Timeform - Beyer - TG Conversions

[bobphilo]

Ok guys, here are my preliminary findings on converting Timeform ratings to Beyers,  Beyers to T-graph and Timeforms to T-graph. I developed a formula similar to the one used to convert Celsius to Fahrenheit and vice-versa. I found a point of equivalence for both Timeform and Beyer equal to a zero in TG. I then found the difference in the scales so for every X points above or below this point of equivalence you add or subtract Y points to the other scale. For my data I compared the Beyer, TG  or Timeform rating earned by each horse for the same race(running line).
This is based on a sample size of 100 races. There may be slight modifications to the formulas as I incorporate more data. This is a work in progress so please give it a dry run before using it in wagering.
 
Here are the points of equivalence:
(Beyer) 101 = (TG) 0,  (Timefrom) 116 = (TG) 0

Ratio Scales:
TG/Beyer = 1/3,  Timeform/Beyer = 1/1,  TG/Timeform = 1/3

So........
To convert Beyers to TG:
TG = (101-Beyer) divided by 3
Example) If Beyer = 107,  TG = (101-107) / 3  
                          TG = -6/3
                          TG = -2

         If Beyer = 92,  TG = (101-92) / 3
                         TG = 9/3
                         TG = 3

Timeform to Beyer:
Beyer = Timeform – 15
Example) If Timeform = 120,  Beyer = 120 – 15 = 105

Timeform to TG: Just convert Timeform to Beyer and then convert Beyer to TG

Good luck everyone,
Bob

[NoCarolinaTony]

Bob,

Very Interesting calculation. Jerry What do you think? What would you say to your customers if we used this as a proxy.. until ground loss and track variant is calulated?

NC Tony

[bobphilo]

Thanks, Tony. I should add that I did not adjust the Beyer and Timeform figures for weight and ground loss since that is the they way they are published and most commenly used. You may, therefore, see some variations in particular races mainly due to ground loss and weight differentials. That\'s partly why the TG ratings will be the more accurate ones in case of conflicts. My next step is to apply this formula to the Timeform and TG figures for the Dubai races. I strongly suspect the 2 will will be very similar demonstrating the value of the TG ratings for European races and shippers to the U.S.
One of my group members is planning to test the conversions one acouple of races involving horses making their U.S debuts from France and Germany. I\'m curious to see what results anyone here gets in similar situations.
I do want to emphasize that these conversions are not a sustitute for TG ratings. What they do is validate the correlation between them and Timeform, which are the \"gold standard\" for European performance.  

Bob

[Caradoc]

Bob: Clear up a couple of things for me if you can.

1.  When you say that these conversions are based on a sample size of 100 races, are you saying that you derived both the Beyer/TG conversion and the Timeform/TG conversion separately on the basis of 100 race samples?  If so, how many horses are included in each sample of 100 races?  If not, how much overlap is there with respect to the sample pools?  

2.  How did you select the 100 races for inclusion?  

3.  Do we know the methodology that goes into production of the Timeform numbers?

[bobphilo]

Caradoc,

for my Beyer-TG comparison I used 100 running lines from the PP\'s of stakes races which I had handicapped. All the data were part of a matched pair of Beyer and TG ratings given to the same horse for the same race. the 2 sets of data were averaged to get 2 means equivalent to each other. I calculated from the ratio of points assigned for beaten legths at the same distance for both sytems and found that 1 TG point equals 3 Beyer points. Using this ratio brought the TG zero equal to Beyer 101. Any deviations from this point would be counted as a 3 to 1 ratio of Beyers to TG.

For the Timeform - Beyer ratio, I used the Dr. Roman study and a smaller number of my own races to determine that the scale was the same 1/1 ratio for both scales with the only difference being the 15 point higher Timeform standard. Therefore one need only substract 15 from the Timeform rating to get the Beyer. Having done this, in order to get the TG from Timeform, one must first transform the Timeform to Beyer and then convert the Beyer to TG, as given above.
I\'m not totally familiar with the methodology used to obtain Timeform ratings except that since European courses vary so much for different races, even at the same track, and the pace can vary so much from race to race that horses are basically compared to their competetors in terms of beaten lengths in the race they ran using the projection method for the most consistent horses that ran an in form race. Jerry could probably give you a more detailed discription of how timeforms are calculated. My personal experience is that they are very skillfully crafted and accurate.

I\'m falling asleep as I write this due to the lateness of the hour and hope i was able to answer your questions. If not, feel free to ask me for any further clarification you may need tomorrow.

Good night for now,
Bob  

[TGJB]

I really don\'t want to get into this conversation (it\'s not in my interest to improve everyone\'s ability to appraise ability unless they are using TG), but if you use some formula as you describe to translate TimeForm you will go broke. I have written about this here before (search?), and Schuback has in DRF as well-- you can\'t use a straight conversion, because a 115 for a 2yo colt does not equal 115 for a 2yo filly does not equal 115 for a spring 3yo, which does not equal a fall 3yo, or an older horse, etc. If you get a TF manual you can read about how they are working with weight-for-age, and a more complete explanation.

Schuback wrote an article in DRF that gave a sliding formula that could be used to convert to Beyer. Then all you have to do is deal with the weight and beaten length issues, but that\'s another matter.

[bobphilo]

Jerry,

I want to re-emphasize that my Timeform conversions are in no way meant to be used as substitutes for the TG ratings. As I noted, I am aware of the differences in methodologies in making the 2. You take into account certain factors that Timeform does not and vice-versa so there will be variations in individual ratings for the same performance.

I was mainly pointing out that I found that when comparing averages of groups of horses and applying my conversion, that a strong correlation becomes evident between your ratings and Timeform - which are considered the \"gold standard\" for European performances. If anything, this actually shows the validity of the TG ratings for foreign races.

Bob

[marcus]

The convertion is better when it\'s just a matter of one shipper , or  two , that are in and the the question is gauging how the horses off TimeForm \"might\" fit into the race . I talked to Jerry at the Thoro-Graph office about this a couple years ago after TimeForm kindly returned an email which I had sent them inquiring about some of these same issues  . TimeForm sent me a rough conversion chart\'s ( TimeForm from Beyer and Beyer from TimeForm  and they also told me that a sliding scale must be used when the beyer is under 100 ) for 2 yo colts / 2 yo fillies , 3 yo colts ( early season ) / 3 yo fillies  ( early season ) , older horses ( dirt ) , older horses and fillies  ( turf ) , older horses ( dirt ) . TimeForm did caution me similarly to what Jerry is saying in his postings and TF & TG both advised me to be very carefull when handicapping or betting  due to the inexactnness of the convertions . Last spring I attempted to make a couple ratings w / numerical designations ( I won\'t call them numbers ) that weren\'t even close - It would have been best had I just tracked money in the exacta pools ... All Jerry\'s euro numbers are super tight and  for figure handicappers  are the only way to go . I\'ll definately wish you the best with this however and be watching for any developments along these lines .

[bobphilo]

Marcus, I do want to re-emphasize that these conversions are not meant to be "substitutes" or "improvements" on any of the figures they are used for. They are general "ball park' guidelines created for the same purpose as the Timeform people had in mind when they created the conversion information you requested from them. There is a similar conversion table on the Roman's site which Bit-player uses and posted here as well. Though I incorporated general equivalences and the differences in scales, the different ratings measure different aspects of performance and will not be identical even for the same race. I often find then useful in showing comparisons in tightness of fit of the different methodologies. I did emphasize that this is a work in progress, which I've been finding useful for years in my own handicapping and only formalized it as a response to requests sent to me from several posters here

I have discussed this privately with Jerry and we agree that since this is not primarily meant to be used to find patterns with TG figures, that further questions and discussions should be directed to me at my forum at Yahoo.

Thanks for your well wishes. If you have any other questions on further progress, feel free to contact me privately or at my group.

Bob