Forum: slam
Page 157
Subject: SSHR Technical Information

  Posted by: Madman - [21020124] Mon, Jan 14, 2002, 04:51
Just wanted to replicate some of my research from the main message board for anyone here that is interested. Also, things tend to last longer in this forum, so this acts as a kind of "backup".

FIRST INSTALLMENT: HOW TO CALCULATE "LENGTHS BACK" BETWEEN HORSES 1 and 2:

Definitions:
Sp1 = Speed number of first horse you are interested in.
Sp2 = Speed number of second horse your are interested in.
f = distance of the race in furloughs
LB = Lengths Back

FORMULA:
LB = (Sp1 - Sp2)*(f-2)*(1.25) / 10

Note1: If the horse #1 is the WINNING horse, subtract .25 from the above "lengths back" calculation.

Note2: Round the above total to the nearest .25.

This method has been tested on both dirt and turf (fast and firm) with experimental race data with the off-line race viewer. In addition, a pseudo-random selection of 3 different horses from my barns were chosen, and the results seemed to be verified.

ADDITIONAL VERIFICATION NEEDED FOR:
a) Do you indeed round to the nearest .25?
b) Do these results hold for off-tracks?

Any other comments / help are solicited. I've had many requests for this info., and just now got around to giving it a serious go at breaking it down.

Other info. to follow as it becomes available.
 
1Madman
      ID: 21020124
      Mon, Jan 14, 2002, 04:54 		Now, a copy of an email I just sent to an inquiring mind, with more detailed technical info. about the raceout.zip format.
_______________________________
Well, since Mr. Coker is calling it quits, I think I am somewhat absolved of my responsibility to my promise to keep from spreading these raceout.zip tips. As per your email request on the message board:

Raceout.zip code illustration:
Raceout 1271, first horse listed in race 0101:
KRISTEENE S., 16714 , 106.742592223185 , 0.276742901833907 ,-1.11308858036995 ,-1.09338490724564 ,-0.190719803571701 , 0.127802343070507 , 0.104566984623671 , 0.123006733059883 , 0.254576377094401 ,-40 , 1291 ,erik

Horse Name, Horse ID, Speed (coded), start%, raceview adjustment1, raceview adjustment2, raceview adjustment3, raceview adjustment4, raceview adjustment 5, raceview adjustment6, odds (coded), post position (coded), owner ID, ownername

Horses are listed by finish place, by race *999999 ends the report. Either 1.0 or 2.0 should start the report. 1.0 refers to very early raceout versions, IIRC.

Code Summary:
1) Speed factor:
sp! is the original field value
in VB Code: sp! = Format(sp! + Left(hid!, 1) - Right(hid!, 1), "##.0")

In other words, take the first digit of the horse's ID, subtract the last digit of the horse's id, and add that to the 3rd field in the raceout.zip

2) odds:
in VB Code:
odds! is the original field value
rshorse!odds = Fix((0.82 - odds!) / odds! * 10) / 10
If rshorse!odds < 0.1 Then rshorse!odds = 0.1

In other words, take (.82-odds)*10 and divide by the odds. Drop any remainder off that, and divide by 10. If the resulting amount is less than .1, then set it equal to .1. Otherwise, this is the amount of the odds column.

3) Post Position:
In VB Code:
hid! = horse ID
ppostemp! = original value in post position from raceout

If Mid(hid!, 1, 1) = 9 Then
ppos% = 49 + ppostemp!
Else
ppos% = 49 + Mid(hid!, 1, 1) + ppostemp!
End If

In other words: if the horse ID starts with "9", then simply add 49 to the value in this field. Otherwise, take 49, add the first digit of the horse ID and add THAT to the value of the field.

There you go. Now you know basically what I know. Good luck. Please don't advertise these too widely, but feel free to use your judgement in helping others out. If you figure out the raceview fields, let me know. I don't even know for sure that they are fields that affect the way the race is run / viewed. But I have some good evidence to that effect, and Mr. Coker thought my hypotheses on that score were reasonable, although not provable.

-- Madman
 
2Madman
      ID: 21020124
      Tue, Jan 15, 2002, 04:34 		WOWWWW. I just got a behind full of info. from Mr. Coker via another guy via Mike. Holy Cow. I'll probably be absent from the boards for the next 24-48 hours as I try to sort through all this stuff and crack some code! Apparently, Mike doesn't even remember how to crack all the fields in the raceout.zip :) He'd have to go back into the code of the off-line viewer, etc. With the information they gave me, I think the better solution (as of right now) will be to apply some serious statistical analysis and code-breaking techniques to the data and see if the answer won't pop out. Based upon what they have shown me, this problem isn't theoretically that complicated; I already know quite a bit about the functions used to calculate stuff (things I learned from my experiments).

Thus, I'm hoping that some basic statistical analysis will finish all the stuff off. Then I'll have to figure out what I want to do with it...
 
3cab
      Donor
      ID: 9115466
      Tue, Jan 15, 2002, 05:55 		Good luck Madman!...You've bamboozled me already!!lol
 
4Madman
      ID: 21020124
      Tue, Jan 15, 2002, 18:12 		Well, I've conclusively proven to myself that the official OFF-LINE viewer uses fields 5-10 of the raceout.zip (the fields I called "raceview adjustment in post #1 above). However, the ON-LINE viewer doesn't!!! This explains why there is such a large discrepancy between what you see on-line versus off!

The Lengths Back at each of the calls in the ON-LINE viewer appear to strictly be a function of the start percentage (as reported in the SSS) and the horse's overall speed number??!!??

I'm a bit worried about this conclusion, because it may mean that it is impossible to construct the on-line lengths back info. from the raceout.zip. I know that Mike calculates Lengths back at 4 different calls for the on-line viewer and uses those calc's. If you could get the same info. from just the start% and speed, then I wonder if he would actually use the LB's at the calls like he does. Seems like you would prefer to use the start% and speed and calculate the stuff from there . . .

Pretty weird. Bad news is that I no longer have a clue about the "functional form" that the lengths back formula takes, although I can identify the variables in it. Could take a lot longer to crack than I thought.