%Median versus %Early in Follow Up Articles

[Ted Craven]

Can anyone point me to articles by Doc Sartin in Follow Ups where he discusses the relative merits of %Median versus %Early readouts.

Both formulae are attempts to measure how much of a horse's inherent energy (at the time of a given race) is exerted to the important 2nd call point (4f in sprints, 6f in routes) - a point marking 2/3 to 3/4 of the distance of most races.

%Early is calculated as (2nd call velocity + 3rd fraction velocity) / 2nd call velocity.

%Median is calculated as (1st fraction velocity + 2nd fraction velocity) / (1st fraction velocity + 2nd fraction velocity + 3rd fraction velocity) or (F1+F2) / Total Energy.

Doc moved away from %Early claiming it was no longer as effective as %Median, which had been documented and made mainstream by Tom Brohamer in Modern Pace Handicapping. Or perhaps it was that he found %Median inherently more useful, or perhaps they are/were both equally useful but he wanted to keep mixing things up, while conveying that he was on the cutting edge...

I'd appreciate any reference to his written thoughts on the matter.

In RDSS (and Speculator, Validator, Synthesis, PaceLauncher) these calcualtions are made after being distance equalized, and adjusted for Daily and Track to Track variants.

The difference between the 2 formulae is that %Early employs the average velocity to the 2nd call point, while %Median uses the sum of the F1 plus F2 velocities - perhaps a more fine-grained approach. I also know that Doc stated later on that %Median itself was overrated as a standalone factor because it didn't measure energy interelationships and decelerations between today's horses (something like that). Also, a hammer is overrated as a standalone tool because you can't use it alone to build a house...

I wonder if both of these factors are interchangeably useful, or are they about equally irrelevant?

Any thoughts and particularly Follow Up references would be appreciated.

Thanks in advance.

Ted

[froggy]

Ted,

One time in Las Vegas I heard Doc Sartin say Median Energy was more accurate.
Some where in his writings I seem to remember reading the same thing.

In Follow Up issue #12, page #6, Dick Schmidt mentions it and then discusses it in length in his computer column beginning on page #19.

I hope this helps.

Froggy

VIVA RDSS

[Ted Craven]

Froggy,

Many thanks for that, just what I was looking for! I'm sure there are some more in there somewhere, but this is an excellent start! Here's that article.

Ted

One of the things that many of our clients find most invigorating about PIRCO (and other clients find most annoying) is that we are always changing, always trying out new ideas and developing new programs and procedures. Those of you who wish we’d just tell you what to do, the best way to do it and then shut up are in trouble. The entire point of PIRCO’s existence is to do research. Pari-mutuel Investment Research Company, in case you ever wondered. Research is literally our middle name!

That means we’re always trying new stuff, making improvements, fiddling around with new ideas. Some of our new ideas are great, but, as with all progress, most of our new and wonderful ideas aren’t much good. The problem is that -it’s very hard to predict which will be
hits and which misses. Percent Median is a classic example. When we started research on the. ENERGY! program, we had the computer display our old % Early, a new % Early, % Median and % Late. After some months of trial and error, we discovered that -the new % Early wasn't very useful, but that % Median far exceeded our expectations. It is good enough that we will eventually incorporate it into all of our computer programs, replacing % Early.

% Median is used exactly like our old, soon to be departed friend % Early. In this case, the improvement is one of degree, not a major
breakthrough. We have found that % Median is superior in two ways to good old % Early. First off, the Win Energy parameters are generally about 25 to 33% tighter with % Median than they were with % Early. This makes the number much more useful, as you will be able to eliminate those horses who don’t fit your new, more exact track profile with greater certainty.

Second, we have found that % Median more sharply defines the upper and lower limits of the profile. For instance, at Del Mar this
year, we found we could almost always safely throw out any horse with a 74 Median of over 70% from win consideration. At other tracks, a very definite lower limit was discovered by observant testers. Once established, these limits tend to hold up far better than those using % Early. Of course, weather and class can still dramatically change a track profile, but % Median is a superior tool.

Take a look at the charts one of our advanced testers Joe Bolger created, based on the recent meet at Del Mar. A quick glance should
be enough to convince you of the superiority of % Median.





As I’m sure you all know, % Early is computed as:

Early Pace / (Early Pace + Third Fraction)


Of course, all numbers are first converted into feet-per-second. An example of this would be:

6 furlong race

45 second call (58.66 F.P.S.)
1:10 final time (52.80 F.P.S.)

58.66 / (58.66 + 52.80) = 52.63%


The new % Median is a close relative, but uses all three Fractions. I’ll condense First Fraction to F1, etc.

(F1 + F2) / (F1 + F2 + F3)

An example would be:

6 furlong race

22 First call (F1 = 60.00 F.P.S.)
45 Second call (F2 = 57.39 F.P.S.)
1:10 Final time (F3 = 52.80 F.P.S.)

(60.00 + 57.39) / (60.00 + 57.39 + 52.80) = 68.97%

(Also, thanks to Joe Bulger for his 20 year old original research )

[socantra]

I'm glad to hear that %median does have some improved accuracy. Still, I think the increase of 14% does not really justify the confusion it has produced over the years.

I like to play with both %early and %median and after trying to get some of the numbers to common ground for comparison purposes, spent a sleepless night recently coming up with the following conversion formulas:

%Early to %Median

%M = (%E * 2) / (%E * 2 + ( 100 - %E ))

%Median to %Early

%E = (%M / 2) / (%M / 2 + ( 100 - %M ))

I'm no mathematician, just a curious geek with a pocket calculator, so I'm not sure about the formatting, or if there is any simpler way to list them. I'm also not completely sure they will work with every velocity combination, though time works just fine, and the 50 or so examples I've worked my way through in the last few days by calculating them both ways and then converting seem to come out fine except for rounding errors of no more than a couple of hundredths of a percent.

I recommend a spreadsheet or pocket computer for intensive use, though any pocket calculator with nested parentheses or multiple memory locations will do the job. Keying in all the parentheses tends to be a rather error prone process.

Dick...

[socantra]

Help!!!

Well, I said I was no mathematician. I just realized that in the above formulas, I've been entering my percentages in as 53.26, or so and getting them out as .6950, In other words, I need to get my multipliers and decimals lined up. On a pocket calculator. you tend to correct that sort of thing in your head

Back to the drawing board to fix it, unless one of you kind souls fixes it for me.

Dick...

[socantra]

I've refined the formulas a little. Current versions are:

%Early to %Median

( %E X 2 ) / ( 1 + %E )

%Median to %Early

( %M / 2 ) / ( %M / 2 + 1 - %M )

In both formulas, %Median and %Early are input and output as 4 digit numbers after a decimal, ie: 53.26% = .5326, 69.50% = .6950, etc.

I think I'm getting down pretty much to basics on %Early to %Median, but the other one could probably still be cleaned up a bit.

Dick...

[CEW]

Hi Guys.............

I've been reading your messages on % E and % M. I use % M, but my feeling is they are damn near one in the same. I'll give the slight edge to % M based on the fact that it is more realistic in terms of what percentage of the race has actually been run to that point.

I'm sure that PIRCO wanted to get away from % E once it went into the mainstream books. But really, when % M adds the the first and second fractions together in the formula, you are getting an average, and thereby you are still using the 2nd call (EP). Howard always said that it is tighter than % E, and he's right, of course.

Here's something for you guys: I've been messing around with the "Biasgraph", which was introduced in Follow-Up # 13 (I think!). This graph was never again discussed in future Follow-Ups, but it is very interesting and revealing. You graph your 4 or 5 contenders by velocity using FW, EP, and SP (the main readouts back then). They get plotted across the graph at their respective % E or % M. The guy who designed it was looking for a way to project the winner like the variagate that came in future programs.

If anyone wants to see one, let me know how I would input an 8 1/2 x 11 print out onto this message board.

Enough ramblings from me for now!

Chuck

[Ted Craven]

Chuck,

I would very much like to see some examples of the Bias Graph.

Ted

[Tim Y]

The early/late number is the same thing as the bias graph. Keep standards for distances/track, then race to race see what wins: same thing.....

The early late also discovers improving early speed, one of the hallmarks of improvement overall.

[socantra]

Originally Posted by CEW

Here's something for you guys: I've been messing around with the "Biasgraph", which was introduced in Follow-Up # 13 (I think!). This graph was never again discussed in future Follow-Ups, but it is very interesting and revealing. You graph your 4 or 5 contenders by velocity using FW, EP, and SP (the main readouts back then). They get plotted across the graph at their respective % E or % M. The guy who designed it was looking for a way to project the winner like the variagate that came in future programs.

If anyone wants to see one, let me know how I would input an 8 1/2 x 11 print out onto this message board.

Chuck

I'd also love to see one. If tou can print out to pdf, you can attach that, I believe.

By the way, I went hunting for the Bias Graph article, and its in Follow Up #9.

dick...