The LyonScoreŽ information system: how it works

The LyonScore information system is designed to provide statistical support for assessments of the ancestries of mares, specifically in relation to an individual sire. The large books typical of stallions that attract major commercial interest contain mares that, taken together, represent several thousand ancestors within six generations. Our system answers two key questions about each of those ancestors: 1) how many mares with that ancestor in their pedigrees produced foals by the stallion? and 2) how many of that group produced superior runners?

In other words, what is the sire's strike rate with mares representing that ancestor? On that basis, the system discriminates between ancestors that made positive contributions to foals by the stallion from ancestors whose contributions had a negative impact.

This large, comprehensive information system is grounded in a simple theory. Any given stallion, especially in an era of rapidly increasing specialization, is going to sire runners within a certain range of type. Some ancestors of his mates will contribute traits that tend to enrich the sire's own contribution while other ancestors of his mates will contribute traits that are generally out of sorts with the sire's range of type.

The LyonScore profiles

Given the sire's strike rates with all six-generation ancestors of a given mare, the system compiles a profile used for evaluating the cross. Let's take the case of the 1991 mare Heavenly Prize (Seeking the Gold ex Oh What a Dance, by Nijinsky II) for the purpose of describing the contents of the profiles and explaining how we use them to evaluate matings.

Heavenly Prize's two best foals were dual-G1 winner Good Reward and G2 winner Pure Prize, both by Storm Cat and both currently at stud. If you open the Storm Cat-Heavenly Prize profile (in a separate window), you'll see that the profile has a score of 99, shown at the bottom of the block of information in the upper left-hand corner of the page. However, the information from which the score is derived is far more important than the score itself.

Strike Rates: Note that Storm Cat's strike rates with mares representing each individual ancestor of Heavenly Prize is shown to the right of the names of her ancestors. For example, the strike rate 5/19 is shown to the right of Seeking the Gold, her sire. This means that Storm Cat sired foals out of 19 mares with Seeking the Gold in their ancestries and that five of these mares produced superior runners. By the way, we define a superior runner as the winner of an unrestricted stakes or a runner that ran at least second in a G1 or G2 stakes.

Heavenly Prize is one of those five mares mentioned above by virtue of her superior runners Good Reward and Pure Prize. So that no individual mare has an undue statistical effect, she is counted only once, regardless of the number of foals or superior runners she produced by the sire.

Ancestor grades: Based on the relation of a strike rate to the overall record of the sire, one of three letter grades is assigned each ancestor. If the sire's strike rate with an ancestor is significantly greater than the sire's overall strike rate, then the ancestor is assigned a letter grade of "A." If the strike rate is significanly lower than the sire's overall strike rate, then the ancestor is assigned a letter grade of "F." All other ancestors are assigned a letter grade of "C."

Note that Heavenly Prize's dam, Oh What a Dance, has a strike rate of 1/2 and is assigned a letter grade of "C" even though 1/2 (50%) is much greater than Storm Cat's overall strike rate. That's because the assignment of an "A" grade requires that the ancestor be represented by at least two dams of superior runners in order to avoid statistical effects that might not be warranted.

Naturally, the status of an ancestor with respect to a subject sire will change over time as the sire accumulates more superior runners. No matter what the strike rate might be, an ancestor represented by only one dam of a superior runner will be a C-ancestor until such time as the number of mares reaches a threshold level beyond which the letter grade falls to an "F."

Points: The number of "Points" is assigned by subtracting the number of F-ancestors from the number of A-ancestors in the six-generation ancestry of the mare, yielding the balance of ancestors to which the sire has demonstrated a favorable response. If the number of A-ancestors is greater than the number of F-ancestors, then the balance will be a positive number, but it will be negative if the number of F-ancestors exceeds the number of A-ancestors.

To put the number of points in perspective, the information block in the upper-lefthand corner of the profile shows the average points assigned mares that have already produced superior runners (SR) by the sire and the average points for mares that failed to do so. Heavenly Prize is assigned 73 points because, as shown on the top line, she contributes 73 A-ancestors to Storm Cat and zero F-ancestors. The average point total of other mares that produced SRs by Storm Cat was around 37, by comparison.

Two other data items help to provide context for the number of points. Note that only seven of the 153 mares that produced one or more SRs by Storm Cat contributed more points than Heavenly Prize. Also, of the 800 mares that failed to produce SRs by Storm Cat, only two contributed more points than Heavenly Prize.

Score: The "Score" reflects the percentile rank of the subject mare's point contribution, relative to other mares that produced foals by the stallion. Therefore, the lowest possible score is zero and the highest possible score is 100. Heavenly Prize ranks in the 99th percentile of Storm Cat's past mates. As shown in the "Points" analysis, seven mares that produced SR's by him have higher scores than Heavenly Prize while only two mares that failed to do so have higher scores.

Frequently, the score assigned a cross has less importance than the array of strike rates for individual ancestors of the mare. In fact, the scores tend to decline in importance as they approach the mean; that is, very high and very low scores tend to have the greatest significance. The mid-range scores are especially subject to pedigree interpretation relating to the specific strike rates of key ancestors, quality of superior runners, etc.

Consider the case of a mare whose ancestry consists entirely of C-ancestors. Because the points on which the score is based as a percentile are functions of the A- and F-ancestors only, a mare with all C-ancestors would have zero points. In most cases, the score would then be around 50, assuming that as many of the sire's mates have negative points as have positive points, and 50 doesn't seem like a very favorable score.

But who's to say that the C-ancestors that comprise her ancestry might not combine to make very favorable contributions? This is why we've given our information system the capability of generating a sire's strike rates with combinations of two or more ancestors. Such cases underline the priority of facts over preoccupation with the score, which represents those facts in only a very broad way.

Bad profiles

Between Pure Prize (1998) and Good Reward (2001)--and before she slipped in 2000--Heavenly Prize produced a foal named Just Reward, by Deputy Minister, in 1999. That earner of $43,000 broke her maiden and placed twice in three starts--not among the best foals of Heavenly Prize even though Deputy Minister was among the best sires of the time.

The Deputy Minister-Heavenly Prize profile strongly suggests that the cross probably had serious issues likely to limit the potential of the foal.

Deputy Minister's strike rate of 1/18 with Seeking the Gold (sire of Heavenly Prize) isn't necessarily prohibitive, but the background numbers do raise concern that Seeking the Gold's influence isn't right for Deputy Minister. The declining trend along Seeking the Gold's female line (1/25 Broadway, 1/27 Flitabout) is especially disconcerting.

Even worse, Deputy Minister was able to get superior runners from only three of the 63 mares that had Nijinsky II in their ancestries, and the same trend of decline is evident in the female line of Nijinsky II (3/79 Flaming Page, 3/82 Flaring Top). Furthermore, all three of those superior runners were more or less minimally qualified, not one graded SW among them.

In this case, though, the summary information in the upper-lefthand corner says it all. Heavenly Prize contributes minus-15 points, for a score of only 11, and 80 of the 81 mares that produced superior runners by Deputy Minister had point totals that were higher than that of Heavenly Prize.

That Deputy Minister, as good a sire as he was, would be the wrong choice for Heavenly Prize is not otherwise obvious, but the profile makes the case in no uncertain terms. This points out an extremely important function of the profiles. Besides identifying stallions whose demonstrated relation to the ancestry of the mare makes them prime candidates, the profiles are very good at eliminating otherwise meritable stallions that have little chance of bringing the best out of the mare.

Proven sires and unproven sires

Clearly, the LyonScore profiles address the question of a mare's pedigree compatibility with proven sires in a vastly more comprehensive and reliable way than other available measures. Without a doubt, the profiles have immense value for breeders using proven sires. But what about unproven sires?

Keep in mind that the profiles for proven sires are supported by an information system that contains the same data for major sires of the 1980s and 1990s as for contemporary sires. The combinations of ancestors that result from crossing a mare with an unproven sire are tested by sampling this ample population of sires to determine whether their strike rates tend to reflect favorably or unfavorably on any given combination. These database resources are perfectly adapted to the purpose of evaluating how different ancestors interact with one another in a populational sense.

The LyonScore information system introduces a new kind of pedigree analysis that is especially appealing to breeders and buyers who understand the importance of a comprehensive factual basis for the decisions they make.

LyonScore is a registered trademark of Roger Lyons Consulting, Inc.