Valiant Thor

Why not rate the courses by grade
You then only have 4 standards to fit to your algo,
...'KISS'... (KEEP-IT-SMALL & SIMPLE ) while building , you can always expand on it later.
Grade 1
Aintree
Ascot
Ayr
Cheltenham
Doncaster
Haydock
Kempton
Newbury
Newcastle
Newmarket
Sandown
Grade 2
Chepstow
Fontwell
Lingfield – Turf
Newton Abbot
Wetherby
Salisbury
Thirsk
Grade 3
Folkestone
Huntingdon
Leicester
Market Rasen
Nottingham
Plumpton
Stratford
Uttoxeter
Warwick
Wincanton
Windsor
Worcester
Grade 4
Bangor
Carlisle
Cartmel
Exeter
Fakenham
Hereford
Hexham
Kelso
Ludlow
Perth
Sedgefield
Southwell
Taunton
 

Why not rate the courses by grade
You then only have 4 standards to fit to your algo,
...'KISS'... (KEEP-IT-SMALL & SIMPLE ) while building , you can always expand on it later.
Grade 1
Aintree
Ascot
Ayr
Cheltenham
Doncaster
Haydock
Kempton
Newbury
Newcastle
Newmarket
Sandown
Grade 2
Chepstow
Fontwell
Lingfield – Turf
Newton Abbot
Wetherby
Salisbury
Thirsk
Grade 3
Folkestone
Huntingdon
Leicester
Market Rasen
Nottingham
Plumpton
Stratford
Uttoxeter
Warwick
Wincanton
Windsor
Worcester
Grade 4
Bangor
Carlisle
Cartmel
Exeter
Fakenham
Hereford
Hexham
Kelso
Ludlow
Perth
Sedgefield
Southwell
Taunton
 

I fail to see the relevance of the above.
 Taking in all JT favs as favs (hence 9 favs @ Haydock and only 7 races)


Fav to runner ratio still shows that Favs outperformed their chances of winning by +2.67%
While non favs under performed by -0.36%

I fail to see the relevance of the above.
 Taking in all JT favs as favs (hence 9 favs @ Haydock and only 7 races)


Fav to runner ratio still shows that Favs outperformed their chances of winning by +2.67%
While non favs under performed by -0.36%

I fail to see the relevance of the above.
 Taking in all JT favs as favs (hence 9 favs @ Haydock and only 7 races)


Fav to runner ratio still shows that Favs outperformed their chances of winning by +2.67%
While non favs under performed by -0.36%

I fail to see the relevance of the above.
 Taking in all JT favs as favs (hence 9 favs @ Haydock and only 7 races)


Fav to runner ratio still shows that Favs outperformed their chances of winning by +2.67%
While non favs under performed by -0.36%

I fail to see the relevance of the above.
 Taking in all JT favs as favs (hence 9 favs @ Haydock and only 7 races)


Fav to runner ratio still shows that Favs outperformed their chances of winning by +2.67%
While non favs under performed by -0.36%

Just use the racing post ratings (RPR) top rated for the 78 rated you wont be a million miles out ?

Michael Wray was from NZ, he used to upload his spreadsheets and ratings weekly.
Back in the early 2000's if anyone remembers the Football Association spat the dummy out and tried to sue anyone putting up the up and coming fixtures for the Leagues as they said they were their "intellectual property".
Michael got a letter from the Football Association marketing board threatening to sue for non payment of the use of their so called "intellectual property" so rather than risk getting sued he stopped posting them up.
Ps the Rating system was ok , (back top rated if so many pts clear blah,blah,blah) but nothing you couldnt knock up yourself in an hour or so using football data spreadsheets and a few stats.
 

In a word 'NO'

The funerals next Wednesday at 20/1

I use the same as everyone else does ,horse, jockey,trainer , runners , course  etc etc .
a bot just reads the output data for the race & puts the bet on without me having to do anything its the algorithm that sorts the data from the database that does the work.
What Im trying to do atm is build an algorithm to price up Hcp races and select a horse that fits the bill within a reasonable  tolerence +/-, but its hard due to the nature of Hcp races thats why Ive stayed away from them. but im bored so thought Id have another crack at it , keeps me busy.
The problem is in handicaps most are inconsistent lower grade garbage , where some are 'not really trying' to get the hcp mark down, ,most runners are usually ridden by lower class jocks, apprentices or Luke Morris ?
etc. They aim them for gambles rather than racing to form due to crap prize money,
Where as top grade races (nonHcp) are running for prestige not to make a few quid .
 

You are probably right although my analysis doesn't include details of the class of the race. I may look at including this in the future.
 

its all automated
The hardest part is getting all the data together and getting it mistake free after that the learning algorithm part is quite easy .
learn the basics on Kaggle for xg boost or gradient boosting (they give you the data bases to try and step you through the learning process ) test it out with there parameters when its working ok , put your data in instead of theirs an play around thats all I did .
You used to be able to get daily racecards in excel format  from HERE (I think you need to log in but still free I think)
then manipulate them how you want with a bit of  VBA then save as a .csv (uses less space than an excel file I think)
Thats half the job done

Hi Alex
Firstly  I do this for a hobby, its not my program (its just my data input)
Its a machine learning program in python that anyone can use, all I do is provide the data for it to read then it does its thing.
For the database input I use impact values for
Horse (win Ratio)
Horse (A/E)
Jockey (win Ratio)
Jockey (A/E)
Trainer (win Ratio)
Trainer (A/E)
Age
Course
Distance
Class
Runners

its all automated
The hardest part is getting all the data together and getting it mistake free after that the learning algorithm part is quite easy .
learn the basics on Kaggle for xg boost or gradient boosting (they give you the data bases to try and step you through the learning process ) test it out with there parameters when its working ok , put your data in instead of theirs an play around thats all I did .
You used to be able to get daily racecards in excel format  from HERE (I think you need to log in but still free I think)
then manipulate them how you want with a bit of  VBA then save as a .csv (uses less space than an excel file I think)
Thats half the job done

Hi Alex
Firstly  I do this for a hobby, its not my program (its just my data input)
Its a machine learning program in python that anyone can use, all I do is provide the data for it to read then it does its thing.
For the database input I use impact values for
Horse (win Ratio)
Horse (A/E)
Jockey (win Ratio)
Jockey (A/E)
Trainer (win Ratio)
Trainer (A/E)
Age
Course
Distance
Class
Runners

I wouldn't say its the "days" that are the relevant factor , more likely to be the class of racing(lower class jocks/trainers/horses) running on these days because they cant get in the decent races @ the weekend . Lower Class greater variance
That is what I stated @ the beginning of this thread .
Atm Im using xgboost as the algorithm on default settings,it works using decision trees on the past data you've trained it with, then it compares that with the present data (daily races) and makes its decision then puts it into pandas (Not the black and white cute ones ? , but an excel like data frame  ) this then sorts it into most likely winner by race and that is what I put up, the algorithm doesn't know or care whether its Sunday or Friday as long as you keep putting in new races it keeps giving out its assessment,  in fact its worst day was 21/05/22 (Sat -12) and its best day was 22/05/22 (Sun 7.5)
Some quite high price selections today which seem a bit strange but hey ho lets see how they go.
30/05/2022 08:56:10 Time Meeting Horse Ex_Odds B365 0 13.15 Lingfield Cu Chulainn  9-10 6.31 7.50 1 17.15 Redcar So Grateful  10-11 6.35 4.50 2 18.00 Ayr Red Force One  9-6 2.46 3.00 3 18.10 Windsor Winnetka  8-11 3.61 17.00 4 18.30 Ayr Cathayensis  10-2 24.50 2.25 5 18.40 Windsor Just That Lord  8-9 4.59 10.00 6 19.10 Windsor Prejudice  8-11 4.46 11.00

I wouldn't say its the "days" that are the relevant factor , more likely to be the class of racing(lower class jocks/trainers/horses) running on these days because they cant get in the decent races @ the weekend . Lower Class greater variance
That is what I stated @ the beginning of this thread .
Atm Im using xgboost as the algorithm on default settings,it works using decision trees on the past data you've trained it with, then it compares that with the present data (daily races) and makes its decision then puts it into pandas (Not the black and white cute ones ? , but an excel like data frame  ) this then sorts it into most likely winner by race and that is what I put up, the algorithm doesn't know or care whether its Sunday or Friday as long as you keep putting in new races it keeps giving out its assessment,  in fact its worst day was 21/05/22 (Sat -12) and its best day was 22/05/22 (Sun 7.5)
Some quite high price selections today which seem a bit strange but hey ho lets see how they go.
30/05/2022 08:56:10 Time Meeting Horse Ex_Odds B365 0 13.15 Lingfield Cu Chulainn  9-10 6.31 7.50 1 17.15 Redcar So Grateful  10-11 6.35 4.50 2 18.00 Ayr Red Force One  9-6 2.46 3.00 3 18.10 Windsor Winnetka  8-11 3.61 17.00 4 18.30 Ayr Cathayensis  10-2 24.50 2.25 5 18.40 Windsor Just That Lord  8-9 4.59 10.00 6 19.10 Windsor Prejudice  8-11 4.46 11.00

Just looked at the 6.30 Ayr where it has Cathayensis  as Ex_Odds of 24.50
Ran that race again and same returns
Seems the upload for that race must be corrupt somewhere as it has all the runners @ Ex_Odds of 24.50 and selected Cathayensis simply because its the 1st horse (no1)
I'll leave it as previously posted as I cant see 1 runner making much difference win or lose.

I wouldn't say its the "days" that are the relevant factor , more likely to be the class of racing(lower class jocks/trainers/horses) running on these days because they cant get in the decent races @ the weekend . Lower Class greater variance
That is what I stated @ the beginning of this thread .
Atm Im using xgboost as the algorithm on default settings,it works using decision trees on the past data you've trained it with, then it compares that with the present data (daily races) and makes its decision then puts it into pandas (Not the black and white cute ones ? , but an excel like data frame  ) this then sorts it into most likely winner by race and that is what I put up, the algorithm doesn't know or care whether its Sunday or Friday as long as you keep putting in new races it keeps giving out its assessment,  in fact its worst day was 21/05/22 (Sat -12) and its best day was 22/05/22 (Sun 7.5)
Some quite high price selections today which seem a bit strange but hey ho lets see how they go.
30/05/2022 08:56:10 Time Meeting Horse Ex_Odds B365 0 13.15 Lingfield Cu Chulainn  9-10 6.31 7.50 1 17.15 Redcar So Grateful  10-11 6.35 4.50 2 18.00 Ayr Red Force One  9-6 2.46 3.00 3 18.10 Windsor Winnetka  8-11 3.61 17.00 4 18.30 Ayr Cathayensis  10-2 24.50 2.25 5 18.40 Windsor Just That Lord  8-9 4.59 10.00 6 19.10 Windsor Prejudice  8-11 4.46 11.00

A nice a 6pts profit on the day to ep and a few of 2nd placed ones again ?
28/05/2022 20:18:18 Time Meeting Horse Ex_Odds B365 Result sp ep_Profit cum_ep sp_Profit cum_sp 0 14.20 Haydock Forza Orta  8-4 5.50 4.50 l 3.75 -1 -1 -1 -1 1 14.40 Beverley Percy's Lad  10-3 3.22 4.33 l 3.00 -1 -2 -1 -2 2 15.39 Catterick Cragside  9-9 2.52 5.50 l 4.33 -1 -3 -1 -3 3 15.50 Beverley Ha Long Bay  9-1 4.81 9.00 l 9.00 -1 -4 -1 -4 4 16.05 Haydock Nationwide  9-5 4.00 5.50 l 6.50 -1 -5 -1 -5 5 16.10 Chelmsford Dubai Station  9-4 1.67 13.00 w 12.00 12 7 11 6 6 16.14 Catterick Yasmin From York  8-11 1.91 3.00 w 3.00 2 9 2 8 7 16.25 Beverley Miss Mulligan  9-8 4.86 7.00 l 4.00 -1 8 -1 7 8 17.20 Chelmsford Dors Toyboy  9-5 6.93 8.00 l 9.00 -1 7 -1 6 9 17.24 Catterick The Armed Man  9-8 4.44 7.00 l 11.00 -1 6 -1 5 Update: Today takes us back into the black with an overall profit of 4.95pts date selections winners return cum_ret yield cum_yield st_rate cum_st_rate 0 2022-05-18 9 3 3.50 3.50 0.39 0.39 0.33 0.33 1 2022-05-19 14 3 -1.00 2.50 -0.07 0.11 0.21 0.26 2 2022-05-20 11 2 0.75 3.25 0.07 0.10 0.18 0.24 3 2022-05-21 12 0 -12.00 -8.75 -1.00 -0.19 0.00 0.17 4 2022-05-22 5 3 7.50 -1.25 1.50 -0.02 0.60 0.22 5 2022-05-23 6 1 -2.00 -3.25 -0.33 -0.06 0.17 0.21 6 2022-05-24 8 3 6.50 3.25 0.81 0.05 0.38 0.23 7 2022-05-25 9 1 -6.75 -3.50 -0.75 -0.05 0.11 0.22 8 2022-05-26 11 2 -1.25 -4.75 -0.11 -0.06 0.18 0.21 9 2022-05-27 6 3 3.70 -1.05 0.62 -0.01 0.50 0.23 10 2022-05-28 10 2 6.00 4.95 0.60 0.05 0.20 0.23

Flat season and we have 4 jumps meetings today, ,just about sums racing up ?
No selections today

Flat season and we have 4 jumps meetings today, ,just about sums racing up ?
No selections today

A nice a 6pts profit on the day to ep and a few of 2nd placed ones again ?
28/05/2022 20:18:18 Time Meeting Horse Ex_Odds B365 Result sp ep_Profit cum_ep sp_Profit cum_sp 0 14.20 Haydock Forza Orta  8-4 5.50 4.50 l 3.75 -1 -1 -1 -1 1 14.40 Beverley Percy's Lad  10-3 3.22 4.33 l 3.00 -1 -2 -1 -2 2 15.39 Catterick Cragside  9-9 2.52 5.50 l 4.33 -1 -3 -1 -3 3 15.50 Beverley Ha Long Bay  9-1 4.81 9.00 l 9.00 -1 -4 -1 -4 4 16.05 Haydock Nationwide  9-5 4.00 5.50 l 6.50 -1 -5 -1 -5 5 16.10 Chelmsford Dubai Station  9-4 1.67 13.00 w 12.00 12 7 11 6 6 16.14 Catterick Yasmin From York  8-11 1.91 3.00 w 3.00 2 9 2 8 7 16.25 Beverley Miss Mulligan  9-8 4.86 7.00 l 4.00 -1 8 -1 7 8 17.20 Chelmsford Dors Toyboy  9-5 6.93 8.00 l 9.00 -1 7 -1 6 9 17.24 Catterick The Armed Man  9-8 4.44 7.00 l 11.00 -1 6 -1 5 Update: Today takes us back into the black with an overall profit of 4.95pts date selections winners return cum_ret yield cum_yield st_rate cum_st_rate 0 2022-05-18 9 3 3.50 3.50 0.39 0.39 0.33 0.33 1 2022-05-19 14 3 -1.00 2.50 -0.07 0.11 0.21 0.26 2 2022-05-20 11 2 0.75 3.25 0.07 0.10 0.18 0.24 3 2022-05-21 12 0 -12.00 -8.75 -1.00 -0.19 0.00 0.17 4 2022-05-22 5 3 7.50 -1.25 1.50 -0.02 0.60 0.22 5 2022-05-23 6 1 -2.00 -3.25 -0.33 -0.06 0.17 0.21 6 2022-05-24 8 3 6.50 3.25 0.81 0.05 0.38 0.23 7 2022-05-25 9 1 -6.75 -3.50 -0.75 -0.05 0.11 0.22 8 2022-05-26 11 2 -1.25 -4.75 -0.11 -0.06 0.18 0.21 9 2022-05-27 6 3 3.70 -1.05 0.62 -0.01 0.50 0.23 10 2022-05-28 10 2 6.00 4.95 0.60 0.05 0.20 0.23