Tentative Derby EL 2.0 Contenders, plus Reasoning

[Ted Craven]

I corrected a few things and re-ran the KY Derby for EL 2.0 purposes. Here's what I got:

22074

Based on the TrackMaster numbers shown - and the (default) rules we have to work with, does anyone think the 2 Early and 2 Late selections are incorrect? If not, I'll post this publicly. At this point, just trying to get the software to perform consistently!

Reasoning:

In this race, the POR is the 106 of Uncle Mo, which is NOT 12 SR points higher than the next highest Low (95 of Comma to the Top). But using 106 as POR results in NO other horses Hi SR qualifying. So we take the next highest Hi SR of Mucho Macho Man (MMM) (102) and this gives us 2 Early contenders.

OR - we could say that Uncle Mo's 11 SR point superiority over next best Low SR - Comma to the Top (95) is close enough to 12 points, and then use the Hi SR of the #1 ranked E% horse (Comma to The Top's 100) and thus make the POR 100. This qualifies a bunch more Early horses, but the same 2 Earlies are still the top 2 ranked Hi: Uncle Mo and MMM. Am I right?


I appreciate that other handicapping may come up with other contenders. Just haven't got that far yet! :o

Thanks,

Ted

[stormbringer]

thanks ! you are the best will you do the card for saturday by any chance?
sorry to be so much trouble.ATTN: UNCLE MO MAY SCRATCH.:eek:

[mikesal57]

Ted..my opinion is that those monster numbers were from last year as a 2yo and the 2 he ran this year would be in line with the rest of the field.

[Ted Craven]

Ted..my opinion is that those monster numbers were from last year as a 2yo and the 2 he ran this year would be in line with the rest of the field.

I can't argue with that at all. If we make his Hi SR this year's 106, he's still the POR and he and MMM still the top 2 Early contenders. Let's see if he still runs...

One frailty of the Newpace approach (as currently implemented) is that the sampled range of SRs for smaller populations (say 5 and less pacelines) and thus for younger and typically improving horses can include either poorer past history or superior past history not relevant today. Current condition and form is not assessed in this. Which is why I tend to favour older horses with fuller histories. Even then, this can be a bit of a broad-brush approach because of the foregoing.

I'll get around soon to my other analysis.

Ted

[Ted Craven]

Gosh, I can't type fast enough :eek: Uncle Mo has scratched, so back to the drawing board.

Ted

[Ted Craven]

OK, here's my reworking of the Derby using only EL 2.0

22091

The reasoning for Comma to the Top as 2nd Early contender is that all Early horses are separated by only 1 SR point in the Hi column, and 1 point is not enough to separate. MMM with 102 is first among equals. To separate ties, in the follow up workshop video, Dave suggested that as an alternative to (or instead of?) Jockey % he began to use lifetime high SR. So Comma to the Top's lifetime high SR (TrackMaster SR) is 105, last race, same as MMM (though 2nd back). And this 105 bests Pants on Fire's 103, also last race. So Comma wins the tie breaker and is the second E horse.

This is setting aside that from my normal handicapping approach for the Derby, I would not even use those horses (earlies or lates) who did not win or finish 2nd very close up in G1 or G2 events in 2011. This basic contender filter has not let me down in 10+ years.

Thoughts?

Ted

[SilentRun]

OK, here's my reworking of the Derby using only EL 2.0

22091

The reasoning for Comma to the Top as 2nd Early contender is that all Early horses are separated by only 1 SR point in the Hi column, and 1 point is not enough to separate. MMM with 102 is first among equals. To separate ties, in the follow up workshop video, Dave suggested that as an alternative to (or instead of?) Jockey % he began to use lifetime high SR. So Comma to the Top's lifetime high SR (TrackMaster SR) is 105, last race, same as MMM (though 2nd back). And this 105 bests Pants on Fire's 103, also last race. So Comma wins the tie breaker and is the second E horse.

This is setting aside that from my normal handicapping approach for the Derby, I would not even use those horses (earlies or lates) who did not win or finish 2nd very close up in G1 or G2 events in 2011. This basic contender filter has not let me down in 10+ years.

Thoughts?

Ted

Ted,

You have followed the "New Pace" methodolgy and therefore I concur with your
analysis above. This is the correct approach because you want to measure "New Pace" performance amoungst all levels of competition.
I feel you should not tweak it any further & go with it as is.

All of the E/L 2.0 contenders have either: won @ G1 (#8, #1) or G2 (#13)
and #6 just missed by a head in his last G1 race on 4/11. So these are viable
contenders.

(As an aside I like 2 other contenders that RDSS came up with....Nehro #19 and Santiva #12)

Ernie

[Ted Craven]

Thanks Ernie! I like those 2 other horses too, especially Santiva. I'll get to work on the rest of my analysis next! I posted this EL2 take on things in the public Kentucky Derby thread.

Ted

[mikesal57]

I can't argue with that at all. If we make his Hi SR this year's 106, he's still the POR and he and MMM still the top 2 Early contenders. Let's see if he still runs...

One frailty of the Newpace approach (as currently implemented) is that the sampled range of SRs for smaller populations (say 5 and less pacelines) and thus for younger and typically improving horses can include either poorer past history or superior past history not relevant today. Current condition and form is not assessed in this. Which is why I tend to favour older horses with fuller histories. Even then, this can be a bit of a broad-brush approach because of the foregoing.

I'll get around soon to my other analysis.

Ted

Ted...this can be a discussion for later on...but when you said that 106 was from THIS year..I said WHAT!!!

as you can in the PP's Uncle Mo's Beyers for his last 2 races this year is 96,95 which is about the same as other entries or worse..but TM has it 106 and towers over the others???

scratching my head

mike

[socantra]

Ted...this can be a discussion for later on...but when you said that 106 was from THIS year..I said WHAT!!!

as you can in the PP's Uncle Mo's Beyers for his last 2 races this year is 96,95 which is about the same as other entries or worse..but TM has it 106 and towers over the others???

scratching my head

mike

Comparing isolated speed ratings from different services will make you do more than scratch your head. Bris has a much tighter scale, Beyers and Trackmaster are on a fairly similar scale, but Trackmaster tends to run 5 to 12 points higher than Beyer, especially at the upper ranges.

The attached image is a quicky of the last 2 SRs from all the Derby horses. If I didn't blow the data entry from trying to handle 3 different sets of PPs. it might help to see the relationship. Hope it helps.

Dick...

[Dave Schwartz]

My Look at the Kentucky Derby

http://www.horsestreet.com/BBSImages/NewPace/2011-05-07-Ky%20Derby.png

[Ted Craven]

Dave, not sure you've got the right race in your screenshot.

Ted

[shoeless]

Ted,

Nice screenshot from RDSS 2.0 and thanks for the reasoning behind it.


Jeff

[Dave Schwartz]

LOL I wasn't even paying attention. I'll fix.

Funny about that.

[Dave Schwartz]

See how little I pay attention to these big races? I didn't even look at the horses' names.

Okay, have corrected it. Just refresh the screen.

I will be around to respond to questions after about 7:30pacific.


Regards,
Dave Schwartz

[Ted Craven]

Dave, before analysing your workup further, how do you assign any ES points to Master of Hounds, given no upclose in his only North American race, and no position info in any other races? His presence in your Early group, affects the horses below him, and your subsequent query about dual Swing horses, I think?

22145

Aside from that, I follow your analysis. We're using 2 different sets of self-coherent Speed Ratings (TrackMaster and HDW). Given that, could you comment on my reasoning in my earlier post above for choosing Comma to the Top versus Pants on Fire as one of the Early horses (i.e. tie break rules when Hi SR within 1 point). That is, what approach would you follow faced with an analogous situation using HDW supplied SR?

Cheers,

Ted

[Dave Schwartz]

Dave, before analysing your workup further, how do you assign any ES points to Master of Hounds, given no upclose in his only North American race, and no position info in any other races? His presence in your Early group, affects the horses below him, and your subsequent query about dual Swing horses, I think?

I am so used to our software handling some things automatically, that I just forget about it sometimes.

First, he has zero for ES Points. Since ES Points are not part of this calculation we'll leave that to a future discussion.

The real question is, "How did we determine his other attributes?" Essentially, I studied foreign horses in their next few races in the U.S. and determined that there was a good way to relate finishes to "likely" pace impact.

The approach I settled upon was to treat the past races as if he stalked or forced the pace based upon distance of the race coming from and the distance of today's race. In other words, a horse that has been sprinting (like Hounds) is more likely to run early in a race of this length than in a race of shorter distance. That is simply logical.

However, the C1 factor takes most of this back because we actually have no proof that the he DID run this way. Considering that the C1 number is a zero, I would expect this horse to actually run somewhat earlier than his "pct" indicates. It is a big leap from "no data" to being able to say this, but that has been my experience.

When you have a true question mark horse in a race, you cannot simply pretend he isn't there. Well, there are some methods for doing that relative to the tote board - I do it all the time with FTS - but it is very difficult when asked questions like, "How will the question mark horse run?"

A logical assumption, since so many winners go to the front, is to assume that IF HE WINS, he will do it at least somewhat early.

Besides, what are the alternatives? To give him all zeros would be to assume that he will be an absolute late closer, which is no more accurate and probably farther from reality.

[Dave Schwartz]

Aside from that, I follow your analysis. We're using 2 different sets of self-coherent Speed Ratings (TrackMaster and HDW). Given that, could you comment on my reasoning in my earlier post above for choosing Comma to the Top versus Pants on Fire as one of the Early horses (i.e. tie break rules when Hi SR within 1 point). That is, what approach would you follow faced with an analogous situation using HDW supplied SR?

Actually, I am not using HDW speed ratings. I am using my own par-based numbers.

The HDW ratings are far superior to my numbers for gathering winners. The top horse for last race speed rating - what we call "BL1" - wins around 29% of the time. My par-based numbers are down around 25%.

However, my numbers return MUCH MORE MONEY in the upper rankings! So, I use them instead.

On to your question.

As I read your previous post, you go into a long explanation about PoR (Pace of Race). The purpose of PoR is to eliminate horses that cannot keep up with a potentially fast pace. However, from a practical sense, it is only used IF WE HAVE A DOMINANT FRONT RUNNER!

Therefore, in a race such as this, you have no need to even LOOK at the PoR. That is why I did not display my RegSpeedLow ratings on the form!

Yes, PoR is used to eliminate horses that do not qualify on SpeedHigh, but in the end, we're going to simply take the two highest (plus ties) anyway so we might as well just move to that column right away.

Thus, in your calculation (as I see it) you wind up with the following:


EARLY
18 Mo 123
13 Mucho Macho Man 102
7 Pants on Fire 101
6 Comma on Top 100
17 Soldat 100
5 Decisive Moment 99
14 Shackleford 97


In this scenario, you clearly have:

18-??? tie break between 13 & 7.


(Hmmm... Code does not line up. Can you point me to how to fix?)

[Ted Craven]

As I read your previous post, you go into a long explanation about PoR (Pace of Race). The purpose of PoR is to eliminate horses that cannot keep up with a potentially fast pace. However, from a practical sense, it is only used IF WE HAVE A DOMINANT FRONT RUNNER!

Therefore, in a race such as this, you have no need to even LOOK at the PoR. That is why I did not display my RegSpeedLow ratings on the form!

Yes, PoR is used to eliminate horses that do not qualify on SpeedHigh, but in the end, we're going to simply take the two highest (plus ties) anyway so we might as well just move to that column right away.

Thus, in your calculation (as I see it) you wind up with the following:


EARLY
18 Mo 123 SCRATCHED
-------------------------------------------------

13 Mucho Macho Man 102 - 105
7 Pants on Fire 101 - 103
6 Comma on Top 100 - 105
17 Soldat 100 - 112
5 Decisive Moment 99 - 102
14 Shackleford 97 - 97
In this scenario, you clearly have:

18-??? tie break between 13 & 7.


(Hmmm... Code does not line up. Can you point me to how to fix? - not sure myself - let me work on that...:confused:)

Thanks Dave, setting aside your discussion of use of POR for a subsequent time, your chart and comments re tie-breaking were actually the meat of my question.

Uncle Mo is scratched. So we're picking between the next 5 horses separated by only 1 SR point. You have suggested (initial video) to break ties on Jockey %. You have also suggested (followup video session - an improvement?) to break ties on highest lifetime SR. I'm sure either way will get you success some of the time, but for consistency - using lifetime SR - do I understand:

(referring to your code table above - I added a second set of bolded numbers showing each horse's lifetime hi SR)

1. comparing #13 to #7 - #13 survives (105 vs 103)
2. comparing #6 to #7 - #6 survives
3. comparing #6 to #17 - #17 survives

Therefore the 2 Earlies would be #13 MMM and #17 Soldat

In your view, is that comparison logic consistent? (I agree, it is different to what I came up with initially). This is not so much to determine Derby contenders (though that's interesting) as to be in synch with your thoughts on best practices applying NewPace.

Thanks!

Ted

[Dave Schwartz]

Hang with me - working on a long post...

[Dave Schwartz]

Changes to the Original Methodology

I know you guys like the M-word. When in Rome...

I am in constant contact with several key users of the NewPace approach, and am working on several interesting developments. One very large one is the melding of Improve/Decline (http://thehorsehandicappingauthority.com/products-page/handicapping-systems-methods/improve-or-decline-video-seminar-cd/) (our form product) which I am very excited about. More to come on that later.

There are several changes - still labeled "experimental" at this point - that I have been using for over a month.

The first is that I have stopped using any of the 12-point/PoR stuff. It is simply too much work. I am not sure if it has hurt my bottom line but do not think so. It occasionally takes me off a very obvious odds-on horse and puts me on a huge longshot that loses most of the time. Frankly, I don't care. My $Net is currently - brace yourself - over $2.30 (after 1,100 races) driven on by high-odds $Net of over $3.00. (That is including the fact I wager over 3 horses per race on average!)

So, ease of use equals more plays. What do I care if it costs me a penny or two?

I have also dropped the complication of the swing horse not being involved in the PoR. Same reason as above: simply too much work.

My effort is to constantly simplify (if possible) the process. While some simplifications result in a significant lowering of $net, a few actually cause improvements. The reality is that the product you are using is the direct result of a serious simplification resulting in a huge positive shift in $net.

I had been playing with NewPace for about 12-15 months using a somewhat conventional pace approach. It was "conventional" in the sense that it used pace ratings to decide which of the E's or L's became final contenders. I actually gave up on it because after all that time (and handicapping roughly 50,000 races manually) I was unable to get within even 10% of being even. Flashes of brilliance, but nothing long term.

Understand that back then there was no "Early Pace Section." It was just two handicapping objects: Early and Late. Take the top 2 E's and the top 2 L's and move on.

So, I went on to work on Improve or Decline and left NewPace behind. (I actually joked to my wife that I should call it "DeadPace.")

One day I decided that I should build a pencil-and-paper handicapping product out of it. If you think NewPace is time consuming now, imagine how tough it would be if you had to MAKE pace figures as well! So the problem was, "How do I replace the pace figures?"

That was when I hit upon the idea to separate the field into two segments. The question, "Who will run early?" was replaced with, "Who is likely to create early pressure?"

(Side note: You will notice how I clearly state within quotes the question I am trying to answer. This is a powerful problem-solving technique. Try practicing it yourself.")

It turned out that the segmentation of the field was critical to the success of the approach! After that it was merely a problem of isolating which speed ratings would best indicate potentially long-term-profitable winners.

Research, research, research!

Okay so, back to developments... In my next post I will simply walk you through my actual REAL process. BTW, being able to change and tweak YOUR process is why we programmers resist just giving you a button that spits out:

E-E
-----
L-L

... which is my shorthand for the NewPace output.

[Dave Schwartz]

The Process

1. EARLY SECTION
I do that as we always have. Add the pcts until reaching 62 or more, then add 1 horse (who is labeled the "swing").

I ask myself, "Is there a dominant horse?" Specifically, does the top horse have a 58% edge over the 2nd horse and at least 25% himself? Notice 58% rather than 62%.

If the answer is "no," then I have no need of a PoR!


2. DOES THE DOMINANT HORSE QUALIFY FOR POR?
If there is a DomES horse, then I do PoR. I simply take the highest of the RegSpeedLow ratings to represent the PoR.

Remember, I am only trying to answer one question: "Does the dominant E horse have a fast enough RegSpeedHigh to qualify?" That is the sole purpose of PoR.

At this point I have either 0 or 1 EARLY horses.


3. SPEED RATINGS for EARLY
Nothing has changed here.
All the EARLY horses are ranked for RegSpeedHigh. The top 1 or 2, (break ties) will be considered for contenders. Obviously, if I have already qualified a DOM-E I only need one more.

Why do I ignore PoR? Because if I come up with only one horse that qualifies, I am going to simply assume that the PoR is at least somewhat unpredictable. Someone is going to chase that big, obvious PaceMonster.

I developed this rule because I had lost one-too-many long-priced winners to a race with an obviously bad pace-of-race prediction.

See, a horse with that big number is talked about by the trainers and jockeys. They make decisions about "what to do if..."


4. SPEED RATINGS for LATE
Everything here is EXACTLY the same. Top 2, break ties.


A few words about tie breaks
First, I chose jockey% because it was designed as a paper-and-pencil method.

We who use software have more sophisticated means available to us. Take advantage of those means. You want to use a high-level tie-breaker.

That is, you want to use a tie-breaker that looks at the horses from an intrinsic "who is the best horse?" question. Personally, I use:

PSR (weight 100)
CPW (weight 62)
RTG (weight 38)

PSR is HDW's "Projected Speed Rating." It correlates very highly with the final odds. Top horse wins 31% of starts. It is not even close to profitable because of the high correlation to odds.

CPW is simply an older version of PSR.

RTG is a high-level rating which uses 38 factors. Lots of ES, class, form and speed all mixed together.

Note that I use the same tie breaker for deciding who to bet under 7/2.

[Ted Craven]

Thanks Dave! I appreciate your time to update your latest thinking and research (which indeed never ends!) I'll be studying this for sure.

Derby day races are catching up to us, and I've some handicapping to do.

cheers,

Ted