09/16/2009 10:30AM

Fractionally speaking


I've been getting numerous posts and questions about the methodology of the pace figures, and specifically why/how a faster-than-par pace that slows to a below-par final figure doesn't necessarily translate into pace figs that are higher than the final fig.

 So at the risk of turning this blog into a forum on pace figures, let me explain the differences in the version you see in Daily Racing Form compared to the versions you find elsewhere.

At Belmont Park, the five-year average raw clockings for a $10,000 claiming race at 6 furlongs are 22.6, 45.9 and 1:10.8.  At one mile they are 23.4, 46.7, 1:11.8 and 1:37.9.   Many pacefig systems are tied directly to par times, with $10,000 pars set at a figure of 100.  That way - removing track variants from the equation for this example - each of those fractional times would correspond to a 100 pacefig.  And if a one-mile race is run at the par final time but the fractions are, say, 23.0, 46.3 and 1:11.4, you might see a pace figure line of 102-102-102-100.

These types of pace figures have advantages.  Mainly, they are easy to use.  At one glance, you can see the pace of the aforementioned one-mile race was two lengths faster than par. You don't need a separate "race shape" line. There is something to be said for simplicity.

Unfortunately, that system can also be misleading when comparing different distances at different tracks.  A horse that ran fractions two lengths slower than par at 6 furlongs might come up with a lower pace figure than a horse that ran two lengths faster than par at a mile, despite the fact the 6-furlong half-mile clocking was 2/5 faster and clearly superior pace-wise. And an added twist at Belmont Park is that fractions for one-mile races are artificially faster because of the long straightaway backstretch run. 

For the best accuracy, my belief is that pace figures should directly correlate to final figures.  In other words, they should be based on the same parallel-time chart values.  This isn't as numerically handy and neat, since in our scale, one length at a quarter-mile equals 3.25 points and a length at 6 furlongs equals 1 point.  But I want to know exactly how the speed of a :44.8 half in a 7-furlong race at Lone Star Park compares to a :45.4 half in a 6-furlong race at Pimlico, and I don't want a pace figure that rates one higher than the other just because it may be faster-than-par.

  Using our pacefig scale and parallel-time charts - and again, excluding any track variants - those raw 6 furlong average times for Belmont $10,000 claimers would translate into an unadjusted pacefig line of 83-88-91, and at a mile it would be 70-81-86-93.  Why do I say unadjusted?  Because to enable Belmont figures to be compared to those of other tracks, I use an adjustment based on how Belmont pacefigs compare to those of other tracks at every distance and class level.  And Belmont fractions tend to be much slower than average. Thus when that adjustment is added, those Belmont 6 furlong figs would wind up at 92-92-91 and the one-mile figs at 82-88-90-90.  This tells you what you need to know: that the horses in the 6-furlong race were running faster at the quarter and half, as you would expect, but also that in case of a turnback in distance, the 6-furlong internal pacefig in the mile race compares favorably to the final figure in the 6-furlong race.

  As a rule, the longer the race, the slower the pacefigs. Combined with the values of the parallel-time chart, this explains (I hope) why it is possible for a race such as Rachel Alexandra's Woodward to have a slightly faster-then-par pace even though the quarter- and half-mile pacefigs themselves are lower than the final fig. 

  Bottom line: we didn't want to sacrifice accuracy for simplicity, so the pacefigs themselves do not reflect whether the pace of a race is faster or slower than par.  We included the Race Shape line to the left of the past performances to help illustrate that, with positive numbers for faster-than-par and negative numbers for slower-than-par.