Paceline picking

[andicap]

I've been fooling around with a paceline selection method as a way of getting my contender list from three to two, a real challenge.

I'm trying to eliminate the horse that does not belong pace or energy-wise. An early horse who will have to work too hard to keep up or a sustained horse who will be too far back or a horse who's energy is just out of whack with the demands of the race and/or the track.
So far, some promising results using this PL method I'd like some feedback on.

Only use pacelines from good races. I'm defining as within 3 lengths at the finish and no worse than 3rd if possible. I want to see what horses can do when they are in contention.

Look for the fastest paces (2nd call) that each horse has successfully competed against, preferably recently and at today's surface/distance structure (although this isn't always possible as we all know).

I'll use one or two races and if I find an aberrant line -- a really good race the horse hasn't repeated, I'll "water it down" with a second more realistic line or maybe ignore it.
-----------------
Examples from today at Belmont in using this method.

Today at Belmont in the 3rd race the pacelines and %E would have pointed to the 1 and the 2.
The 4's %E of 53.17 suggested he would go too fast too soon. it was the highest in the race and I'll consider that kind of an energy figure only when he horse is lone speed and is unpressured so he can take the %E down to %52 or so.

The 3 had the lowest, 51.40 and I don't like taking the lowest unless the track is really playing that way or the race is incredibly contentious pace-wise.

In the 5th race the 1, 8 and 11 were my three final contenders (as well as everyone's) and I wondered if the 1 at 5/1-9/2 fit with the 11.

Obviusly the 1 is a play on the basis of lone speed -- 8 QUirin points to 4 for the next highest horse, but I wanted to use energy and running styles to understand why the 11 had little shot.


The average %E of the race was 50.19. I like to use a race average to see how a horse fits with those animals. I'll use the track model as a guideline, but horse's win all the time with %E that don't fit the "profile." That's why people throw up their hands and don't use energy.
The 11 was an Early horse wiht a 49.90 %E !!!. That's a red flag because it means he won by taking it easy early. If he faces any pressure he could be fried. That's what happened.

The 8 at %49.31 as an "S" horse would probably too far back to do any damage while the 1 at %50.74 wasn't too high and wasn't too low. Just in the middle where I often like it.

I'm not redboarding here -- I didn't bet the races, just analyzed them.

[DJofSD]

andicap,

One comment: don't ignore or eliminate horses because of what appears to be a bad paceline. Sometimes they're "better than they look."

DJofSD

[andicap]

Thanks,

Actually I agree with you on that. But in eliminating a horse on pace, I need to see what kind of a pace he runs very well against.

This is part of the elimination theory I posted here last week following my talk with the handicapper I ran across Belmont Day.
I have guidelines for getting to 6, then to five, then to four and then to three, but none yet for two. I find when it's too tough to make a decision on eliminating a horse I either a) pass the race or b) demand higher odds on my key.

Often I post these long notes as reminders to myself so that if I lose my focus I can re-read the notes and remember what I was thinking as I was handicapping that day. It's kind of like my own personal web-log.
I guess I should just start a web-log so I don't have to bore y'all.
:)

[Tuffmug]

Andicap,
Enjoy your notes on your search for handicapping niggets. Thanks for posting.

Question: As you use %E, have you noticed any correlation between average %E or winning %E and classes of the races?

[NoDayJob]

Originally posted by andicap
Thanks,


I have guidelines for getting to 6, then to five, then to four and then to three, but none yet for two. I find when it's too tough to make a decision on eliminating a horse I either a) pass the race or b) demand higher odds on my key.



What percentage of your winners come from your top three horses? If it's high enough the average of the last 4 purses that the horse has run for may be of some value for final separation.

NDJ

[andicap]

Good question.
I've noticed for sure with maidens and younger horses that higher %E's win but not yet among say, Claiming 17K and claiming 50K.
---------------------
(WARNING: MORE MUSING TO MYSELF COMING. DON"T READ IF YOU ARE OFFENDED BY NAVEL-GAZING HANDICAPPERS WHO ARE WORKING OUT THEIR 'ISSUES" IN PUBLIC.)

My current thinking is to only compare %E among my final three or four contenders (though in relation to the entire field's average). So if the entire field averaged a 50.85 and my final three have

52.10
50.90
50.10

I'll eliminate the 50.10 if I think that type of sustained pace can't win today either because of the pace demands of the race or the way the track's been playing in general.

In the past I haven't been able to make money on %E in comparing the entire field. Now I'm solely using it as an elimination factor in the latter stages of handicapping. The paceline selections are NOT geared toward picking a winner on velocity, etc., but
a) eliminating a horse based on %E, early or late.
b) finding "E" or "EP" horses who can't win today because their %E has been too high against today's pace demands.

Let's redboard EMD's 8th, a race I quickly did and watched but didn't bet.
Generally for 8.5f, EMD demands in the mid 51% range based on the model. After paceline picks based on horse's good performances against the fastest possible pace I found the race average to be %52.32. and the pace scenario seemed pretty contentious.

The 1, 6, 4 were my finalists with %Es of

51.32 -- 1
52.67 -- 4
50.31 -- 6

52.37 -- 3, a close contender.

The 4 has to go because the %E is too high for the track and today's race, since no way will he be able to whittle that 52.67 down to the 51-52 range today. In a lone speed scenario it might be different if the horse had shown the abilty to rate.

that left the 1 and 6. The top two favorites. Whoopee. No edge here.

It finished 6-3-9-1.
.
The elimination factors are tricky. If I had used %E with four horses to get down to three, I'd have the 1,6, 3 as my final three and doubts about the 3 since its %E was still higher than the race average and the track profile.

Rather than make these fine-edged choices that are bound to be wrong a number of times -- and which I don't have the time to do a full-throttle handicapping effort -- it might behoove me to just find decently priced horses who meet all of my four or five elimination factors.

In other words, be negative down to three horses, and turn positive after that looking FOR reasons to bet a horse instead of finding reasons to dismiss it: Strong trainer angle, key race, performed well vs. bias, excellent HTR strength of race rating in a good race recently, potential F1 longshot, high HTR workout rating, good form cycle pattern, horse for course or distance, among top two velocity ratings at attractive odds.

Now if I could only do all that in 10 minutes ...

[andicap]

Originally posted by NoDayJob
What percentage of your winners come from your top three horses? If it's high enough the average of the last 4 purses that the horse has run for may be of some value for final separation.

NDJ

Too early to tell using this new process that is about two weeks old but incorporates all my raw tools I've been using, but in a chaotic, scattered approach. I've been searching for a systematic -- but not a "system" -- approach to handicapping each race.

I feel very strongly many of us have the skills to be successful, but are not organized or structured enough in the process to make effective decisions. I find in my "real" life I am less productive when I have a million things to take care of, but no real plan to attacking them. An author will use an outline before writing a book. Why shouldn't a handicapping need a framework to approach his craft?

I'll check out that purse angle, thanks.

[GR1@HTR]

AC...check out your stats in excel/access when your top fr1/fr3/ep/ap/sp/lv/vel horses go off at odds above the vBet (odds line). You might have some good results.

[BeatTheChalk]

So who won the race !!!! I am losing it here....like the robber
in the pic with Clint Eastwood....He is lying on the ground with
Eastwood pointing the gun ....anyway the punch line was
I JUST GOTS TO KNOW !! help :) thanks cheers E

[BeatTheChalk]

My apologies .. I found the winner in your postings !! Looks like
the most sustained of the bunch...won the race. I will re read the
postings to get my head straight :) " Percent Early " is the key
to picking winners and elminating losers.

[andicap]

Well, that's the theory, anyway. Won't know for sure until I road-test it a bit. The problem is I've been blowing off a project I've got due in two weeks to work on this, so after today I'll have to take a break from this stuff.

Work sucks!!!

:(

[andicap]

Originally posted by GR1@HTR
AC...check out your stats in excel/access when your top fr1/fr3/ep/ap/sp/lv/vel horses go off at odds above the vBet (odds line). You might have some good results.

Thanks, GR.
Of course I have to learn to use Excel first, but I know it's not that difficult.

[chickenhead]

Hi Andy,
I was wondering if you found any definate correlation between race avg. %E and track profile.

For instance:

Race Avg. %E > TP %E, then winner %E < TP %E
Race Avg. %E < TP %E, then winner %E > TP %E

If there is a pretty good correlation, I would think at a certain gap between avg. and TP will start to push the winner to the other side of TP. Your EMD was an example of this, wondering how often it actually works out that way, and how big the gap normally is for it to happen. Also, do you track to see how close they ran in todays race to the pacelines you are using?

Good stuff.

[andicap]

The first race I checked your theory, it worked!!! :D

seriously I'll keep tabs on it -- good thought. I actually did something like that a few years back when I used Colts Neck, and it picked a lot of longshots. Why give it up -- it was a royal pain to use because as a DOS program I had to redo all the scatches by hand and recaluate everything.

I try to go back and ck the pacelines but time is not on my side.

Also the pacelines are not designed to pick the winner, but find the true energy demands of the race. But it's always worthwhile to second guess yourself. I made a few mistakes on Sunday at Belmont which cost me one race -- I gave the 12 horse in the 9th race way too much credit as well as the 1 horse in the 8th -- even tho I averaged the 1 horse's best performance with his 2nd best to smooth it out. It was still on top and the horse ran poorly at 8-1 with my $$$ on it. The 1 horse was a pig -- one peak performance 2 races back and a bad trainer. Bounced off that race last time and there's no evidence he was back in form. Bad PL picking.

I am keeping tabs tho on how well my pacelines do in picking a winner and if there is overlay value there in sticking to the top two choices. Can't always do that because at times I'll pick a PL that's not designed to be representative of its current form, but WHEN it is in form. I have to think about that because if a horse isn't going to run his race then I shouldn't be picking PL based on him giving a competitive performance.
I also want to see how a horse will do against the toughest possible pace scenario he's ever faced and compare those pace figs to the rest of the field. If an "E" horse has never had success against more than a 101 pace and there are two "E" horses today who can go 103, the first horse becomes a non-contender.

Some Sunday examples
First rae, 8F maiden turf, using the average for all races (the maiden sample had 2 races not too different than overall), of 50.98.
Ave %E was 50.84 so it was pretty close to the norm. The field had some cheap speed in it so I thought a lower %E was the way to go.
of the four main contedners, Theatrical Cat had a 49.33 -- Ulla had a 50.04, Angelina a 50.01 and Suncoast a 50.13.
Out goes Theatrical.
Finished Ulla-Suncoast. Ulla had better velocity figs and a higher strength of race rating too.

In the 6th race the ave. of 50.56 was higher than normal (taking both the wet and firm %E ave.) and there was plenty of speed. Time to go below the race average.
of my 5 contenders, the qualifers were the 4, 8 and 10 (who's %e was the race ave.)
the 8's was way too low at 49.57 and the 4 had better velocity figures. I hated the 10's post, trainer and its early run style.
the 4 won at 4-1.

Won an interesting bet in the 7th.
The 7 was odds on in a race with an 52.44%E, about the track par. Not much speed in the race and the 7 had a 52.53 %E, a very nice match. The 5 was second choice was a higher %E, 53.19, but an excellent form cycle pattern for a young 3 yr old and Pletcher training.
I decided to see if there was a counter-horse I could put on the back end, a horse that ran counter to the winner, someone at a price.
Looked for the best horse below the par. That was easily the 4, Affair in the Air at 19-1.
Played him to show and in exacta with the 7 and 5. He showed, paid $4.80 and I made a profit on the race.


I lost the 8th, an interesting illustration of your point.
Race ave was 51.24, a good point higher than the track profile.
I tossed the winner, Brandala for the win because I thought his %E of 49.65 was too low.
From a Quirin Points standpoint the race looked "normal," was a 7, a 5 and two fours. But the %E gave it away.
The 4/5 chalk On the Bus had a 50.88 which I thought would be low enough, but the post position killed him because he had to use his speed to get position and he had to expend too much energy too early and lost ground to boot.

Had to remind myself -- speed on the outside near a turn when there's other speed inside is not a good thing to have. If On the Bus had been able to get clear before the turn and settlle in near the rail he would have won the race.

[BETKING]

Andicap:

During my years with the Sartin group, I remember that %E was replaced with %M. Have you checked this out. It was supposed to be a more accurate measurament.

I think that I still have that formular somewhere. If anyone is interested I will try to find it and post it.

BETKING

[cj]

I'd be curious to hear what %M is, I've never heard of that one before.

Thanks in advance...

[chickenhead]

if you don't mind me rambling a bit more I had a further thought to this....

I always had a bit of a problem with the way that %E was done as a profile for a track, I didn't know what it was, but it kind of clicked today.

If it's just an average of winners at that distance/class, it's hard to seperate whether it's telling you much about the track specifically or really just giving you the average pace scenario for races at that track.

If the avg. race %E is a good measurement of pace scenario, I was thinking you could profile a track in the following way:

Using charts, classified by distance/class whatever, take race avg. %E and winner %E. Group them by avg. %E, and then avg. the winners %E for that group, so for instance your profile would look like:

Race Avg. %E Winner %E Ratio Winner/Avg.
48-49 XX.x 1.xxx
49-50 XX.X 1.xxx
50-51 XX.x 0.99xxx
etc.

So you'd have kind of a lookup table based on todays race avg. Use that ratio to adjust todays avg. to get winners avg.

And the nice thing to boot, is the difference between ratios for given %E (like 48-49) from track to track would give you a nice look at how the tracks stack up to each other when they have like pace scenarios.

[BETKING]

Fomulae for %E and %M:

%E : EP / (EP + 3F)

%M : F1 + F2 / (F1 + F2 + F3)

EP = 2ND FRACTION

F1 = FRACTION ONE

F2 = FRACTION TWO

F3 = FRACTION THREE

"Dick Schmidt"

% Median is used exactly like the old %E. We have found that % Median is superior in two ways to %E. First off, the Win Energy parameters are generally about 25 to 33% tighter with %M.

Second, we have found that %M more sharply defines the upper and lower limits of the profile."

BETKING

[andicap]

BK,
My HTR software only calcuates %E so I use that.
%M was something like the first two fractions divided by all three fractions, I think.

[JimL]

CJ, % Median uses all three fractions (F1+F2)/ (F1+F2+F3) I find this number more useful in eliminating horses that do not fit the profile. Bet King can probably add more. JimL

[BillW]

Originally posted by BETKING
Fomulae for %E and %M:

%E : EP / (EP + 3F)

%M : F1 + F2 / (F1 + F2 + F3)

EP = 2ND FRACTION

F1 = FRACTION ONE

F2 = FRACTION TWO

F3 = FRACTION THREE

BETKING

I'm having a rough day ... what is the difference between EP and F1+F2? I thought EP WAS F1+F2 and the above was just a typo?

Is the difference in the route calculation only?

Bill

[cj]

Bill,

F1 + F2 is E2, but I think they are talking F1 velocity rating plus F2 velocity rating, and ditto for the denominator.

So, a horse runs 22, 45, 1:10
E2 = 58.67 fps
F1 = 60.00 fps
F2 = 57.39 fps
F3 = 52.80 fps

So, for %E, you would have 58.67 / (58.67 +52.80) = 52.63

I'm going to assume (F1 + F2) should have parentheses around it.

For %M, you would have (60.00 + 57.39) / (60.00 + 57.39 + 52.80) = 68.98

Did I do this right? I don't know the significance of course as this is brand new to me.

[BillW]

Originally posted by cjmilkowski
Bill,

F1 + F2 is E2, but I think they are talking F1 velocity rating plus F2 velocity rating, and ditto for the denominator.

So, a horse runs 22, 45, 1:10
E2 = 58.67 fps
F1 = 60.00 fps
F2 = 57.39 fps
F3 = 52.80 fps

So, for %E, you would have 58.67 / (58.67 +52.80) = 52.63

I'm going to assume (F1 + F2) should have parentheses around it.

For %M, you would have (60.00 + 57.39) / (60.00 + 57.39 + 52.80) = 68.98

Did I do this right? I don't know the significance of course as this is brand new to me.

Aha! Duh!

[BETKING]

cjmilkowski,

You did good.

Dick Schmidt where are you???

[stuball]

Could you please explain %M ? New term there.....

Appreciate it..

Stuball

[stuball]

Could you please explain %M ? New term there.....

Appreciate it..

Stuball

[stuball]

sorry about the double post
Thanks for the clarification......

Stuball

[DJofSD]

Yes, the definitions and derivations for %E and %M are correct.

Why is %M better than %E? It is more accurate. It's a better representation of the distribution of energy to that point in the race.

Let's use a 6 furlong race to illustrate. Using the previous numbers, we came up with 52.98%. This is for a measurement at the 2nd call, the end of the 2nd fraction.

But in a 6 furlong race how many furlongs have you run at the end of the second fraction? 4 furlongs that last time I checked. Distance wise, 4 furlongs is 66% of the total distance of the race. From an energy expenditure perspective, do you think you've burned 50+% or 66+% of the energy to get to the end of the 2nd fraction?

If you have studied some calculus, remember L'hospital's rule. Draw a curve on a graph. If you need to find out the area under the curve, one approach is to draw rectangles and to add up the area of all the rectangles. Need more accuracy, draw more, smaller rectangles. In essance, we say there is a curve which is the distribution of the energy of the horse during a race. We want to get an estimate of the energy. We can do that by drawing two rectangles (%E) or three rectangles (%M). Which is more accurate?

DJofSD

[BeatTheChalk]

You are correct sir ( Ed McMahon impersonation) Howard
went to Median Energy .. instead of Per cent Early. I use them
both. Sadly I dont have the will .. the "energy" and the time
to do things correctly. One must must keep a model. Since I
dont...I have to guess. Sometimes right others not right .There
was a red hot fave the other day...in a sprint race..with over
54 % early ! Those horses win every blue moon. So I go lucky
and tossed him. Got the winner at a great 5 to 1. Sometimes It
just aint worth it :)

[chickenhead]

I brought this up on another thread and everyone ignored me, same will probably happen again, but these energy equations don't have anything to do with energy. They have to do with work and force and probably other things as well, but that is not the same as energy. They are not really related. I'll shut up now. But one example first.

Horse A goes from a dead stop to the speed of light for a subnanosecond, then slows down very quickly and crosses the 1/4 at the same time as horse B who ran like a normal horse, 22.0 They both did the same amount of work in the same amount of time, yes, you tell me they used the same amount of energy?

[DJofSD]

chickenhead,

Your correct. The label leaves something to be desired if you understand basic physics.

Your right, by definition the amount of work performed when moving from point A to B is independent of the path taken.

DJofSD

[Tom]

Energy here I think, refers to the amount of "effort" or velocity, a horse has to ration out during a race. Don't think in terms of physics energy. The calulations are all based on velocity.
Substitute "Race Units" for energy.

[chickenhead]

I agree with you both, and I'm not just trying to be a stickler, but I think it undervalues and understates the advantage that a high velocity horse has over the field when he catches a slow pace, which is hard to quantify in any way other than real energy. At least that is my theory.

[DJofSD]

OK, let's get technical.

Why not just call it what it is, change in velocity per unit time, delta v divided by delta t -- that's acceleration, or in Doc's paralance, deceleration.

He called that next program after T'mation (EXDCX) Decleration Pars - that was pretty descriptive and accurate. But then it was reworked and offered it as Entropy. Now how does the 2nd law of thermodynamics get into the mix?

Maybe he needs to come out with 'LABU' -- lactic acid build up. There's a sure fire winner!

Didn't the Bard say 'a rose by any other name smells just as sweet?' I'll still use the concepts regardless of the label.

DJofSD <VBG>

[chickenhead]

well, let's get technical then. %M and %E don't measure acceleration.

Or, maybe better put nontechnically, they are kinda sorta a very obscure glance at acceleration.

What I want to know is, if accleleration is important, why not just measure it? You've got the formula, you've got the data. What is it about this particular formula (%E,M) that causes it to work so well? I am truly not being sarcastic or trying to be difficult, I am just wondering why out of all the particular pace cocktails people could come up with, why THIS one correlates so well? I am assuming that they came up with it experimentally, i.e. they tried a whole bunch of different formulations and found one that seemed to click. My question essentially is why does it click? What is the underlying logic that makes it stand out over, say, sum of positive acceleration?

[BeatTheChalk]

We modeled Per cent Early .. so I guess I still look at it. It is..
obvious that the Median Figure is better. You know more about
physics than I ever will. Darnit I shudda taken the course in
high school...but the folks who did ...were the nerds and the
eggheads. Me ? Hey I played Baskets and Baseball. Shows you
where it got me :) Thanks as always for your input