MONDAY PICKS

[formula_2002]

THESE PICKS ARE BASED ON 8760 BACK FITTED RACES (PART OF A 165,889 HORSE DATA BASE)
RETURNING 22% WINNERS AT A 2% LOSS. PICKS AR E 100% COMPUTER GENERATED.
FINAL ODDS MUST EXCEED THE MIN. ODDS
UNACCOUNTED SCRATCHES MAY AFFECT PICK SELECTION.

RESULTS SINCE THAT TIME;
122 RACES, 22% WINNERS 3.6% LOSS



TRACK DATE RACE # H PG# ML MIN.ODDS A B PI 'E' PI 'E'-'E' WNET DBT WIN N230 N231 N232 N233
TUP20031202 2 WOLFMON 4 2.50 2.27 0.00 1 0
TUP20031202 3 FRENCHI 4 2.00 2.91 0.00 1 0
TUP20031202 4 TWILIGH 5 6.00 2.89 A 0.00 1 0
TUP20031202 6 BRADFOR 5 1.80 2.08 E 0.00 1 0
TUP20031202 7 SHEET L 6 2.50 3.00 A 0.00 1 0
TUP20031202 8 NEWARK 7 3.00 3.58 0.00 1 0
TUP20031202 9 SNOWBAL 7 3.00 4.29 0.00 1 0
0.00 7.00 0.00

[JustMissed]

Joe-Which is correct-Monday or 20031202.

Thanks,

JustMissed

[formula_2002]

Justm...how time goes by.

It is indeed tuesday's picks

thanks

Joe M

PS its the 5 in the 4th that tends to bring home the bacon.

[JustMissed]

Joe, you hit 1 out of 7 races. The one you hit, the 6th paid $3.00;) .

All the other races your were off the board except for a place.

You need ask Santa Claus to bring you some new software for Christmas or a subscription to the Wizards picks.

See ya,

JustMissed

P.S. What exactly does "back fitted" mean.

[Larry Hamilton]

In a nutshell, backfitting means you takes todays average of anything and apply that average to races in the past. When this technique is used, trainers with actual 30% winners will appear to be 70% winners. It is the curse of those who store data and are not careful about how they calculate averages.

There are a million lies in this method and all SHOULD BE AVOIDED!

[cj]

Larry,

Love that quote you use at the bottom of your posts...I didn't realize the trainer of the last two BC Mile winners was such a philosopher! :D:D:D:D:D

[formula_2002]

Originally posted by JustMissed
Joe, you hit 1 out of 7 races. The one you hit, the 6th paid $3.00;) .

All the other races your were off the board except for a place.

You need ask Santa Claus to bring you some new software for Christmas or a subscription to the Wizards picks.

See ya,

JustMissed

P.S. What exactly does "back fitted" mean.

larry, its my own software.. it produces no worst than anyother and better then most.

How is the wizard been doing for the last few hundred win bets?

Backfitting is what most people do, I just do it to a greater number of races (this system used almost 9000 plays).

It's looking for patterns, you know, if a trainer wins two races out of town with a grey horse then next time out under those conditions, its a bet.
:confused:

[formula_2002]

Originally posted by JustMissed
Joe, you hit 1 out of 7 races. The one you hit, the 6th paid $3.00;) .




I think only two of yesterdays picks qualified for a play.
The final odds must exceed the min posted odds.

[Larry Hamilton]

what is clear is that we do not have the SAME understanding of what back fitting is.

As to the quote, it is one of my favorites too.

[JustMissed]

Originally posted by Larry Hamilton
In a nutshell, backfitting means you takes todays average of anything and apply that average to races in the past. When this technique is used, trainers with actual 30% winners will appear to be 70% winners. It is the curse of those who store data and are not careful about how they calculate averages.

There are a million lies in this method and all SHOULD BE AVOIDED!

Larry, I still don't get it.

I thought we(handicappers)took what happened in past races and projected the results forward in order to help predict future events.

If you have ever watch the Dr. Phil tv show, one of his favorite sayings is "Past behavior is the greatest indicator of future behavior". Now your telling me that our friend Formula2002 is using today's results and applying those results to the 'past'.

I don't get it.

JustMissed
:confused:

[formula_2002]

Originally posted by JustMissed
Larry, I still don't get it.

I thought we(handicappers)took what happened in past races and projected the results forward in order to help predict future events.

If you have ever watch the Dr. Phil tv show, one of his favorite sayings is "Past behavior is the greatest indicator of future behavior". Now your telling me that our friend Formula2002 is using today's results and applying those results to the 'past'.

I don't get it.

JustMissed
:confused:


JM
this is an example of the type of back fitting I look at.

as posted for WED picks.

"Worth a note regarding the 2nd at aqu.

when the picks min odds<1.and.the final odds are >then min odds, and the pick is the favorite, the results were as follows;

348 races, 50% winners, 6% profit.
Tough beast to bet against!"

This type of back fitting is ok because it is somewhat relating to the odds.

Good Luck

[Larry Hamilton]

I thought we(handicappers)took what happened in past races and projected the results forward in order to help predict future events.

This is exactly correct as presented.

It becomes BACKFITTING when you take that same average to update past averages. For instance, if you calcs show that Baily has an average win rate of 42% on the grass (just made that up). Now you a build a model that will be used to predict how well Baily did a year ago--you insert 42%. That is backfitting. The error is you have used the future to ADJUST the past.

Why wouldnt you expect the averages of all to improve? The longer you use this back fitting, the better they appear to get.

[Larry Hamilton]

As I reread this, I noticed that many would say, I woulld never adjust averages in the past. Let me tell you how you may be doing it without even realizing it.

Say Bailey 42% is calculated every day and this is the average as of today. In your db there is a point where your code says Insert Jockeys win rate on the grass. That .42 is inserted in every row all the way back to the beginning of the db--BACKFITTING!

Bailey is .42 only one time--today. Any other average identical is purely coincidental.

[formula_2002]

Just M

The first two races at AQU today(Wed) is almost as good as it gets for the system.

It made the 1st race fav 5-1 and it lost.

After the 7 scratched in the 2nd it made the winning favorite 1/2 and it won at 1-1

[formula_2002]

Originally posted by Larry Hamilton

It becomes BACKFITTING when you take that same average to update past averages. For instance, if you calcs show that Baily has an average win rate of 42% on the grass (just made that up). Now you a build a model that will be used to predict how well Baily did a year ago--you insert 42%. That is backfitting. The error is you have used the future to ADJUST the past.

Why wouldnt you expect the averages of all to improve? The longer you use this back fitting, the better they appear to get.

Larry, I'm using the term backfitting (sample) as follows;
When I say the picks are based on 8760 races, that means that results of the 8760 produced a dollar loss of 2%.

Then I proceed to use the "formulas" that produced the 2% loss on all races outside of that sample.
Currenly , I accumulated 191 race outside of the sample. Those 191 picks have produced a 10% loss.

Those live picks never go back into the origional sample so as to affect the formula.