DRF Incorrect Rounding of fraction/final times

[mhrussell]

This has troubled me for some time and I wanted others to comment. I have used BRIS data for the past 4 years but I know BRIS and DRF have a relationship and maybe these rounding errors propagate into the BRIS speed and pace figures as well. I have sent BRIS an email to this effect asking them about this issue. ITS on their website has just posted comments on this inaccurate rounding and this topic has been discussed at the last two Handicapping Magic Seminars that I have attended.
What impact does this have? Anyone really concerned? I myself have had sufficient success with the BRIS figures over the past several years as to think these differences are not significant. But for those who use the raw times; maybe so (this includes HM users).

Comments anyone/everyone?..

[Tom]

My thoughts are "So what?"
If you had a way to use fractional times with great precision, it might matter, but you don't. No one knows or agrees what a beaten length is worth, and if they did, no two chart callers would call 'em the same anyways. Look at some horses' PPs - they run varrying times race to race, so even if each one were prefectly timed and reported and you could know exactly what each horse ran itself (gps?) it all goes out the window wehn you figure daily variants into it. Not to mention track to track.
To me, looking at times in hundredths is a exercise in futility. But some people will not agree with me. That's ok, too, since we play against each other. But when I see two horses that go for the lead, one in 45.1 and the other in 45.3, I see two closely matched horses, and give either one the change to out-pace the other on any given day.
Now, when they round down the PAYOUTS, that's where I get antsy! :mad:

[cj]

Agree with Tom, it doesn't matter much when the BL measurements are so inprecise. That said, I do my software so that any fraction is rounded a little more accurately.

I do the following:

22 flat: 22.10
22 1/5: 22.30
22 2/5: 22.50
22 3/5: 22.70
22 4/5: 22.90

DRF gives anywhere from 22.00 to 22.19 a 22 flat, therefore a possible error of .19 seconds.

At least my method reduces the possible inaccuracy to .10 seconds. This is all because I'm a math geek, not that it really matters in the long run.

[Jed]

Should it matter? Are not all beaten lengths and variants estimations?

[delayjf]

My understnading is the the drf doesn't round up or down but rather truncates down. But why use fifths when hundreds are available?

[VetScratch]

mhrussell,
The times are accurate enough considering how inaccurate the distances travelled are. No two horses in the same race are likely to run the same distance, so figures accurate to hundredths are moot. Until the distance each horse actually runs is measured to a degree of accuracy better than let's say a neck, you can't begin to treat times as if a carefully controlled scientific experiment was conducted.

Even the distance run by wire-to-wire winners is variable due to post positions and lane preferences (influenced by track biases).

[kenwoodallpromos]

even if more exact timing is used, at what distance back does time/lengths back become moot? For example, it is hard for me to belivee that 34 1/4 lengths back would need to be accurate even if timed in 1/100ths!!

[Brian Flewwelling]

I did a little check on the times supplied by BRIS in their "drf" files and compared them to the times supplied by Equibase in their results charts.

All times were given to 2 decimals and were exactly the same.

It is my observation that the 'old' way of reporting times was to the previous fifth, so that 23.39 would be reported as 22 and 1 fifth, and that may be the way they appear in some paper past performances. BRIS' new pp reporter offers the option of hundredths or fifths, and they may round down for the fifths, but i haven't checked.

I agree, when other data is gets to the accuracy of fifths, and develops some real predictive value, somebody may care.

Brian