Our methodology is multifaceted.  First, we use linear regression (involving fractional times and lengths behind) to generate the slope and intercept of the best straight line that models the individual performance.  Slope is a measure of fatigue.  The higher the slope, the longer the time required to negotiate an additional distance.  Slopes will generally fall between 0.9 (slow early-fast late) and 1.2 (fast early-slow late). Intercept is a measure of early speed and is inversely related to the slope.  In this case, the lower the intercept, the greater the relative early speed.  Intercepts also will usually fall between 0.9 (fast early) and 1.2 (slow early). Correlation Coefficient is a measure of how efficiently a horse's speed is expressed throughout the whole race.  Perfect efficiency is equal to a Correlation Coefficient of 1.00000.  Any value below 1.00000 represents less than ideal efficiency, although it is almost always greater than 0.99900.  The very best horses continually display Correlation Coefficients above 0.99990 while lower quality horses may not display Correlation Coefficients that high. We can subsequently use the slope and the intercept for a race to calculate a predicted time at any distance.  These times include a projected ten-furlong time (10f), a projected turn time in a ten-furlong race (10fTT, the time to negotiate the distance between the six furlong and eight furlong markers) and a projected last quarter-mile time in a ten-furlong race (10fLQ).

Next, we calculate Sartin Methodology-based Brohamer pace numbers ("Modern Pace Handicapping", by Tom Brohamer, William Morrow and Company, Inc,. New York, 1991) for the same races.  Evaluating only route races these are limited to a 3Fr (Final Fraction) number which is the speed in feet-per-second from the six furlong call to the finish and a %E (Percent Early) number which is a relative measure of energy used through the six-furlong call.  Speed types display %E figures significantly higher than off-the-pace types, although the absolute numbers are greatly affected by the distance of a race.  We emphasize 3Fr and %E from among the various available Brohamer figures because these are directly related to a horse’s ability to conserve energy and to finish its races with something left in reserve.

The other Brohamer figures include EP (Early Pace), the velocity in feet-per-second to the six-furlong call; SP (Sustained Pace), the average of Early Pace and the Final Fraction, thus relating a horse's speed to the six-furlong call and his ability to finish; AP (Average Pace), the average of EP and SP; 1Fr (First Fraction) and 2Fr (Second Fraction), the average velocities in feet-per-second between each of the first two calls (start to four furlongs and four furlongs to six furlongs); and TE (Total Energy), the sum of EP and 3Fr (the total available energy based on current conditions of distance, surface and track as well as inherent ability).

Finally, we use Performance Figures, based on the same linear regression technology described earlier, as a measure of the absolute quality of the races being analyzed. 

The relevant figures for this year’s leading Derby contenders are here. 

For reference, guideline numbers (all times are in seconds; e.g., 122.34= 2:02.34) for all Derby winners since 1984 plus Spectacular Bid and Triple Crown winners Secretariat, Seattle Slew and Affirmed are:  

 
The figures for the individual Derby winners are:

The following table represents one way, but certainly not the only way, of using pace parameters and Performance Figures as a guide to Kentucky Derby performance.  For each year since 1998, the starters are rank ordered according to the five pace-related parameters derived from Derby prep races at a mile or more on dirt.  They are 1) the fastest 10f; 2) the fastest 10f LQ 3) the fastest 10f TT; 4) the fastest 3Fr; and 5) the lowest %E.  Starters with the top five values in each category pass the initial screen.  In the second screen, only those passing the first screen and which have achieved a PF of -55 or better in a graded Derby route prep race are considered for the win.  To test the usefulness of this analysis, all qualifiers are bet across the board and in exacta combinations.  The results are displayed on the right.  Since 1998, this methodology has identified all seven winners as well as four exactas and two trifectas.  The results demonstrate that horses with exceptional performance in a Derby prep race at less than ten furlongs and which have also shown the ability to finish their races with very good energy are most likely to win.  There is nothing mysterious about the outcome.  However, it is also true that many legitimate contenders have often been overlooked by emphasizing factors that seem not to be as important. 
 

 
The next table presents the best prep race pace parameters for the 2005 Derby entries followed by the same analysis used previously (updated after the post position draw, May 4).
 

Afleet Alex, by the way, is also one of only two Dual Qualifiers among the potential starters, the other being Sun King, which also appears among the pace parameters top five but falls short on Performance Figures.

Bellamy Road, weighted at 110 pounds on the Experimental Free Handicap (EFH) but the 11th best juvenile of 2004 based on The Thoroughbred Times Performance Rates could be considered a "conceptual" Dual Qualifier, a term reserved for three-year-olds with a Dosage Index (DI) of 4.00 or less and which were among the best juveniles of the previous year.  Earlier Derby winners fitting that description include Winning Colors, Sunday Silence, Grindstone, Funny Cide and Smarty Jones.  All were either weighted too low on the EFH or totally left out because of the rules for inclusion.  Bellamy Road's Dosage Figures are well within the historical guidelines and, based on what he has done in 2005, he may have been ranked too low by the EFH.  The Thoroughbred Times assessment of his two-year-old form coupled with his performances in 2005 confirm his extraordinary talent.