Application of Regression Analysis To Greyhounds

[thebrowngrowler]

Hi,

I am interested in learning about the application of regression analysis to both Horse Racing and Greyhound Racing. However, my interest is primarily Greyhound Racing.

Would anyone please be able to point me in the direction of any articles or info regarding studies of the application of regression analysis to greyhound racing results and its use as a predictor.

I realise that we are in an equine forum but thought that some members may have cross code interests in our canine friends.

rgds
growler

[Paging]

The biggest issue to overcome with greyhound racing is the constant lack of quality information contained within the charts.


If you think the chart writing with horses can be "iffy" watch a few greyhound races and check the charts. This has been a constant complaint over on a greyhound message board. Most of the greyhound tracks these days employ a chartwriter that was a leadout (walking the dogs to the starting box) the previous meet. Of course some tracks are better than others.

Years ago it was unreal. Graded "C" dogs from one track appearing in maiden races at another. Dogs often missing there most recent last few past performances etc.. But once again it was opportunities abound $$$.

[thebrowngrowler]

Paging,

The race results data sets I wish to apply the regression analysis to are from tracks based in the UK where they are operating under a fairly strict set of guidelines and rules set up by the National Governing Authority. There is a lot of uniformity associated with the race results data so it would appear to be a reasonable safe model to analyse.

What I hoped to find out was the results of any past studies and how the regression analysis was designed.

thx
thegrowler

[ryesteve]

Didn't someone here recently reference their experience in modeling greyhound races? I think maybe it was Traynor. If you search the board, you're likely to find it.

[1st time lasix]

My simulcast venue is a dog track clubhouse. Been watching guys lose betting on the hounds for many years. It may be the single worst way to gamble in the world I can think of....small pools, big takes, only eight horse fields or less, dogs getting hammered at the start and getting wiped out on the first turn. At least in Vegas you get a free room and beverages for throwing your money away in the slots. Do yourself a big favor....do not bet another dog race the rest of your life!

[1st time lasix]

My simulcast venue is a dog track clubhouse. Been watching guys lose betting on the hounds for many years. It may be the single worst way to gamble in the world I can think of....small pools, big takes, only eight horse fields or less, dogs getting hammered at the start and getting wiped out on the first turn. At least in Vegas you get a free room and beverages for throwing your money away in the slots. Do yourself a big favor....do not bet another dog race the rest of your life! Perhaps this may sound a bit harsh...{hey to each his own vice}....but it is difficult enough to win with the horses and they have rules about bumping and fouls.

[Paging]

My simulcast venue is a dog track clubhouse. Been watching guys lose betting on the hounds for many years. It may be the single worst way to gamble in the world I can think of....small pools, big takes, only eight horse fields or less, dogs getting hammered at the start and getting wiped out on the first turn. At least in Vegas you get a free room and beverages for throwing your money away in the slots. Do yourself a big favor....do not bet another dog race the rest of your life!

Have to totally disagree with you. I personally have found the greyhound to be quite profitable. That is more or less a past statement as the pools have diminished. Believe it or not I've come across more professional greyhound players (ones that sole income was the dogs for 20+ years) in my life that horse players. Maybe just the circles I hang out in?

Also, playing the greyhounds in the UK one would receive fixed odds with a bookmaker so pools aren't an issue.

[Paging]

Paging,

The race results data sets I wish to apply the regression analysis to are from tracks based in the UK where they are operating under a fairly strict set of guidelines and rules set up by the National Governing Authority. There is a lot of uniformity associated with the race results data so it would appear to be a reasonable safe model to analyse.

What I hoped to find out was the results of any past studies and how the regression analysis was designed.

thx
thegrowler

I'm not familiar with how accurate the charts are from you UK so as you say it maybe just fine.

However, I would take any past studies from the United States with a grain of salt as the input data may not be that accurate to start with.

[thebrowngrowler]

lasix,

As "paging" has said, in the UK there are trackside bookmakers who offer bets on the racing. They compete with one another for business so you do not have to bet into a pool with massive take outs by the track operator.

There are a multitude of options for betting that do not avoid a pool run by the track operator. For example, there are online betting exchanges operating on a back or lay basis, there are also online bookmakers offering odds as well as the track bookmakers.

By looking around it should be possible to bet to a 103% to 107% book, at times the bookmakers bet overbroke, ie, you could back every runner to appropriate stakes and win.

It is a very different scene in the UK as regards betting on greyhounds when compared to the USA.

growler

[1st time lasix]

lasix,

As "paging" has said, in the UK there are trackside bookmakers who offer bets on the racing. They compete with one another for business so you do not have to bet into a pool with massive take outs by the track operator.

There are a multitude of options for betting that do not avoid a pool run by the track operator. For example, there are online betting exchanges operating on a back or lay basis, there are also online bookmakers offering odds as well as the track bookmakers.

By looking around it should be possible to bet to a 103% to 107% book, at times the bookmakers bet overbroke, ie, you could back every runner to appropriate stakes and win.

It is a very different scene in the UK as regards betting on greyhounds when compared to the USA.

growler Thank you....I appreciate the international knowledge. I wasn't aware of that possibility. Learn something new every day. One thing I have learned already based on decades of observation....no one..... and I mean no one beats the dogs in the US if they play regularly now. Show me somebody who says they do....and I will show you a gambling degenerate who has had his heart broken on the first turn so many times he has lost all his positive attitude and perception of reality. There is a reason the US dog venues are 10%-20% of capacity. They have burned out the majority of their patrons. As the pools get much smaller....and the take stays high....a player can't win. Only reason they even exist now in this country is for horse simulcasting rights, poker rooms and the possibility of slots. Otherwise they wouldn't be able to stand on their own. Derby Lane in St Petersburg Florida may be the best dog track in the country. Beautuful track, long tradition, good food and great weather. Yet it is a ghost town of gamblers except where the horse simulcast is being shown and where the poker rooms are. Across town....Tampa Bay Downs has one of the fastest growing para-mutual handles anywhere.

[Paging]

lasix,

As "paging" has said, in the UK there are trackside bookmakers who offer bets on the racing. They compete with one another for business so you do not have to bet into a pool with massive take outs by the track operator.

There are a multitude of options for betting that do not avoid a pool run by the track operator. For example, there are online betting exchanges operating on a back or lay basis, there are also online bookmakers offering odds as well as the track bookmakers.

By looking around it should be possible to bet to a 103% to 107% book, at times the bookmakers bet overbroke, ie, you could back every runner to appropriate stakes and win.

It is a very different scene in the UK as regards betting on greyhounds when compared to the USA.

growler

Also they run 6 dog races in the UK vs the field of 8 in the US, often with an outside lure.

[46zilzal]

They BLEND in the individual aspects of a single race into an average "truism" which makes a particular contest's unique nature disappear.

[robert99]

46zilzal,

Don't like to make negative posts but you are absolutely right.
The average dog/horse loses. A winning dog/horse is exceptional.
Any analysis that does not seek to find the cause of that exception but goes for fuzzy averaging is of little long term use ie the same as economists that make up this econometric regression nonsense.

If you bet on UK dog tracks then chose the major, larger circumference ones. Track bookie odds are diabolical and person to person betting does not cater for any substantial betting for all meetings - not that they will take money from USA at the moment.

[garyoz]

Hi,

I am interested in learning about the application of regression analysis to both Horse Racing and Greyhound Racing. However, my interest is primarily Greyhound Racing.

Would anyone please be able to point me in the direction of any articles or info regarding studies of the application of regression analysis to greyhound racing results and its use as a predictor.


The best stuff I have read about regression vis a vis handicapping has been on this board. It doesn't work well, too many highly correlated variables plus violation of basic assumptions of this statistical approach (violation of the assumption of random distribution of error terms). Do a search on regression or multiple regression on the site. Nothing published (to my knowledge) in the area because of the above problems.

Traynor may jump on and try to snow you, but these are the facts.

[traynor]

The best stuff I have read about regression vis a vis handicapping has been on this board. It doesn't work well, too many highly correlated variables plus violation of basic assumptions of this statistical approach (violation of the assumption of random distribution of error terms). Do a search on regression or multiple regression on the site. Nothing published (to my knowledge) in the area because of the above problems.

Traynor may jump on and try to snow you, but these are the facts.

Nope. The sterling advice already doled out pretty sums up the body of knowledge. I have nothing to add.

[cliftonarms]

Growler

I am a UK greyhound follower with a huge amount of complete electronic form data from all the UK tracks.

So far I have tried stats, neural, decision tree and decision forest prediction techniques. I have not so far tried linear regression but did once test on binominal regression.

My best continual correct prediction rate is 33% and you need 31% to beat the bookmakers in the UK, so a few more percent and I'll be there.

Perhaps we should talk.?

Cliftonarms