|
JCapper Message Board JCapper 101
|
| Home |
Register
Log In |
| By | PScore | |
| 59tpepejr 5/16/2019 2:32:24 PM | Could someone please tell me the difference in the numbers. Example 5 horse Pscore is 64 7 horse is 110 Does the lower score for the 5 horse mean the public will be betting this horse or does the 7's higher score mean the public will be betting this horse? Thanks
| |
| jeff 5/17/2019 1:36:19 PM | Here's what I did prior to posting the Data Window output below: I. • I used the UDM Wizard to create a new playlist file UDM named pl_AP1 with the following factor constraints: CompoundAP minrank=1 maxrank=1 • I brought up the Data Window and used the folder nav icon to point the Data Window at my Q3 2018 folder. II. • From the Data Window MENU, I selected Exports, Quick Index File - Create New File, and used the resulting dialog box to define a new quick index file named pl_AP1.txt • I clicked the UDM button in the Data Window to execute the pl_AP1 UDM against my Q3 2018 folder. • After the query results had been returned I now had a quick index file named pl_AP1.txt (populated with CompoundAP minrank=1 maxrank=1 starters only) sitting on my Q3 2018 folder. • I then cut and pasted the pl_AP1.txt file from my Q3 2018 folder to my Q4 2018 folder. III. • From the Data Window MENU, I selected Exports, Quick Index File - Append to Existing File, and used the resulting dialog box to select the pl_AP1.txt file on my Q4 2018 folder. • I clicked the UDM button in the Data Window to execute the pl_AP1 UDM against my Q4 2018 folder. • After the query results had been returned I now had a quick index file named pl_AP1.txt (populated with CompoundAP minrank=1 maxrank=1 starters only for both Q3 2018 and Q4 2018) sitting on my Q4 2018 folder. • I then cut and pasted the pl_AP1.txt file from my Q4 2018 folder to my Q1 2019 folder. IV. • From the Data Window MENU, I selected Exports, Quick Index File - Append to Existing File, and used the resulting dialog box to select the pl_AP1.txt file on my Q1 2019 folder. • I clicked the UDM button in the Data Window to execute the pl_AP1 UDM against my Q1 2019 folder. • After the query results had been returned I now had a quick index file named pl_AP1.txt sitting on my Q1 2019 folder. --Note: At this point, the file is now populated with CompoundAP minrank=1 maxrank=1 starters only data for three consecutive quarters: Q3 2018, Q4 2018, and Q1 2019. --Note: I created the Quick Index File because doing that allows me to span multiple folders in the Data Window while working in playist file mode. (Plus, working with quick index files is faster than working with the much larger pl_profile.txt file.) V. • I then used the UDM Wizard to create a new playlist file UDM named pl_AP-test with the following factor constraints: UDM Definition: pl_AP-test VI. • From there, I executed the above UDM against the quick index file named pl_AP1.txt sitting on my Q1 2019 folder. Here's a cut and paste of the Data Window results with the data broken out by simple PScore: query start: 5/17/2019 10:52:41 AM By: PScore VII. After performing the above preliminary steps, I am now in a position to answer your question. You asked: "Does the lower score for the 5 horse mean the public will be betting this horse or does the 7's higher score mean the public will be betting this horse?" My reply: Neither. If you look at PScore using an ALL button query, PScore probably isn't something you're likely to find (remotely) useful. But if you look at a dataset of runners with a few hidden positives in their past performance records, when you break the data out by PScore, sometimes you'll see something useful. PScore is designed to be an indicator of how likely it is that the public will mis-bet or mis-price (or be fooled by) a horse given the hidden positives in its past performance data. It's not an indicator of how much money or how little money a given runner will attract. There's a subtle difference between the two. Imo, the stronger the hidden positives contained in the past performance records of the runners of a given dataset, the more likely it is you may find PScore to be of some use. (There are exceptions, but usually - the higher the PScore, the more likely it is a horse's past performance data contains attributes that can fool the pubic.) Notes: I needed to create something like the above UDM before trying to answer to your question. Otherwise, the PScore data breakout wasn't going to show the effect I was shooting for. About that UDM: The above UDM is based on a development sample spanning 10-01-2018 through 03-31-2019. I haven't forward tested it on fresh data nor am I using it for live play. I purposely chose the factors that I did because all of them are available in JCapper Silver and in playlist file mode. Hope I managed to type most of that out in a way that makes sense. -jp . ~Edited by: jeff on: 5/17/2019 at: 1:36:19 PM~
| |
| 59tpepejr 5/20/2019 11:00:58 AM | Yes, you did. Thanks! Tom
| |
| jeff 5/22/2019 10:38:15 AM | About four lines up from the bottom, there's something in my post (above) that needs editing -- --Change this: About that UDM: The above UDM is based on a development sample spanning 10-01-2018 through 03-31-2019. --To read as follows: About that UDM: The above UDM is based on a development sample spanning 07-01-2018 through 03-31-2019. -jp .
| |
| NYMike 3/20/2024 10:17:40 PM | It seems the higher the PScore is typically better but there seems to be something about a PScore of 45 that is different than just below or just above. Any reason for that?
| |
| jeff 3/22/2024 2:54:38 PM | I wrote the algorithm using 45 as the default. The algorithm evaluates MLine and an implied fair odds strike price as inputs and generates PScore_HDW as the output. If I were to graph all of the horses fed into the algorithm on a bell curve - the ones in middle area of the curve are the ones that get PScores. The ones to the far left and far right are the ones that retain the default 45. Of course I should probably redo the algorithm altogether. I created it using data from 2004-2005. At the time, favorites won at a lower rate with lower flat win pool roi than present day, and horses scoring out at about 95 or higher were near break even in the win pool. In today's game that's no longer the case (even though you can still find the occasional exception.) -jp .
|
| Reply |
|
Copyright © 2018 JCapper Software back to the JCapper Message Board www.JCapper.com |