Racing data trial - free and its over 2 weeks Flat,NH, All weather and international

hrstats

Who would like to be part of a trial for horse racing data machine learning exercise - the output would be more data accurate fields and provide the punters with what they really want in terms of meaningful information

The Tetrarch

I don't understand your post. Please give an example, or a detailed explanation. Are you selling anything?

del roy

Do you mean something like these...

http://www.gaultstats.com/

del roy

or this

Who?

Calle Malva @ 4/1 BOG

...in a 7-runner, Class 4, Mares Novices Handicap Hurdle for 4yo+ over 2m6½f on heavy ground worth £4,289 to the winner...

Why?...

This 5 yr old mare has won one of her four starts to date, another Class 4 hurdle on heavy ground at relatively nearby Hexham, so conditions won't be too unfamiliar today. That win is the only win recorded by any of the runners in this contest, as her six rivals have only mustered a combined 4 places from 21 runs to date.

Regular jockey Blair Campbell is the saddle again today, once more hoping to use his 3lb claim effectively enough to land another Kelso winner for trainer Lucinda Russell, who is 30 from 117 (25.6% SR) for 60.5pts (+51.7% ROI) with horses sent off in the Evens to 8/1 price bracket in handicaps at this venue since the start of 2015. These include of relevance today...

28/103 (27.2%) for 67.1pts (+65.1%) at Classes 3 & 4
18/74 (24.3%) for 46.7pts (+63.2%) with horses unplaced last time out
18/74 (24.3%) for 33.6pts (+45.3%) at Class 4
18/56 (32.1%) for 54.7pts (+97.6%) in fields of 6-8 runners
14/43 (32.6%) for 29.9pts (+69.5%) during March to May
13/45 (28.9%) for 46.1pts (+102.4%) with a jockey claiming 3-6lbs
12/36 (33.3%) for 41.3pts (+114.6%) over hurdles
9/31 (29%) for 9.6pts (+31%) on heavy ground
8/18 (44.4%) for 19.9pts (+110.8%) in March
7/20 (35%) for 17.7pts (+88.5%) with Blair Campbell in the saddle
and 7/16 (43.75%) for 31pts (+193.9%) with female runners...

...whilst risking diluting the data too far...those unplaced LTO and now running at Class 3/4 in fields of 4-7 in March are 5 from 9 (55.6% SR) for 15.66pts (+174% ROI) profit...

...giving us...a 1pt win bet on Calle Malva @ 4/1 BOG as was available from Bet365, BetVictor & Hills at 8.00am Monday with plenty of 7/2 BOG elsewhere, but as always please check your BOG status. To see what your preferred bookie is quoting...

...click here for the betting on the 3.40 Kelso

hrstats

The Tetrarch wrote: »

I don't understand your post. Please give an example, or a detailed explanation. Are you selling anything?

We are looking at a new set of variables - at the moment we look at 434 different variables , as an example we review weather , ground conditions, predicted ground conditions, distance travelled , incline , decline , course maps , training maps and then the outcome is scored and horses are graded on that score , we want to get peoples views and opinions , of what data they would like but is not available etc , it’s all free not selling anything, we would like to build a temporary models on as many of the recommendations as possible and then trial the outcome against future races

hrstats

del roy wrote: »

or this

Who?

Calle Malva @ 4/1 BOG

...in a 7-runner, Class 4, Mares Novices Handicap Hurdle for 4yo+ over 2m6½f on heavy ground worth £4,289 to the winner...

Why?...

This 5 yr old mare has won one of her four starts to date, another Class 4 hurdle on heavy ground at relatively nearby Hexham, so conditions won't be too unfamiliar today. That win is the only win recorded by any of the runners in this contest, as her six rivals have only mustered a combined 4 places from 21 runs to date.

Regular jockey Blair Campbell is the saddle again today, once more hoping to use his 3lb claim effectively enough to land another Kelso winner for trainer Lucinda Russell, who is 30 from 117 (25.6% SR) for 60.5pts (+51.7% ROI) with horses sent off in the Evens to 8/1 price bracket in handicaps at this venue since the start of 2015. These include of relevance today...

28/103 (27.2%) for 67.1pts (+65.1%) at Classes 3 & 4
18/74 (24.3%) for 46.7pts (+63.2%) with horses unplaced last time out
18/74 (24.3%) for 33.6pts (+45.3%) at Class 4
18/56 (32.1%) for 54.7pts (+97.6%) in fields of 6-8 runners
14/43 (32.6%) for 29.9pts (+69.5%) during March to May
13/45 (28.9%) for 46.1pts (+102.4%) with a jockey claiming 3-6lbs
12/36 (33.3%) for 41.3pts (+114.6%) over hurdles
9/31 (29%) for 9.6pts (+31%) on heavy ground
8/18 (44.4%) for 19.9pts (+110.8%) in March
7/20 (35%) for 17.7pts (+88.5%) with Blair Campbell in the saddle
and 7/16 (43.75%) for 31pts (+193.9%) with female runners...

...whilst risking diluting the data too far...those unplaced LTO and now running at Class 3/4 in fields of 4-7 in March are 5 from 9 (55.6% SR) for 15.66pts (+174% ROI) profit...

...giving us...a 1pt win bet on Calle Malva @ 4/1 BOG as was available from Bet365, BetVictor & Hills at 8.00am Monday with plenty of 7/2 BOG elsewhere, but as always please check your BOG status. To see what your preferred bookie is quoting...

...click here for the betting on the 3.40 Kelso

We’ve seen these before, some of the info is useful but the model will be different

PillarToPost

all **** , every race should be studied on its own merits

del roy

PillarToPost wrote: »

all **** , every race should be studied on its own merits

all stats are there to be turned around, just like form.

hrstats

PillarToPost wrote: »

all **** , every race should be studied on its own merits

But it’s finding those individual merits and calculating an outcome by using historical databases etc would mean more and more accurate information and more predictive outcome , there is still a huge factor that it’s a 1/2 tonne animal that has a mind of its own

The Tetrarch

hrstats wrote: »

We are looking at a new set of variables - at the moment we look at 434 different variables , as an example we review weather , ground conditions, predicted ground conditions, distance travelled , incline , decline , course maps , training maps and then the outcome is scored and horses are graded on that score , we want to get peoples views and opinions , of what data they would like but is not available etc , it’s all free not selling anything, we would like to build a temporary models on as many of the recommendations as possible and then trial the outcome against future races

I have a lot of opinions.
Is it flat or jumps or both?
My feeling is that 434 different variables is unlikely. Probably many are related or part of another variable, or derived from a variable.
If you use data as presented by the industry e.g. ground conditions, then you will be using wrong data.
And win rate, earnings, result in previous race - you need to be careful.
Do we get to see the 434 variables?

weather: the only factor to bother about here is wind strength and direction, and the draw. If the wind is from the right, and your pick is drawn on the left, that is a positive.
ground conditions: I can give you info from one of Nick Mordin's books where he proved going descriptions are worthless. Even worse, they are misleading.
The other thing you probably are not considering is the course soil. This is a major factor.
distance travelled: Is this the distance from the training yard to the racecourse?
incline , decline: very important. horses run slower down hill (although one poster told me I was daft and gave one example, the 5f Epsom course as proof).
course maps, training maps: this could be interesting. It can be tricky where a racecourse has two tracks and you do not know which was used.

hrstats

hrstats wrote: »

Who would like to be part of a trial for horse racing data machine learning exercise - the output would be more data accurate fields and provide the punters with what they really want in terms of meaningful information

This is a sample for today’s racing - the model is built to cover two types of punter , fav and value , green for fav and yellow for the more valued recommend horse

The Tetrarch

hrstats wrote: »

This is a sample for today’s racing - the model is built to cover two types of punter , fav and value , green for fav and yellow for the more valued recommend horse

That needs a bit of work.
You have Kelso 02:10 Temple Guide in green (favourite) at 50/1

A problem to me is I have no idea why you picked these.

hrstats

The Tetrarch wrote: »

That needs a bit of work.
You have Kelso 02:10 Temple Guide in green (favourite) at 50/1

A problem to me is I have no idea why you picked these.

So you look at one negative and ignore all the positives 😂, there is still some bugs in the tool but that horse falls down the list once the bug was removed

Daraa

Rest of the days data appears to be here if anyone wants to judge based on future performance and not just past.

https://www.hrstats.com/membersarea100


Edit: If you load up the website. https://www.hrstats.com and just click Mondays Racing it will load.

marieholmfan

hrstats wrote: »

Who would like to be part of a trial for horse racing data machine learning exercise - the output would be more data accurate fields and provide the punters with what they really want in terms of meaningful information

Keenly interested hrstats

The Tetrarch

hrstats wrote: »

So you look at one negative and ignore all the positives ��, there is still some bugs in the tool but that horse falls down the list once the bug was removed

You say "äll the positives".
All I see is a list of horses, with no indication why they are the selections.
Has this been trialled, and are the results positive?
By trialled I do not mean based on past races. Was it trialled on races where you did not know the result? That is how punting works. You bet on an unknown.