For the broadest spectrum of yearling buyers, the sale of choice is the world's largest yearling sale, Keeneland September. This was not always so.

Until a few years after the collapse of the 1980s bloodstock market bubble, the majority of buyers at the prestigious selected Keeneland July, Fasig-Tipton Saratoga, and Tattersalls Houghton sales completely ignored the horses offered at Keeneland September and did not even attend. The assumption of those buyers, who saw themselves as interested in only the best-quality racing stock available, was that the horses available at Keeneland September were essentially culls.

From its inception as an independent sale in 1951 to the end of the 1980s, that was basically true. Traditionally, the majority of Kentucky's commercial breeders nominated all their yearlings to the Keeneland July sale, and Keeneland chose the ones they wanted for their premier showcase, with the selection process more heavily weighted toward pedigree over conformation than is now the case. The rejected yearlings were automatically eligible for Keeneland September with no additional nomination fee, and those summer-sale rejects made up the bulk of the yearlings on offer.

Of course, Keeneland September also usually included those few yearlings forced to withdraw from Keeneland July by minor injuries and many of those that failed to meet their reserve prices, but the latter too were essentially culls.

That traditional stratification began to change gradually in the mid-1980s. The Maktoum brothers' saturation-bombing raids on Keeneland July and Saratoga made many longtime American buyers attending those sales feel they simply could not afford those venues. Those buyers began to turn their attention toward the less-expensive offerings available at Keeneland September.

When 1987 tax code changes made owning racehorses even more costly, more buyers went in the direction of Keeneland September. At almost the same time, the Maktoums drastically reduced their spending as their stud farms began to produce runners to fill their racing stables. As a result of all these factors, official average at Keeneland July plummeted from a then-record $544,681 in 1984 to $233,325 in 1994.

Many of Kentucky's commercial breeders may be traditionalists, but they are smart enough to figure out that it makes more sense to sell their best products in the market with the most buyers. By the early 1990s, several large breeders began to hold back some or all of their best yearlings to sell at Keeneland September. With plenty of buyers competing for what once had been bargains, prices began to rise in 1993-the first segment of the fall yearling market that began the revival that has continued to this day.

Racetrack results

As shown in Table 1, these changes in buyer and consignor focus predictably resulted in higher prices, immediately followed by an increase in the number of horses on offer. The number of horses sold at Keeneland September declined by about 400 between 1990 and '93. When average rebounded 23% to $35,168 in 1993, total horses sold immediately jumped back up by 313 horses the following year.

In addition to sales statistics, Table 1 details the racetrack performance (includes data from all countries in the Jockey Club database) of every graduate of the sale from 1990 through 1997, the most recent year for which racing data are reasonably complete. The 1997 sale, of course, was for foals of '96, now four-year-olds. The majority of those foals that will win stakes or graded races have already done so, though figures for that sale will doubtless still improve slightly over the next couple of years.

Racing data for the 22,305 horses sold at Keeneland September during those eight years show that the sale is indeed a very good place to buy racehorses. During that period, 1,468 (6.6% of the total) stakes winners were sold at Keeneland September, including 454 graded stakes winners (2%). Over the same time period, the comparable percentages for Keeneland July were a yearling-industry high 11.1% and 5.9%.

The difference, of course, is that yearlings at Keeneland July during the period cost an average of $285,783, while the average yearling at Keeneland September costs $38,390.

In addition, buyers overall at Keeneland July did not come anywhere close to making money at the racetrack, posting almost a $300- million deficit. Keeneland July buyers may well have made up that deficit through residual value, but Keeneland September buyers did not have to wait that long. The $856-million expended on Keeneland September yearlings yielded purse earnings of more than $1-billion, with a substantial amount more to come from those four-year-olds and up still running. Even though training costs will put that figure into the red, the average Keeneland September yearling from 1990-'97 did relatively well on the track based on sale price.

The bad news for prospective Keeneland September buyers may be that, in a sense, the sale's best days are already past. In terms of racetrack performance, 1993 was easily the best year overall in the sale's history. That sale-the smallest in number sold of the decade-produced the highest number of stakes winners (201), second-highest number of graded stakes winners (65), and the highest percentages of stakes and graded stakes winners of any year (2.6%). That sale also produced the highest Standard Starts Index (SSI), a measure of earning power adjusted for inflation, and the highest ratio of average sales price to average earnings (1.44).

That sale did include some very high-class runners, such as dual Breeders' Cup Mile (G1) winner Da Hoss, French champion and English classic winner Pennekamp, Grade 1 winners Gal in a Ruckus, Star Standard, and Kentucky Derby (G1) runner-up Tejano Run. But that group is certainly no better and probably not collectively as good as the best horses from the 1994 sale: champion Hidden Lake, Belmont Stakes (G1) winner Editor's Note, and Grade 1 winners Cara Rafaela, Capote Belle, Formal Gold, Honour and Glory, and Will's Way.

More horses, less quality

As mentioned earlier, the 1994 sale had 313 more horses than the sale in 1993. The sale's version of a midriff bulge grew to 464 more horses the following year, and in 1999 more than 3,000 horses-an increase of more than 500 yearlings from 1993-were sold at Keeneland September.

That increase in number of horses sold necessarily dilutes the percentages of stakes winners and graded stakes winners sold; but curiously, the actual number of stakes winners and graded stakes winners sold appears to be declining as well.

Surely even if all 500 additional horses were less well-bred than the 2,494 sold in 1993-an unlikely supposition given the way consignors have shifted better-bred stock to the sale-sheer numbers should have led to an increase in the number of winners at the highest level. As noted earlier, the 1996 and '97 sale years will doubtless add a few stakes winners and graded stakes winners, but the chances of matching the numbers posted in 1992-'95 appear to be slim. This is not to say that more recent Keeneland September sales have not produced top-class racehorses. The 1997 sale included Lemon Drop Kid (arguably the best horse in the country now), Menifee, Yes It's True, and Marley Vale, among others.

The percentage of top winners, however, has clearly declined despite the unarguable fact that more and more consignors are retaining what they perceive as their best yearlings for September.

It is interesting to note that average win distance for graduates of the sale appears to be climbing steadily, a presumed reflection of the higher number of classically bred yearlings now reserved for the sale.

Table 2 divides the global data presented in the totals line of Table 1 into market segments. Seven-figure yearlings were uncommon at Keeneland September until th