In Not All Favorites Are Created Equal – Part 1, I discussed the difference between strong and false favorites. I also introduced the concept of the qualifier, or a horse that shows most of the criteria used to judge the favorite and how the presence of a qualifier also impacts the chances of the false favorite.
To summarize from Part 1, the breakdown of favorites by type:
While I have established that there are different types of favorites, are all strong or false favorites still considered equal?
The answer is no.
Favorites can be further broken down by a factor I developed called the “Favorite Likelihood Factor”, or FLF for short. The logic here is simple. It is based on data relating to how morning line favorites fared under different conditions. From this, I created a set of factors that correspond to the likelihood of a favorite winning under different conditions.
Conditions considered when computing the FLF (race dependent):
- Field Size
- Race Type / Distance / Surface
Conditions considered when computing the FLF (entrant dependent):
- Morning Line Odds
- Layoff / Recency
- Speed Figure Ranking
- Class Rating of the horse
I use a series of positive and negative values when evaluating each of the seven criterion above. The FLF is adjusted for scratches and changes since these will impact several of the criterion. Once totaled, if the FLF has a factor of 0, it implies that the favorite has the same chance of winning as the average favorite. When the FLF is a negative value, the favorite has less of a chance of winning. Likewise, the higher the FLF value, the greater chance the favorite has of winning.
The following graph shows the correlation of the FLF to the favorite’s win percentage in non-maiden and non-two-year-old races. I have included the coefficient of determination (R-squared) values to show how well the factors fit a linear regression line (the closer the R-squared value is to 1.0, the more predictable the relationship between the factors is).
Overall, we can see that the data follows a linear pattern. The FLF of 0 has all favorites in this sample hitting at a rate of 30%. As the FLF value increases into double digits, the favorite’s win percentage approaches 50%. Likewise, as the FLF decreases to negative double digits, the win percentage plummets to about 20%. The R-squared value indicates a very predictable relationship between the FLF and the favorite’s win percentage.
If we look at only strong favorites, which are a small sample of the overall favorite data, you can see that at the FLF value of 0, the win percentage is about 35%. By virtue of having more positive factors in their favor, strong favorites tend to have higher FLF values in their races than false favorites. As you can see, as the FLF gets higher, the winning percentage exceeds 50%. The smaller sample size of strong favorites in this data yields a less predictable fit of the data in linear regression (lower R-squared value).
False favorites with no qualifiers make up the largest subset of data in this study and their statistics closely mirror the overall results since they comprise the bulk of the data. This subset ranges the full spectrum from the lowest to the highest FLF values. In many cases, the difference between the strong and the false favorite may only be one item, while the race exhibits positive FLF factors.
False favorites with qualifiers are a smaller subset of data. Here you can see that being a false favorite with a horse that would qualify as a strong favorite if it was the morning line favorite yields a similar linear pattern. The data towards the lower and upper FLF figures starts to skew the R-squared factor. If we focus in on the range from -2 to 5 the data appears much more predictable.
So what does this tell us?
By coupling factors that are both race- and entrant-specific, you can gauge the increased (or decreased) chances that a morning line favorite has in today’s race before you start your in depth race analysis.
Where do I get these figures?
On my new website I will soon have reports available for select races that will include the Favorite Likelihood Factor (FLF), Pace-Based Speed Ratings (PBRs), Ability Ratings (ARs) and will show the Quality Starts over the horse’s past performances.