You are sitting in an airport bar, impatiently waiting for your flight to Louisville. You strike up a conversation with the guy sitting next to you. He says that he loves the Kentucky Derby. Never misses it. Watches it every year. On television.

You decide that you might have a pigeon ready to be plucked.

You say, "I’ll bet you an evening of drinks that you can’t name two horses that have won the Kentucky Derby by more than two lengths."

Our pigeon begins thinking. You can just smell the smoke. You know that he got Secretariat right away. Everybody knows that one. Click! He has it! He’s sure he’s got you. "You’ve got a bet he says."

He starts by proudly announcing, "Secretariat," and waits for your pain to begin. You smile.

He continues, explaining about all the different possibilities he had considered, and that he finally settled upon Real Quiet as his second selection because he was sure you would remember him. Now he smiles big.

You say, "You lose." He asks, "Why?" That is when you explain that no "horse" as ever won the Kentucky Derby because the races isn’t for "horses." It is for "three-year old colts, geldings, and fillies."

Okay, so while your wheels are turning around who your own personal pigeon for this trap is going to be, let’s get on with the real story.

Although we not typically taken the Kentucky Derby too seriously from a profit standpoint, we find that our annual analysis is generally right on the money.

As with any analysis, we have to start with some races. We decided to use all races from our database which fit the following criterion:

1. three-year old males only
2. dirt
3. distance of 1 3/16 to 1 1/4 miles
4. January through September
5. Handicaps and Graded Stakes
6. Purse value of $100,000 or more

Our search produced 25 such races, including the last 7 Kentucky Derbies. The results are very enlightening, to say the least.

The Tote Board

The tote board tells much of the tale at the Derby. First, let’s look at how the different public choices rank:

Before going on, let’s make sure everyone understands our column labels:

Starts - Number of horses with this characteristic
Pays - Number of winning horses
Pct - Win percentage
$Net - Return for each $2.00 wagered
IV - Impact Value (wins / expected wins)
PIV - Pool Impact Value (wins / pool expectation)
AvPay - Average Payoff
Av Odds - Average Odds of all starters

The fact that the favorites in this type of race have not done well, should come as no surprise. But notice that the second choices have more than taken up the slack.

The critical point here is that a whopping 23 of the 25 races were won by one of the lowest 5 public choices at the race track.

The logical thing to do is to divide the horses into 3 groups:

If we look at the raw odds themselves, we find that the odds groups break nicely into three categories:

In this type of race, the public is usually wrong, picking the wrong favorite. But they are not too wrong. While betting on favorites is usually a bad idea, betting on horses beyond the fifth public choice or above 8/1 is the kiss of death.

And notice that 0.70 PIV? That means that favorites only win 70% of the races that the public thinks they should win based upon how the money in the win pool is distributed.

Early Speed

Not too much to recommend here, accept that true early horses (7 or 8 points) do not typically last at these distances.

Earnings per start has long been used as a yard stick for determining who the better horses are in a given race. In our calculation of earnings per start,we looked for a 10% edge between horses to make an edge in the rankings.

Notice that after the top 5 ranks the IVs drop off drastically, as do the odds. In other words, the public is well aware that horses with greater earnings win more often.

Total Earnings, Last 2 Years

In Dr. Frederick Davis’ manuscript, Percentages & Probabilities (Wildwood Publications) from 1973, he suggested that a good way to measure class was to simply rank the horses by total money earned in the last two years.

This factor deserves some serious attention. First, notice that the first two rankings (and three of the top four) produced a significant profit. But before you get all excited and decide that you can bet the Derby on earnings alone, notice the relationship between Average Pay and Average Odds.

See how the average pay is way out of line with the odds? That points to the fact that these rankings were inflated by some higher than normal winners.

But that isn’t necessarily bad news. Perhaps this is a way for us to point to those few horses that did pay big mutuels.

The way to use total earnings is as an excuse to put back in as contenders some of those longshots on the tote board.

Average Purse Value, or APV as it is usually called, measures the size of purse for which each animal has been competing. For a more in-depth definition, look in Dr. WIlliam Quirin’s Winning at the Races (referenced earlier in this article).

In the table above we have ranked the horses, demanding a 10% edge over the next horse in the rankings. Notice that 23 of the 25 winners were ranked in the top 5.

What if we just looked at the final time ratings of each horse? The first question is, "How do we look at final time?"

We are searching for ways to use final time as a contender selector here. In looking at these 13 different final time factors we have chosen the most effective ranking as a contender selector and ranked the factors by impact value.

In most cases, we broke the contender/non-contender status at the 6th ranking. That is, the unmarked ones above (such as B2L3) include the totals for horses ranked 6th or worse in their field. Wherever a different ranking was more effective we marked it appropriately.

In studying the above table, we do not suggest that "Best 2 Ever" is the factor to use. Notice that we had to drop down to the 7th ranking to use it as an elimination. That is because the 6th-ranked horses actually performed respectably.

Some of the factors in the above table actually allowed us to move up a notch to the 5th ranking. Stand back from this table and look at the order. You will notice a pattern.

The absolute worst way to use Speed Ratings was "Last Race," followed closely by several "Best-of-Last-Somethings." In other words, no single, big race is going to point to a winner in the Derby! It requires multiple big races!

Look at the PIVs. Remember that this is "Pool Impact Value." This means that horses which did not rank in the top 4 for B3L4 (Best 3 of Last 4) won only 11% of the races that the public thought they should have won. This is a very significant finding and should be taken very seriously.

By the way, we feel that using the Beyer number or any other high quality approach to speed ratings should work well. Just remember that if you are using the Beyer number, you must look for at least a two-point edge in the ratings.

One of the other questions that you would logically be asked would be, "Should sprint races be used in computing these rankings?"

Our approach to speed was to use all races, all distances, all surfaces, and to demand a full length of difference in the speed ratings to break a tie.

Speed ratings offer a simple, yet powerful way to pick contenders in the Kentucky Derby. No single race qualifies a horse as a contender. Throwing in one clunker in the last 4 is okay, but more than one is enough to strongly downgrade a horse’s chances.

Early Pace

In studying early pace, we simply looked at the early pace rating earned by each horse, just as we did the speed ratings. Obviously, this is not as easy to apply as the final time ratings mentioned earlier because it takes more work to calculate the ratings. The extra effort may well eliminate an extra horse or two on Derby day.

It appears that best 2 of last 4 races is the best way to apply the early pace factor. Notice that this would have eliminated 134 starters and only 1 winner! This factor was powerful enough that it deserves a more in depth look.