
C 16.5 20.09
D 20.1 24.59
F 24.6 Infinity

Is this a much more strict system, yes. You'd have to be below 20% essentially to get a C. Anything above that is pretty egregious, so that aligns with my views just fine anyway.

Take the above grade for each individual pool, and weight them together using pool size to get the overall takeout grade -- and I think you probably have a nice improvement over the current system. I'm sure it would still result in something wacky, something else would be disturbing about the results. But at this low level view, it would make more sense (to me at least).

[InsideThePylons-MW]

If I knew what that perfect method was I'd suggest it, but I don't. If anyone else does, please do.

I don't know what is perfect but I do know what to add to it.......

One ounce of common sense.

A horseplayer's organization should not give a C grade to the 64th worst track out of 71 that has 26% takeout.

A horseplayer's organization should not give an overall takeout grade of B (which led to a rank of 3rd on best tracks list) to a track that has 25.9% takeout rates on tris/supers and a higher than avg 17.5% WPS takeout rate which combined accounts for 2/3 of their total handle.

[Horseplayersbet.com]

I don't know what is perfect but I do know what to add to it.......

One ounce of common sense.

A horseplayer's organization should not give a C grade to the 64th worst track out of 71 that has 26% takeout.

A horseplayer's organization should not give an overall takeout grade of B (which led to a rank of 3rd on best tracks list) to a track that has 25.9% takeout rates on tris/supers and a higher than avg 17.5% WPS takeout rate which combined accounts for 2/3 of their total handle.
What formula should be used to rate takeout? Personally, I think average takeout is the only way to go. But if you have any suggestions here, lets see them. It has to be done by formula, or the takeouts just might as well be posted and not have any weight whatsoever in the final ratings.

[InsideThePylons-MW]

or the takeouts just might as well be posted and not have any weight whatsoever in the final ratings.

If 25.9% on tris/supers and 17.5% on WPS is going to equate to an overall B and 3rd ranked track by horseplayers......I vote for takeout having no weight whatsoever.

I'd rather have takeout mean nothing than giving a double thumbs up with a B grade and 3rd best ranking to those rates.

[InsideThePylons-MW]

What formula should be used to rate takeout?

Not the one you guys are using.

[chickenhead]

that's constructive.

[InsideThePylons-MW]

that's constructive.

I've been trying to be constructive. All of my posts have a suggestion of what not to do.....that is the best advice to figure out what to actually do.

We have a bunch of smart people who are doing the most important thing they do all year for HANA and nobody could figure out that 26% takeout rates which rate tracks in the bottom quintile should not be given a C grade? When Dean posted here about the formula being online, within a second of seeing it, I immediately said that this can't be possible.

Why can't the formula involve #'s and common sense?

I could look at takeout rates coupled with individual importance of each one pertaining to that tracks handle and give out a letter grade that would be realistic to how they rate.

[chickenhead]

I agree with you that your most constructive point is factoring in the pool size that the takeout is employed on. I think that would change the relative rankings more than most people realize.

[InsideThePylons-MW]

that's constructive.

I'm not a computer/math/formula/data guy. I'm 99% common sense and it's tough for me to compete with all of you because that is your specialty.

I can't come up with sophisticated ratios and stuff like you came up with in your posts above.

It would take me a year to do what you did in minutes. But I can obviously look at something and know through common sense it's wrong.

You thought on a glance that Tampa was a D or F but when I told you it was a C, you knew it wrong too.

So that is why I can't be too constructive on what you guys are needing me to be constructive about other than what I've said should be obvious.

[chickenhead]

look, for my own part --

I am friendly with the people, and interested in HANA. I talk to them all the time. And I was given every opportunity to offer HANA input like the above, for the past 2 years, and I didn't. I don't know if it would have mattered, but it possibly would have. I am pretty lazy, and busy doing real life things, and I frankly never dug down into the takeout grading details. I was just happy that Remington and Will Rogers Downs and whoever else weren't ranked so highly this year. You are right, I assumed a 26%+ takeout would have gotten a D or an F.

So for my own part -- being that I agree with you, being that I don't think a 26%+ takeout should get a C yet I didn't pay enough attention to even notice it and throw flags -- I apologize to HANA.

[InsideThePylons-MW]

CH,

Nobody is blaming you for anything.

I'm not blaming HANA or anyone there specifically either. I'm blaming the process and how the process let this happen.

I haven't got a reply from HANA so maybe I'm wrong and HANA doesn't care that this happened. If that's true, then I blame HANA.

Did the Officers/Advisory Board look at the rating systems and see 26% as a C and shake their heads in agreement.....Did they not look at the ratings system....Did they look but not see it? This is stuff that needs to be looked at in my opinion.

[DeanT]

I haven't got a reply from HANA so maybe I'm wrong and HANA doesn't care that this happened.

Wednesday at 12:37PM in response to the letter grade questions from a couple posters.

http://www.paceadvantage.com/forum/showpost.php?p=865782&postcount=22

"We talked about not even using letter grades, or only using letter grades and possibly tweaking that. A couple of you (ITP is another one and one or two more via email) who thinks we should change it for next year."

"I think we probably will have a change for you guys next time."

"If you have any more feedback on grading or whatever that you think might make things better, or to get horseplayers' points across in a better fashion, shoot us a mail at info @ hanaweb.org."

I have answered upwards of 250 emails, posts, media inquiries and phone calls the past four days regarding the HANA list and metrics. I answered your question on Wednesday.

Regards,

Dean

[InsideThePylons-MW]

Dean,

I read that the other day but it really wasn't in response to what I was talking about.

I'll admit it, I'm steamed silly at the ratings....but I guess that's good because it shows how much I care.

[DeanT]

There are probably 15 pages of items on the HANA work board with suggestions and tweaks and concerns for the ratings system moving forward. Your concerns and suggestions are there and will be enacted upon.

We have ten months to make these better, and just like last year when the same discussions happened, we will do our best. Looking backwards is not an option, making them better next time is the only concern.

That's about all I can tell you. I am done with the ratings for a bit, so I am going to enjoy the weekend racing now.

Regards,

Dean

[InsideThePylons-MW]

I am done with the ratings for a bit, so I am going to enjoy the weekend racing now.

Enjoy your weekend Dean.

I'm done causing trouble on the ratings. ;)

[ronsmac]

Anything over 12 % is too high, and quite frankly maybe even 10 %.

[turfnsport]

I don't know what is perfect but I do know what to add to it.......

One ounce of common sense.



That's a good equation from a guy who admits he is not a "computer/math/formula/data guy". :ThmbUp: