JCapper Message Board
Sql rider trainer win%
|| Sql rider trainer win%
|I have created a "connections" negative expectation UDM where one of the conditions I have is (valf34+valf35) <= .15, where f34 is rider win% and f35 is trainer win%. In theory this somewhat "weeds out" low performing connections. Then I noticed that on breeders cup day I was getting a connections eliminate flag where the trainer was Graham Motion and the Jockey was John Velasquez. Certainly not bad connections! I then figured out that the win percentages must pertain to the current track. Since Motion and Velasquez rarely compete at Santa Anita, they could in theory have very low percentages at that track. I am trying to find a work around where legit connections are not flagged by the UDM (e.g. Bob Baffert and Martin Garcia are the jockey and trainer at Finger Lakes) my UDM would flag them as "bad" even though that is not the case. I'm not sure I can come up with a solution to this conundrum. Any ideas?|
~Edited by: SILVER01HDW on: 11/25/2014 at: 3:11:01 PM~
|First, the long answer...|
I'll preface this post by saying - in my own way of doing things - about 98% of the horses I bet on are bet in the first place because they are flagged by UDMs based primarily on speed-pace-form.
I use rider and trainer info as 'layering' info only - and almost never as the sole basis for betting a horse.
Within this context 'layering' info means ANY type of supplemental info that can provide a deeper level of understanding about the horses I bet.
For example, a horse flagged by one of my good speed-pace-form UDMs means one thing when it is not flagged by any 'layering' UDMs at all - and can mean something else entirely when a horse flagged by that same speed-pace-form UDM is also flagged by multiple 'layering' UDMs.
Other players may use rider or trainer info differently but all I can post about are my own experiences.
I approach this by running sql expressions through the Data Window with the objective being to create speed-pace-form UDMs to get me basic contenders first.
I'll then run speed-pace-form (basic contender) expressions through the Data Window with the data broken out by layering info such as BY TRAINER.
From there I can use the Name Selection Tool to create layering UDMs for under the radar trainers.
On race day, when I see a horse flagged by a speed-pace-form UDM that is also flagged by a 'layering' UDM designed to tell me that same horse just so happens to be trained by an under the radar trainer - my level of understanding about that horse deepens.
I can also run individual trainer data for 'contenders' broken out BY RIDER to get an idea of hierarchy within a barn to see who gets 'first call' for that barn's live mounts - and again use the Name Selection Tool to create layering UDMs based on first call riders for individual trainers. -- With the result being that the entire UDM profile for a horse flagged by that initial speed-pace-form UDM now gives me a more in depth understanding of that horse at a glance.
The short answer to the question posed at the top of this thread is a suggestion to create layering UDMs based on rider and trainer names. Doing so means that you are limited only by the data in your StarterHistory table and your own imagination - not the way the stats are compiled within the program and/or data file.
For example, if we are talking about the Breeders Cup, use names from either the Breeders Cup history only - or another idea might be to use names for say the past few years of Graded Stakes races only.
Hint: search the board for the phrase Name Selection Tool with me as author for links to a video as well as links to various threads where I've posted examples of using the Name Selection Tool to generate lists of rider and trainer names that can be pasted into sql udms in conjuction with an INSTR command.
Hint: Link to an older thread that contains some discussion about name based connections sql UDMs:
Hint: Link to an older thread where I posted basic operating instructions for using the Name Selection Tool to paste a list of names into a sql UDM:
|Thanks for the reply Jeff, what I ended up doing is creating multiple OR statements so that I would not flag riders/trainers that would not be considered "poor" performing. As a side note somewhat; in my personal opinion, the user has much more control and can design much better udms in SQL mode. I have noticed great improvements in my game since switching to SQL mode.. the ability to peform math expressions, the speed of the queries, etc.. I would highly recommend SQL mode to anyone if they are not using it...just my .02 cents|