A Way Of Measuring "Foundation"?

[TreadHead]

One of the many \"arrows in my quiver\" that I try to use for the Derby is an evaluation of foundation.  This is much more an art than a science, but it is rooted in the perception that many horses who do well are coming off patterns that show consistency, repeated low figures, and the absence of large jump-ups to get to one very good race.

Aside from the fact that the analysis is purely a dogmatic look at past figures and does nothing to predict how condition changes may mean the horse is going to run better in the Derby (although history shows that rarely happens), I am certainly open to other critiques/suggestions for improvement.

To measure foundation, I simply pick the 3 best races in the horse\'s history and take an average and standard deviation from those.  This is where the art part comes in, because you have to pick relevant races (for instance, throwing out Pyro\'s last.  But do I use the sloppy track race from last year?  Is that \"relevant\"?)  

The other obvious issue is with a horse like Big Brown, having only run 3 races total.  Having to throw his 2 yr old race into the evaluation may not be \"fair\", but since the concept is foundation, it may lend to the fact that he has little.  In either of these cases, you may want to run the numbers both ways and see what it tells you.

What we are looking for is a horse with both a low average and a low standard deviation.  How low is low on SD?  Well the only horses to win the derby with a SD greater than 2 1/2 TG points from the derby archives are War Emblem (1,1,8 figures) and Charismatic (3.3 3.3 8 which is barely over the limit).

Here is a look at data from last year, horses who were in the 0-1 range coming into the race only.  (I multiplied the raw figures by 4 to deal with whole numbers.  The avg is shown in TG points)

   R1   - R2   - R3   - Total   - Avg -   Dev
SS   -8   6   9   7   0.58   9.07
HS   6   8   8   22   1.83   1.15
Curlin   -2   1   2   1   0.08   2.08
CQ   -4   10   9   15   1.25   7.81
AGS   6   3   9   18   1.50   3.00
ScatD   4   9   9   22   1.83   2.89

Granted this analysis may not have put you on SS as the winner.  But if you are looking for horses to use in the exotics, it lets you know that hard spun and curlin are pretty damn solid.

The results from this year\'s data are much more dramatic.  No matter how you pick races for some of the questionable horses, there are only 2 horses who have an average TG figure less than 3 and a SD of less than 2 1/2 TG points (10.0).  They are easily enough determined for those who have already purchased the data.  Needless to say, those horses will be on all my exotic tickets.

For fun, here is the data from some of the other famous derby performers from the last 10 years using data going into the race, using what I thought were the right 3 races to use.

   R1   - R2   - R3   - Total    - Avg   - Dev
barb   6   15   18   39   3.25   6.24
Bgcat   4   10   12   26   2.17   4.16
giac   20   18   21   59   4.92   1.53
CA   14   13   18   45   3.75   2.65
Funny   6   14   15   35   2.92   4.93
smarty   -7   -15   0   -22   -1.83   7.51
waremb   4   4   33   41   3.42   16.74
monar   15   1   12   28   2.33   7.37
invis   20   23   6   49   4.08   9.07
fupeg   9   19   19   47   3.92   5.77
chars   14   32   32   78   6.50   10.39
menifee   15   15   22   52   4.33   4.04
RQ   15   17   28   60   5.00   7.00
VG   20   22   34   76   6.33   7.57

[Chuckles_the_Clown2]

You may want to revisit this post.

In opposition to your stated theory you have a host of horses with standard deviations exceeding the 2.5 threshold. You have Street Sense with a 9.07 deviation.

The post is indecipherable.

TreadHead Wrote:
-------------------------------------------------------
> One of the many \"arrows in my quiver\" that I try
> to use for the Derby is an evaluation of
> foundation.  This is much more an art than a
> science, but it is rooted in the perception that
> many horses who do well are coming off patterns
> that show consistency, repeated low figures, and
> the absence of large jump-ups to get to one very
> good race.
>
> Aside from the fact that the analysis is purely a
> dogmatic look at past figures and does nothing to
> predict how condition changes may mean the horse
> is going to run better in the Derby (although
> history shows that rarely happens), I am certainly
> open to other critiques/suggestions for
> improvement.
>
> To measure foundation, I simply pick the 3 best
> races in the horse\'s history and take an average
> and standard deviation from those.  This is where
> the art part comes in, because you have to pick
> relevant races (for instance, throwing out Pyro\'s
> last.  But do I use the sloppy track race from
> last year?  Is that \"relevant\"?)  
>
> The other obvious issue is with a horse like Big
> Brown, having only run 3 races total.  Having to
> throw his 2 yr old race into the evaluation may
> not be \"fair\", but since the concept is
> foundation, it may lend to the fact that he has
> little.  In either of these cases, you may want to
> run the numbers both ways and see what it tells
> you.
>
> What we are looking for is a horse with both a low
> average and a low standard deviation.  How low is
> low on SD?  Well the only horses to win the derby
> with a SD greater than 2 1/2 TG points from the
> derby archives are War Emblem (1,1,8 figures) and
> Charismatic (3.3 3.3 8 which is barely over the
> limit).
>
> Here is a look at data from last year, horses who
> were in the 0-1 range coming into the race only.
> (I multiplied the raw figures by 4 to deal with
> whole numbers.  The avg is shown in TG points)
>
>    R1   - R2   - R3   - Total   - Avg -   Dev
> SS   -8   6   9   7   0.58   9.07
> HS   6   8   8   22   1.83   1.15
> Curlin   -2   1   2   1   0.08   2.08
> CQ   -4   10   9   15   1.25   7.81
> AGS   6   3   9   18   1.50   3.00
> ScatD   4   9   9   22   1.83   2.89
>
> Granted this analysis may not have put you on SS
> as the winner.  But if you are looking for horses
> to use in the exotics, it lets you know that hard
> spun and curlin are pretty damn solid.
>
> The results from this year\'s data are much more
> dramatic.  No matter how you pick races for some
> of the questionable horses, there are only 2
> horses who have an average TG figure less than 3
> and a SD of less than 2 1/2 TG points (10.0).
> They are easily enough determined for those who
> have already purchased the data.  Needless to say,
> those horses will be on all my exotic tickets.
>
> For fun, here is the data from some of the other
> famous derby performers from the last 10 years
> using data going into the race, using what I
> thought were the right 3 races to use.
>
>    R1   - R2   - R3   - Total    - Avg   - Dev
> barb   6   15   18   39   3.25   6.24
> Bgcat   4   10   12   26   2.17   4.16
> giac   20   18   21   59   4.92   1.53
> CA   14   13   18   45   3.75   2.65
> Funny   6   14   15   35   2.92   4.93
> smarty   -7   -15   0   -22   -1.83   7.51
> waremb   4   4   33   41   3.42   16.74
> monar   15   1   12   28   2.33   7.37
> invis   20   23   6   49   4.08   9.07
> fupeg   9   19   19   47   3.92   5.77
> chars   14   32   32   78   6.50   10.39
> menifee   15   15   22   52   4.33   4.04
> RQ   15   17   28   60   5.00   7.00
> VG   20   22   34   76   6.33   7.57

[TreadHead]

> there are only 2
> horses who have an average TG figure less than 3
> and a SD of less than 2 1/2 TG points (10.0).

Remember, I multiplied the TG figures by 4, so the SDs are also multiplied by 4.  Therefore, a SD of 10=2.5 TG points.

[shanahan]

Thead - pls explain further..the SD?  and who are the two you reference?

[TreadHead]

Sorry, guess I should have tried to make this a little more user friendly, my fault.

SD is the Standard Devation (in terms of the whole numbers I used) of the horse\'s 3 best (and/or most relevant) races.  The concept is that a lower standard deviation between a horse\'s 3 best figures represents a solid \"foundation\".

I choose to work with whole numbers as a matter of personal preference, and that may have confused things.  But basically what the data says is horses with a 3-race standard deviation of less than 2.5 TG points (or 10 after I multiplied by 4) and who are also fast tend to do better than the \"fast\" horses with a larger standard deviation.

[Chuckles_the_Clown2]

ok, i missed the multiplication. It makes sense now.

TreadHead Wrote:
-------------------------------------------------------
> > there are only 2
> > horses who have an average TG figure less than 3
>
> > and a SD of less than 2 1/2 TG points (10.0).
>
> Remember, I multiplied the TG figures by 4, so the
> SDs are also multiplied by 4.  Therefore, a SD of
> 10=2.5 TG points.

[Chuckles_the_Clown2]

Well, looking at it again briefly, reveals I still don\'t understand it. The a negative effort seems to impact the whole thing immensely. Aren\'t you getting way too busy with numbers? Thats the problem with figures alone, they stand in a vacuum.

TreadHead Wrote:
-------------------------------------------------------
> One of the many \"arrows in my quiver\" that I try
> to use for the Derby is an evaluation of
> foundation.  This is much more an art than a
> science, but it is rooted in the perception that
> many horses who do well are coming off patterns
> that show consistency, repeated low figures, and
> the absence of large jump-ups to get to one very
> good race.
>
> Aside from the fact that the analysis is purely a
> dogmatic look at past figures and does nothing to
> predict how condition changes may mean the horse
> is going to run better in the Derby (although
> history shows that rarely happens), I am certainly
> open to other critiques/suggestions for
> improvement.
>
> To measure foundation, I simply pick the 3 best
> races in the horse\'s history and take an average
> and standard deviation from those.  This is where
> the art part comes in, because you have to pick
> relevant races (for instance, throwing out Pyro\'s
> last.  But do I use the sloppy track race from
> last year?  Is that \"relevant\"?)  
>
> The other obvious issue is with a horse like Big
> Brown, having only run 3 races total.  Having to
> throw his 2 yr old race into the evaluation may
> not be \"fair\", but since the concept is
> foundation, it may lend to the fact that he has
> little.  In either of these cases, you may want to
> run the numbers both ways and see what it tells
> you.
>
> What we are looking for is a horse with both a low
> average and a low standard deviation.  How low is
> low on SD?  Well the only horses to win the derby
> with a SD greater than 2 1/2 TG points from the
> derby archives are War Emblem (1,1,8 figures) and
> Charismatic (3.3 3.3 8 which is barely over the
> limit).
>
> Here is a look at data from last year, horses who
> were in the 0-1 range coming into the race only.
> (I multiplied the raw figures by 4 to deal with
> whole numbers.  The avg is shown in TG points)
>
>    R1   - R2   - R3   - Total   - Avg -   Dev
> SS   -8   6   9   7   0.58   9.07
> HS   6   8   8   22   1.83   1.15
> Curlin   -2   1   2   1   0.08   2.08
> CQ   -4   10   9   15   1.25   7.81
> AGS   6   3   9   18   1.50   3.00
> ScatD   4   9   9   22   1.83   2.89
>
> Granted this analysis may not have put you on SS
> as the winner.  But if you are looking for horses
> to use in the exotics, it lets you know that hard
> spun and curlin are pretty damn solid.
>
> The results from this year\'s data are much more
> dramatic.  No matter how you pick races for some
> of the questionable horses, there are only 2
> horses who have an average TG figure less than 3
> and a SD of less than 2 1/2 TG points (10.0).
> They are easily enough determined for those who
> have already purchased the data.  Needless to say,
> those horses will be on all my exotic tickets.
>
> For fun, here is the data from some of the other
> famous derby performers from the last 10 years
> using data going into the race, using what I
> thought were the right 3 races to use.
>
>    R1   - R2   - R3   - Total    - Avg   - Dev
> barb   6   15   18   39   3.25   6.24
> Bgcat   4   10   12   26   2.17   4.16
> giac   20   18   21   59   4.92   1.53
> CA   14   13   18   45   3.75   2.65
> Funny   6   14   15   35   2.92   4.93
> smarty   -7   -15   0   -22   -1.83   7.51
> waremb   4   4   33   41   3.42   16.74
> monar   15   1   12   28   2.33   7.37
> invis   20   23   6   49   4.08   9.07
> fupeg   9   19   19   47   3.92   5.77
> chars   14   32   32   78   6.50   10.39
> menifee   15   15   22   52   4.33   4.04
> RQ   15   17   28   60   5.00   7.00
> VG   20   22   34   76   6.33   7.57

[spa]

Treadhead,sorry but your whole concept is flawed. The old school formula is, grade 1 winner at 2, three starts before the big dance.

[TreadHead]

LOL, well that formula is extremely old school, given it does not apply to more than half the winners in the last 10 years.  The data clearly shows you do not need to be a Grade 1 winner at 2, and in many cases, you dont even need to have tried a Grade 1 at 2.  So the number of races and the condtions defines \"foundation\" and not how fast each race was?  You have it backwards IMO, the effort involved is the most important factor of building foundtaion.