Find Selections Usi...
 
Notifications
Clear all

Find Selections Using Impact Values  

Page 3 / 4
  RSS

AndrewP
(@andrewp)
Reputable Member
Joined: 1 year ago
Posts: 429
28/02/2020 3:07 pm  

Thanks Michael. So are these figures about right ( I have allowed an extra expected winner in 2019 for the two missing odds).

  2019 2020
RUNNERS 684 124
WINNERS 205 38
SR 30.0% 30.6%
EXP WIN 168.00 30
A/E 1.23 1.267
POT Level 14% 9.44%
POT Prop 15% 20.07%
Chi2 0.10% 9.34%

ReplyQuote
Michael Wilding
(@mikeywilding)
Member Admin
Joined: 11 years ago
Posts: 1649
28/02/2020 3:26 pm  

@andrewp without cross-checking, they look about right.

 

The Race Advisor now has an Instagram channel. You can follow us on Instagram here.


ReplyQuote
MICHAEL CLARKE
(@michael-clarke)
Prominent Member
Joined: 5 years ago
Posts: 854
21/05/2020 9:52 pm  

@mikeywilding

Michael, have all the issues with historical data now been resolved? 

I recall the main issues included a lot of missing historical data and ratings on the day being different when they were exported at a later date. Also there were a lot of missing BSPs.

I am contemplating having another bash when racing resumes but I want to avoid the problems I encountered last time.


ReplyQuote
John Waters
(@johnwaters1965)
Honorable Member
Joined: 11 months ago
Posts: 662
22/05/2020 9:26 am  

So let me get this straight - we are looking at All Weather 6 furlongs and below

 

PFP Rank 1st-3rd

5278 CFR Rank 1st

and then sorting by BSP?

 

Just wondering if it's worthwhile also looking at breaking the results down over Course - so we would have results for the different surfaces and draw biases (if there is any) - I know I always look at Southwell differently to other AW Courses so maybe a difference between fibresand/polytrack/tapeta could be spotted?


ReplyQuote
Barry
(@budgie-1)
Active Member
Joined: 2 years ago
Posts: 10
22/05/2020 10:03 am  

i think i may try this thanks john


ReplyQuote
Michael Wilding
(@mikeywilding)
Member Admin
Joined: 11 years ago
Posts: 1649
22/05/2020 10:13 am  

@michael-clarke we now have multiple additional data quality control points. Horse racing data is very noisy and full of holes, but we are now catching as many of those holes as possible. We are currently in the final phase of our quality control upgrades, which is a restructure and optimisation of our database. This final piece will correct any outstanding known issues. We are trying to get this completed before racing restarts, but it may take a week or two longer.

Betfair SP is out of our hands, we currently use the results files provided by Betfair and sometimes these contain errors such as spelling mistakes, which means we can't match them to runners in. the database. I would need to check, but I believe we have well over 90% of UK and IRE Betfair SP's. Where we have been unable to get Betfair SP's the SP is of course available. 

The Race Advisor now has an Instagram channel. You can follow us on Instagram here.


ReplyQuote
Michael Wilding
(@mikeywilding)
Member Admin
Joined: 11 years ago
Posts: 1649
22/05/2020 10:36 am  
Posted by: @johnwaters1965

So let me get this straight - we are looking at All Weather 6 furlongs and below

 

PFP Rank 1st-3rd

5278 CFR Rank 1st

and then sorting by BSP?

 

Just wondering if it's worthwhile also looking at breaking the results down over Course - so we would have results for the different surfaces and draw biases (if there is any) - I know I always look at Southwell differently to other AW Courses so maybe a difference between fibresand/polytrack/tapeta could be spotted?

Yup, that's the building blocks to start with 🙂 I was wanting to show a method that you can use to start building methods from the ground up. 

It's always worthwhile looking into breaking results down, as long as the breakdown doesn't reduce the quantity of selections too low. There should also be a solid reason for seeing results, if you can't see a logical reason why results may be better or worse in a segment, based on the ratings you're using, then you're best of not using them.

The Race Advisor now has an Instagram channel. You can follow us on Instagram here.


ReplyQuote
John Waters
(@johnwaters1965)
Honorable Member
Joined: 11 months ago
Posts: 662
22/05/2020 11:56 am  

@mikeywilding

I've spent a little time looking at the breeding of horses on the All Weather courses in UK & Ireland and Southwell stands out as an exception in some ways.

Looking at the country of the Sire for example gives an unusual slant for Southwell:

Country of Sire

Runners

Winners

Strike Rate

Expected Wins

A/E

France

306

45

14.71

36.10

1.25

Australia

576

78

13.54

72.29

1.08

Ireland

3970

481

12.12

473.69

1.02

GB

4604

504

10.95

523.36

0.96

USA

2351

250

10.63

269.86

0.93

Rest of World

221

44

19.91

32.34

1.36

I've only highlighted Nations with Sires who've had 20+ winners since 2015 - as you can see USA is the lowest ranked of these nations in terms of A/E with France the highest ranked.

 

Looking at the other A/W Courses:

Country of Sire

Runners

Winners

Strike Rate

Expected Wins

A/E

France

3811 

361 

9.47 

381.33 

0.95 

Australia

6701 

788 

11.76 

764.07 

1.03 

Ireland

 48070

5091 

10.59 

33.79 

0.99 

GB

45498 

4603 

10.12 

4610.15 

1.00 

USA

21390 

2331 

10.90 

2273.86 

1.03 

Rest of World

2096 

218 

10.40 

226.75 

0.96 

This time the table has been turned on it's head with USA the highest A/E and France the lowest.

Obviously some of the samples are too small to be statistically significant but I do think there is something there.

Incidentally I'm looking at following Horses for Southwell that have been Sired by Lethal Force (Ireland) or Street Cry (Ireland) - I'll try and remember to post the full list of Sires for AW courses and then you can follow and see what, if any, progress I make.


ReplyQuote
MICHAEL CLARKE
(@michael-clarke)
Prominent Member
Joined: 5 years ago
Posts: 854
22/05/2020 2:30 pm  

@mikeywilding

I think often the BSP issue is where there is an apostrophe in the name, e.g. FAVORITO BUCK'S in the RA data is shown as FAVORITO BUCKS in the BSP file. It is relatively easy to match up these anomalies.

I would suggest that as part of the data cleansing exercise these missing BSP prices are manually adjusted. It is a bit time consuming but BSP is such an important piece of data that it is worth the effort. 

Out of interest how do you deal with the missing data in the various analysis of data that you have performed? In this current analysis based on the 10% of missing data you would have over 1,000 horses with no BSP.


ReplyQuote
AndrewP
(@andrewp)
Reputable Member
Joined: 1 year ago
Posts: 429
22/05/2020 5:14 pm  

I maintain my own database of pre-race ratings and results culled from several sources. I've learned the hard way to resolve all sources to one common unique identifier, a combination of;

Date - As Microsoft date

Scheduled Off Time - As Microsoft time

Course - With Newmarket split into July and Rowley

Horse Name - Upper case with all punctuation and leading/trailing spaces removed

Country Suffix - e.g. (IRE), (GB), (FR)

Twice I've come across horses with the same name running at the same course on the same day. Not yet in the same race as well but it's only a question of time before that happens. If it can possibly happen, it will one day (or has already either here or in some other quantum or universe). :0)


ReplyQuote
Michael Wilding
(@mikeywilding)
Member Admin
Joined: 11 years ago
Posts: 1649
22/05/2020 8:39 pm  

@michael-clarke I am pretty sure we were looking for apostrophe's, but would need to double check. We now have a better way of doing it on current and future races. In the examples I use, if there is no BSP then I use the SP.

The Race Advisor now has an Instagram channel. You can follow us on Instagram here.


ReplyQuote
Michael Wilding
(@mikeywilding)
Member Admin
Joined: 11 years ago
Posts: 1649
22/05/2020 8:40 pm  

@andrewp if it can happen it will in horse racing data! We've seen horses running at 150mph before 😀

The Race Advisor now has an Instagram channel. You can follow us on Instagram here.


ReplyQuote
Michael Wilding
(@mikeywilding)
Member Admin
Joined: 11 years ago
Posts: 1649
22/05/2020 8:41 pm  

@johnwaters1965 very interesting, definitely worth getting more data on.

The Race Advisor now has an Instagram channel. You can follow us on Instagram here.


ReplyQuote
Barry
(@budgie-1)
Active Member
Joined: 2 years ago
Posts: 10
23/05/2020 8:15 pm  

on the bsp narewe looking for a high or low number here


ReplyQuote
Barry
(@budgie-1)
Active Member
Joined: 2 years ago
Posts: 10
23/05/2020 8:16 pm  

sorry i meant to type on the bsp are we looking for a high or low number here please john


ReplyQuote
Page 3 / 4
Share:
Back to top button
Close
Close