Question on Precision of Figures

[BitPlayer]

TGJB -

I\'m not interested in getting involved in the argument about the relative accuracy of Thorograph and Ragozin figures, but it does bring to mind another question I\'ve been wondering about.  How precise do you think TG figures are?

You seem to acknowledge that there\'s room for error in measuring ground loss (otherwise you wouldn\'t have 3 guys do it for the BC), wind speed and direction, the amount by which the track speed has changed from day to day or race to race, etc.

In statistics, there is the concept of a 95% confidence interval.  As I understand the concept, it acknowledges that you don\'t know for certain what some number is, but you\'ve made your best estimate, and there\'s less than a 5% chance that you\'re off by more than some error factor.  You then express the number as your best estimate plus or minus the error factor.

Based on your experience, what do think the error factor is for TG figures?  Does it vary between sprints and routes?  Between turf and dirt?  From track to track?

Thanks for your time.

[TGJB]

This probably deserves more than a quick answer, but I would say this-- even if all your variant decisions are right, there is play in your figures WITHIN a race because of rounding off (and general innaccuracy) of charting of ground loss, and some small rounding off of the beaten length info. The wind doesn\'t matter that much unless you make unwarranted assumptions about the underlying data and formulas being rock solid, and make dogmatic decisions, rather than use the previous histories of the horses.

Race to race, the more data (solid figure histories) you have, the more solid the figures you assign. In other words, if you have a race with lots of first time turfers, there is more of a chance of getting it wrong, and if it\'s the only grass race that day (or after a rain), it\'s even tougher. You could come up with all kinds of degrees of gray when it comes to this-- the other extreme would be a BC grass race, where you have solid figure histories of horses who run in a tight range, and ground looked at by more than one guy.

I can\'t really give you a range or percentage on this, but a figure can certainly be off by a half point either way within a race, even if you get the variant right. That may change when they go to  electronic tracking in the future, but for now it\'s true for everybody.

[Chuckles_the_Clown2]

Thats a .2 margin of error for TGraph. Which means a possible full point swing, so within that point the election is too close to call. Thats the TGraph margin of error, the Rags margin of error is another matter entirely. If you\'re missing fingers on your counting hand after your tour of Iraq, you might want to take your socks off.