Horse Predictor

[bobs71]

tell u what
Tell the average to go somewhere
Was ready to swear there
Consider the fastest

StellaBot - trying to decipher your incomprehensible posts is much harder than trying to put a figure on a handicap winner at Chelmsford on a wet and blustery Thursday night - if you had bothered to read my post , i did say "averages don't work" - standards are created by comparing a dataset of past times against expected to actual ability , and the use of statistical techniques involving standard deviation and variable percentiles - It is easier to run faster at 5f than it is to run faster at 16f , so that basic fact of equine bio-mechanics has to be factored in , along with normalisation in relation to the factors that effect times - This process of benchmarking has to be done in the first place to "consider the fastest" otherwise you are working with smoke and mirrors. There is also an ongoing calibration process ,simply because new data is added every single racing day. One aspect that nobody has mentioned is the amount of "false" published or official times that get passed through as if they are accurate - the video analysis (which i know Timeform do as well) is a necessary - to give you an example , have a look at a horse called Promote - who won recently at Newcastle in an "official" time of 71.02 - video analysis shows that was out by ~2.00 full seconds which would equate to around 40 odd lbs at that distance using modern estimates (25lbs per second at 5f) - or that at Chelmsford the timer is calibrated wrongly and "undersells" every time by ~0.20 to 0.50 depending on the distance - these kind of errors are common place in the game and not many people know about them.

Promote.PNG

Promote 2.PNG

[bobs71]

Maybe Bob can explain
https://www.britishhorseracing.com/pres ... om-1-june/

The reason the courses were re-measured was highlighted in the first place by Time analysts - go on the Racing forum and search "Wetherby distances suspect" or read Simon Rowlands posts on Timeform back in 2014 - it was time data that brought this problem to light in the first place.

[bobs71]

The time from a to b
In differing conditions can be estimated

Exactly , "estimated" is key here , but not by your inadequate methods of a basic spreadsheet and inputting sporting life raw times.

[bobs71]

SeaHorse - how's it going mate? - Hope your still punting strong.Final timefigures - Kempton 08.09.2018.PNGSectionals - Kempton 08.09.2018.PNG

Yes bud.

Well still a little sore after the Son Of Rest dead heat from Saturday. Hope your well?

Regards to your headwind post.
Yep still sat waiting for going readings between each race and some decent live wind speed data coming from the tracks.
Trying to push something behind the scenes... well questioning it to certain people within racing. Hopefully than I may consider speed ratings full time one day.

Good to hear Buddy - give me a shout if you are needing any of that "data" updated - been a few months since we last spoke.

[bobs71]

Agree with bobs71, just subscribe to Timeform - they know what they're doing!

Phil Bull use to say Timefigures tell you how bad a horse isn't.

In other words, a good time proves a horse is good, a bad time doesn't prove one way or the other how good the horse is, because ultimately a horses time depends on how fast the front runners go, not on how good the horse is. Any method that rates horses purely on time will overlook top quality horses that haven't yet put up a good time.

If you're going to guess how fast a horse would run if the jockey didn't lose it's irons, why not just guess what times the non-runners and unseated riders would have gone?

Derek , i can tell you that in the example i put up - Timeform got it wrong that day - they had Enable running faster than Kessaar in the last 3f in their sectional archive - don't know why or don't know how but it happens - see attached snapshot of the two races head to head timed from a common point - Kessaar clearly was faster that day and confirmed that promise on Saturday at Newbury

Kempton Enab v Kess.PNG

[bobs71]

Agree with bobs71, just subscribe to Timeform - they know what they're doing!

Phil Bull use to say Timefigures tell you how bad a horse isn't.

Any method that rates horses purely on time will overlook top quality horses that haven't yet put up a good time.

Agree with your second sentence if only using final time data - the sectional data that breaks down the race is needed to provide context and both work hand in hand - especially in European racing where the early pace is completely different to other jurisdictions (slower) - It's one of the reasons why in UK flat racing there is a big bias towards front runners - looked at on a % of runners to winners basis , frontrunners vastly outrun their final odds and is basically where the money is

DmGJ4xAX4AA_jhm.jpg

That covers a large sample of data on all UK Flat racecourses and uses SP as a benchmark - there is a lot of money in predicting frontrunners , although i will say that most of the profits come from the unexpected types who change runstyle from past history or from jockeys making pace dependant decisions at the start of a race but still a lot can be inferred from that data.

[jimibt]

tell u what
Tell the average to go somewhere
Was ready to swear there
Consider the fastest

StellaBot - trying to decipher your incomprehensible posts is much harder than trying to put a figure on a handicap winner at Chelmsford on a wet and blustery Thursday night - if you had bothered to read my post , i did say "averages don't work" - standards are created by comparing a dataset of past times against expected to actual ability , and the use of statistical techniques involving standard deviation and variable percentiles - It is easier to run faster at 5f than it is to run faster at 16f , so that basic fact of equine bio-mechanics has to be factored in , along with normalisation in relation to the factors that effect times - This process of benchmarking has to be done in the first place to "consider the fastest" otherwise you are working with smoke and mirrors. There is also an ongoing calibration process ,simply because new data is added every single racing day. One aspect that nobody has mentioned is the amount of "false" published or official times that get passed through as if they are accurate - the video analysis (which i know Timeform do as well) is a necessary - to give you an example , have a look at a horse called Promote - who won recently at Newcastle in an "official" time of 71.02 - video analysis shows that was out by ~2.00 full seconds which would equate to around 40 odd lbs at that distance using modern estimates (25lbs per second at 5f) - or that at Chelmsford the timer is calibrated wrongly and "undersells" every time by ~0.20 to 0.50 depending on the distance - these kind of errors are common place in the game and not many people know about them.Promote.PNGPromote 2.PNG

bob -tbh, i've monitored some of the posts from this individual for quite some time and notice periods of erratic behaviour, followed by weeks of silence. at best impulsive but i feel there's a lot complex issues going on that get ranted out on these pages. your efforts to engage and explain are admirable and I for one completely appreciate and understand your research, methodology and approach.

[SeaHorseRacing]

tell u what
Tell the average to go somewhere
Was ready to swear there
Consider the fastest

StellaBot - trying to decipher your incomprehensible posts is much harder than trying to put a figure on a handicap winner at Chelmsford on a wet and blustery Thursday night - if you had bothered to read my post , i did say "averages don't work" - standards are created by comparing a dataset of past times against expected to actual ability , and the use of statistical techniques involving standard deviation and variable percentiles - It is easier to run faster at 5f than it is to run faster at 16f , so that basic fact of equine bio-mechanics has to be factored in , along with normalisation in relation to the factors that effect times - This process of benchmarking has to be done in the first place to "consider the fastest" otherwise you are working with smoke and mirrors. There is also an ongoing calibration process ,simply because new data is added every single racing day. One aspect that nobody has mentioned is the amount of "false" published or official times that get passed through as if they are accurate - the video analysis (which i know Timeform do as well) is a necessary - to give you an example , have a look at a horse called Promote - who won recently at Newcastle in an "official" time of 71.02 - video analysis shows that was out by ~2.00 full seconds which would equate to around 40 odd lbs at that distance using modern estimates (25lbs per second at 5f) - or that at Chelmsford the timer is calibrated wrongly and "undersells" every time by ~0.20 to 0.50 depending on the distance - these kind of errors are common place in the game and not many people know about them.Promote.PNGPromote 2.PNG

As we know this is the case... what I’d like to know.. how accurate are the finishing distances from the winner.

Have you researched any of this?

If they can get simple race times wrong. How wrong are the actual finishing positions in lengths?

[bobs71]

tell u what
Tell the average to go somewhere
Was ready to swear there
Consider the fastest

StellaBot - trying to decipher your incomprehensible posts is much harder than trying to put a figure on a handicap winner at Chelmsford on a wet and blustery Thursday night - if you had bothered to read my post , i did say "averages don't work" - standards are created by comparing a dataset of past times against expected to actual ability , and the use of statistical techniques involving standard deviation and variable percentiles - It is easier to run faster at 5f than it is to run faster at 16f , so that basic fact of equine bio-mechanics has to be factored in , along with normalisation in relation to the factors that effect times - This process of benchmarking has to be done in the first place to "consider the fastest" otherwise you are working with smoke and mirrors. There is also an ongoing calibration process ,simply because new data is added every single racing day. One aspect that nobody has mentioned is the amount of "false" published or official times that get passed through as if they are accurate - the video analysis (which i know Timeform do as well) is a necessary - to give you an example , have a look at a horse called Promote - who won recently at Newcastle in an "official" time of 71.02 - video analysis shows that was out by ~2.00 full seconds which would equate to around 40 odd lbs at that distance using modern estimates (25lbs per second at 5f) - or that at Chelmsford the timer is calibrated wrongly and "undersells" every time by ~0.20 to 0.50 depending on the distance - these kind of errors are common place in the game and not many people know about them.Promote.PNGPromote 2.PNG

As we know this is the case... what I’d like to know.. how accurate are the finishing distances from the winner.

Have you researched any of this?

If they can get simple race times wrong. How wrong are the actual finishing positions in lengths?

Good question Scott - and i can see your thinking - although nowadays a length is a function of time anyway and taken from the photo-finish equipment which is calibrated according to this scale https://www.britishhorseracing.com/wp-c ... tables.pdf- the elapsed time from each runner to runner is then converted to "lengths" as per that scale - but yes , i believe there are occasional discrepancies although on the whole appear accurate enough to infer meaningful data. On the same note the BHA use a handicapping scale which is correctly variable over distance but starts at around 3lbs per length at 5f and works from that basis when assessing a race. This starting point is at odds with what hard data tells us and there is evidence that this starting point should be in the region of 4.5 lbs per length at 5f and worked from there - why does that matter?- well , the average horse and the breed as a whole are being "undersold" in terms of ability for one. The other maybe more relevant point is in Handicap racing , where it has been shown that the average rise in ratings to winners or placed horses is nowhere near large enough to stop them winning more than their fair share of races again , so ethics and fair play come into it. Timeform & others have been at the forefront to try and implement these changes to the official body but the powers that be have really turned a deaf ear and continue with the same scale they have used for literally donkeys. One forward move though is that the BHA are now publishing the finishing times for all horses and not just the winner (see attached) and also their own "performance" figures which is their assessment of individual races and not just the general handicap "mark"

BHA 1.PNG

[SeaHorseRacing]

StellaBot - trying to decipher your incomprehensible posts is much harder than trying to put a figure on a handicap winner at Chelmsford on a wet and blustery Thursday night - if you had bothered to read my post , i did say "averages don't work" - standards are created by comparing a dataset of past times against expected to actual ability , and the use of statistical techniques involving standard deviation and variable percentiles - It is easier to run faster at 5f than it is to run faster at 16f , so that basic fact of equine bio-mechanics has to be factored in , along with normalisation in relation to the factors that effect times - This process of benchmarking has to be done in the first place to "consider the fastest" otherwise you are working with smoke and mirrors. There is also an ongoing calibration process ,simply because new data is added every single racing day. One aspect that nobody has mentioned is the amount of "false" published or official times that get passed through as if they are accurate - the video analysis (which i know Timeform do as well) is a necessary - to give you an example , have a look at a horse called Promote - who won recently at Newcastle in an "official" time of 71.02 - video analysis shows that was out by ~2.00 full seconds which would equate to around 40 odd lbs at that distance using modern estimates (25lbs per second at 5f) - or that at Chelmsford the timer is calibrated wrongly and "undersells" every time by ~0.20 to 0.50 depending on the distance - these kind of errors are common place in the game and not many people know about them.Promote.PNGPromote 2.PNG

As we know this is the case... what I’d like to know.. how accurate are the finishing distances from the winner.

Have you researched any of this?

If they can get simple race times wrong. How wrong are the actual finishing positions in lengths?

Good question Scott - and i can see your thinking - although nowadays a length is a function of time anyway and taken from the photo-finish equipment which is calibrated according to this scale https://www.britishhorseracing.com/wp-c ... tables.pdf- the elapsed time from each runner to runner is then converted to "lengths" as per that scale - but yes , i believe there are occasional discrepancies although on the whole appear accurate enough to infer meaningful data. On the same note the BHA use a handicapping scale which is correctly variable over distance but starts at around 3lbs per length at 5f and works from that basis when assessing a race. This starting point is at odds with what hard data tells us and there is evidence that this starting point should be in the region of 4.5 lbs per length at 5f and worked from there - why does that matter?- well , the average horse and the breed as a whole are being "undersold" in terms of ability for one. The other maybe more relevant point is in Handicap racing , where it has been shown that the average rise in ratings to winners or placed horses is nowhere near large enough to stop them winning more than their fair share of races again , so ethics and fair play come into it. Timeform & others have been at the forefront to try and implement these changes to the official body but the powers that be have really turned a deaf ear and continue with the same scale they have used for literally donkeys. One forward move though is that the BHA are now publishing the finishing times for all horses and not just the winner (see attached) and also their own "performance" figures which is their assessment of individual races and not just the general handicap "mark" BHA 1.PNG

I have noticed that BHA have put actual times up. Have they done that for all tracks?

[bobs71]

Yes Scott , been ongoing now for a couple of years with the published times , although the "performance" rating is relatively new - think its a bid to increase transparency , so owners/ trainers can understand their thinking behind the movement of "marks"

[Derek27]

tell u what
Tell the average to go somewhere
Was ready to swear there
Consider the fastest

StellaBot - trying to decipher your incomprehensible posts is much harder than trying to put a figure on a handicap winner at Chelmsford on a wet and blustery Thursday night

+1 I don't know why we bother wasting time with this guy. At least our talented little champion made his best effort to explain himself!

[Derek27]

Agree with bobs71, just subscribe to Timeform - they know what they're doing!

Phil Bull use to say Timefigures tell you how bad a horse isn't.

Any method that rates horses purely on time will overlook top quality horses that haven't yet put up a good time.

Agree with your second sentence if only using final time data - the sectional data that breaks down the race is needed to provide context and both work hand in hand - especially in European racing where the early pace is completely different to other jurisdictions (slower) - It's one of the reasons why in UK flat racing there is a big bias towards front runners - looked at on a % of runners to winners basis , frontrunners vastly outrun their final odds and is basically where the money is DmGJ4xAX4AA_jhm.jpg That covers a large sample of data on all UK Flat racecourses and uses SP as a benchmark - there is a lot of money in predicting frontrunners , although i will say that most of the profits come from the unexpected types who change runstyle from past history or from jockeys making pace dependant decisions at the start of a race but still a lot can be inferred from that data.

I'm well aware of how sectionals are used, I was just explaining what Phil Bull said - we didn't have sectionals at that time.

[Derek27]

Derek , i can tell you that in the example i put up - Timeform got it wrong that day - they had Enable running faster than Kessaar in the last 3f in their sectional archive - don't know why or don't know how but it happens - see attached snapshot of the two races head to head timed from a common point - Kessaar clearly was faster that day and confirmed that promise on Saturday at NewburyKempton Enab v Kess.PNG

I've always had a lot of respect for Timeform's race-readers, products, and their attention to detail, but in fairness, with thousands of races and plus ten thousand horses to assess using error-prone techniques, the errors are bound to be there. Even more so when it comes to rating horses - maiden winners often have to be uprated or downgraded after their next run. The real benefit is having a team putting in the hours for a job that's too much for one man.

[StellaBot]

Anyone got it?