Same Effort

[horsegoer]

One question I have regarding the numbers. How is it possible that Spider Rock in the 6th at Saratoga yesterday runs four basically identical races in a row last year. It seems almost impossible that a horse could do that. I have also seen this with other horses on the Thoros.

[ronwar]

I think I\'ve heard this argument before, but do not really understand.  IMO the number is an performance figure, which means the horse performed at or about the same level for those four races you speak of.  Same surface, same distance, roughly same competition, roughly same finishes.  Why would you expect it to to be different?

[miff]

\"roughly same competition, roughly same finishes\"


Ron,

...if ROUGHLY the same, then how can the numbers be EXACT? The fig maker is in fact saying the figs are NOT roughly the same,he is saying they are exact, which,is highly unlikely.

Don\'t remember the sheet in question, but whether the figs are exact or very close is minutia.Projection does lend itself to back fitting on occasion.


Mike

[ronwar]

Hey Miff,

I just took a quick look at the race in question in the Redboard room.  We Know nothing is exact about creating figures considering wind, surface, wideness on turns, run ups.  You can get close, but exact?  Nobody can do that.  So IMHO when I look at a TG figure, I know its the figure makers best guess taking into consideration all the tools he uses including projecting.  As long as you\'re comparing apples to apples (a TG 9 to a TG 6) then that is the best you can hope for.

\"...if ROUGHLY the same, then how can the numbers be EXACT?\"

Maybe I\'m way off, and it wouldn\'t be the first time, but when I think of a performance figure I view it as a number that quantifies the horse performance.  What\'s so hard to believe about a healthy athlete performing at a consistent level versus similar competition at the same distance?

[miff]

Ron,

Agree,no problem with \"performing\" at a consistent level.Performing exactly 4 races in a row, highly unlikely.


Mike

[BitPlayer]

My sense from reading this board and from trying to anticipate some of TGJB\'s figures is that, in the final fine-tuning stage of assigning a set of figures to a race, he prefers to give at least one horse in the race (and preferably more) a figure that exactly pairs a prior figure.  The phenomenon that Horsegoer observes is a by-product of that approach.

Any method of making figures has to have a way of dealing with the need to assign precise figures in the midst of unavoidable uncertainty.  I think that the only test of whether TG\'s approach is better than a more subjective approach (e.g, this horse got a more trouble-free trip today and therefore probably should receive a slightly better figure) or a more conservative approach (prejudiced towards keeping the variant constant across multiple races) is how well the figures hang together over the long term.

[fkach]

I think the complexities of making speed and pace figures (now that there is also a Race Shape product) are so numerous, this debate sort of misses the point. No one makes perfect figures. As long as they tend to isolate the contenders, non contenders, and standouts accurately, debates about a length or less here or there are probably moot. I suppose some people might argue that their pattern reads are dependent on very small differences like that, but I find it hard to believe that differences like that mean more than a whole bunch of other important factors that determine the outcome between horses of similar ability (not to mention the odds).

IMO, the only time these debates matter much is when the difference between various sources is so large, an entirely different conclusion could be reached.

[bobphilo]

Actually, there is nothing surprising of a given horse among thousands getting 4 identical figures from a probabilities point of view. Remember there are figures being made for thousands of horses. For example, it would be highly unlikely for one person tossing a coin to get 10 heads in a row. However if there are thousands of tossers, it would be very likely that a number of them  would get ten heads, just like any other number sequence. And that’s with either combination being equally likely. With figures for the same horse running under the same conditions, the probable range would be much narrower so that even more matching patterns would occur.

In terms of predicting future figures, fkach is right, the figures are never completely precise, so that given a standard error; it doesn’t matter whether the figures are identical or similar

Bob