Chef-de-Race: Lost Code
dkb/br.h., 1984 (Codex-Loss or Gain, Ack Ack)
(September 18, 2010)

Chef-de-Race: Codex
ch.h., 1977 (Arts and Letters-Roundup Rose, by Minnesota Mac)
(September 18, 2010)

Many people believe that Dosage is a breeding system or pedigree handicapping angle. They are wrong, although the results from Dosage studies can be applied to both breeding and handicapping as part of a comprehensive approach.  In reality, Dosage is simply a pedigree classification system that facilitates the study of aptitudinal type with a focus on how to measure it, how it relates to real-world performance and how to track its evolution in the Thoroughbred over time.  The term "aptitudinal type" refers to the inherent characteristics of a horse that define a performance niche within which it should most comfortably fit.  Recognition of those characteristics is based on an assessment of the speed/stamina contributions inherited from selected ancestors identified as prepotent for those qualities.  Aptitudinal type covers the range from early-maturing dirt sprinters to late-developing turf marathoners and everything in between.  For a more detailed explanation of the link between Dosage and aptitudinal type, see here.  Calculated numbers known as Dosage figures locate where on that performance spectrum a pedigree lies.  They also allow us to measure how quickly and in what manner aptitudinal type in the Thoroughbred is changing. 

Rarely has the change in aptitudinal type from one generation of sires to the next been as dramatic as that observed between classic winner Codex and his Grade 1-winning son, Lost Code.  As an illustration, the average winning distances (AWD) in North American open stakes races for runners by Codex or out of Codex mares are 8.95 furlongs and 8.50 furlongs, respectively.  The  comparable figures for Lost Code are 7.06 and 6.37 furlongs.  In other words, the preferred competitive distance of runners sired by father and by son differs by about a quarter mile.  In a sport where a horse's optimum distance range often is measured in fractions of a furlong, a quarter mile separation between father and son indicates a huge shift in aptitudinal type.

Equally dramatic are the differences revealed in the following graphic showing the percentage of progeny stakes wins in sprints (less than 8 furlongs), middle distance races (between a mile and a mile-and-a-sixteenth) and routes (at least a mile-and-an-eighth) for each of the two stallions.  The reversal in type is virtually complete within a single generation.

Codex was a very good race horse, winning two other Grade 1 events in addition to the Preakness Stakes (G1) as a three-year-old, The Hollywood Derby and the Santa Anita Derby.  That year, 1980, he was ranked third best three-year-old colt behind champion three-year-old Temperence Hill and champion sprinter Plugged Nickle.  He may be best remembered for a bumping incident at the quarter pole in the Preakness when, while in the lead, he and jockey Angel Cordero, Jr. forced rallying Kentucky Derby winner Genuine Risk very wide into the stretch and out of contention.  The stewards denied a foul claim against Codex and the decision was later upheld by the Maryland Racing Commission on appeal.

Codex went to stud at Tartan Farm in Ocala but died tragically after siring just three crops.  However, from only 106 foals he got 10 stakes winners (9%) including Grade 1 winners Lost Code, Badger Land (which went on to become one of South Africa's premier sires, leading the general sire list twice) and Coup de Fusil.  In addition, his runners achieved an extraordinary Average-Earnings-Index (AEI) of 3.28, a figure superior to that of A.P. Indy or Storm Cat and better than any sire among the current North American leaders in 2010.

Lost Code, out of Loss or Gain, by Ack Ack, was among the leading runners of his generation, winning fifteen times in 27 starts over three years with earnings in excess of two million dollars.  He won the Grade 1 Arlington Classic at three and the Grade 1 Oaklawn Handicap at four, as well as seven other graded stakes.  He stood first at Vinery in Kentucky and then at Legacy farm in Virginia where he died prematurely at age 17 in 2001.  During his relatively short stud career he sired at least 53 stakes winners (9% from foals) including the multiple Grade 1-winning, top class sprinting mare, Kalookan Queen.  His daughters have produced champion and Breeders' Cup Sprint (G1) winner Squirtle Squirt, Japanese champion miler Hat Trick, Grade 1-winning sprinter Fabulous Strike, Grade 1 sprint winner and Arkansas Derby (G2) winner Gayego, multiple graded stakes-winning sprinter Rite Moment and multiple graded stakes-winning sprinter/miler Desert Code.  Lost Code ranked among the top 100 leading North American broodmare sires every year between 2001 and 2009

The AWD of North American open stakes races won by Lost Code is 8.91 furlongs while that for Codex is 9.17 furlongs.  Below is a summary displaying their AWDs as a runner, a sire and as a broodmare sire of North American open stakes winners.

Despite the similarity of their AWD on the track, the difference in their distance influence on succeeding generations is striking.  This is an excellent illustration of a basic premise of Dosage methodology, first articulated by Varola.  He noted that the characteristics transmitted by chefs-de-race at stud are not necessarily those they expressed on the track themselves.  This concept is extremely important when trying to observe prepotence for aptitudinal type.

Recently, when Unbridled was named a chef-de-race, we introduced a new metric to help identify prepotence for type.  The intention was to determine whether the winning distance profile of an individual sire's runners differs significantly from an opportunistic pattern that merely reflects the average distance distribution for the runners of all sires.  Large deviations from the average suggest a greater predictability of aptitudinal type transmission, a key component of prepotence.  The following graphics clearly show that both Codex and Lost Code have profiles greatly skewed from the average.

We can verify the discrepancy between Lost Code's notional type contribution (i.e., without invoking a prepotent influence of his own) and the real-world performance of his close up descendants through an examination of Dosage figures.  The data derived from 69 North American open stakes races in which Lost Code is either the sire or broodmare sire reveal an AWD of 6.76 furlongs with an average Dosage Profile (DP) of 4.80-3.78-6.00-0.19-0.59, equivalent to an average Dosage Index (DI) of 3.28 and an average Center of Distribution (CD) of 0.74.  Based on the results of over 25,000 races since 1983, the predicted average DI and CD for North American open stakes winners at 6.76 furlongs are 4.79 and 0.98, respectively.  Accordingly, Lost Code's contribution to succeeding generations currently fails to capture the the additional speed he predictably transmits.  His present contribution to the DP of his foals is 0-2-3-0-1 (equivalent to DI 1.40 and CD 0.00) which hardly reflects his attributes as a consistent sire and broodmare sire of high class sprinters and sprinter/milers.  The objective, then, is to bring the figures for descendants of Lost Code back in line with those of the general population having similar performance characteristics.

By assigning Lost Code as a split Brilliant/Intermediate chef-de-race and Codex as a split Intermediate/Classic chef-de-race, Lost Code's revised figures are average DP 11.06-13.17-9.13-0.72-0.59, average DI 5.07 and average CD 0.96, close to the predicted figures.  His revised contribution to the DP of his descendants is now 8-14-7-0-1 (equivalent to DI 5.67 and CD 0.93) as a sire and 4-7-3-0-0 (equivalent to DI 8.33 and CD 1.07) as a broodmare sire.  These are more realistic and they parallel the decrease in the AWD for Lost Code between his first- and second-generation influence. 

Similarly, Codex's figures change from average DP 3.64-3.32-4.36-0.50-1.89, average DI 2.25 and average CD 0.31 to average DP 3.64-10.46-11.50-0.50-1.89, average DI 2.63 and average CD 0.47.  The predicted figures are average DI 2.35 and average CD 0.54, reasonably close to the new figures.  Additionally, the figures for Badger Land as a sire and broodmare sire change from average DP 2.95-4.55-6.41-1.00-0.91, average DI 2.03 and average CD 0.35 to average DP 2.95-10.55-12.41-1.00-0.91, average DI 2.46 and average CD 0.47.  The predicted figures at his AWD of 8.99 furlongs in North America are average DI 2.20 and average CD 0.51, confirming the new figures are a significant improvement.

Having brought the figures for Codex and his sons satisfactorily in line with those for the breed at large, we are confident in assigning Codex to the Intermediate/Classic categories and Lost Code to the Brilliant/Intermediate categories as the 212^th and 213^th chefs-de-race.