Forumites experiences of number-crunching/stats-based approach

I don't do the horses, only footy, but i suppose the last quote is the same in all betting - because heads and tails has evened out to around 50% each over the course of a thousand coin flips doesn't mean that the next 1000 won't produce 800 heads and 200 tails.

It's just very unlikely. Form and results are the only things we have to go on to give us the information we need. Man United will probably beat Norwich this weekend (hope they do because i'm going to it) - how can we say this? On the day Norwich might be better? Yeah, but all we have to go on is the fact that in the past United have been much better.

ALWAYS study form, otherwise you're just picking blindly in my opinion.

Duncan - I don't study form / statistics or anything like that, I use a purely mathematical approach.  I did read a post on here some time ago where someone stated something along the lines of "results inform statistics, statistics don't inform results".

I agree with this, although I'm not knocking form study / statistics analysis if you can get an edge from it, which is what it's all about in my opinion.

If you're betting pre-off, there isn't much to go on other than form. Just taking the figures at face value won't really get you anywhere though. You've got to know when a particular statistic is likely to be wrong, and how.   And which statistics, collectively, give you the best indication of a team's overall advantage or disadvantage over today's opposition - and to what extent.  Only then can you have any real idea of whether the price being offered will allow you to make money in the long term.

What you really want to unearth are those initially hidden statistics that seem to go against common wisdom - the average side that virtually never loses at home, the highly rated side that tends not to score until the 2nd half, the mid-table side that's got some of the best current form in the league etc.  If you can catch those trends before everybody else wakes up to them, you might have something.

Cheers, fellas. Yes, form has always been my thing and there is no doubt that form study, done correctly, gives you an edge. Have only recently branched out into stats/number-crunching, and my jury is still out, hence the thread!

wayne, i appreciate there are a lot of players on here that adopt purely a mathematics approach like yourself and in some respects i'm jealous of that! Alas, i know that's not where my strengths/interest lies, so it was more stats than maths i was referring to in OP.

duncan - speak to swift tuttle on fr froum about hrb - he uses that alot

fr froum = hr forum

cheers delta- yes, he's red hot is swift

Don't forget bookmakers also have all the access to stats. They have raceform interactive so the top rated in 2010 was 5-1, today it's only 4-1, so there goes your profit. As people get smarter and a wider range of stats become readily available, The value now lies in going against the stats. But because of commission you can't just Press lay, the value is betting something Else in horse Racing and in other markets (not match odds) football.

But As has been Said a thousand times before, the only thing that matters is value. If All the stats and thousands of people are on the same thing, the value is someplace Else.

I read a book about value and i Think the author used raceform in his systems so probably hundreds more are using raceform so it's lost its edge. There are hundreds of systems out there, some last forever. One system is not enough, you need to find lots, to feed the turnover required. Remember 5-10% roi is the max, so turning over enough bets and money is the problem.

Don't make the analysis too complex, the simplest methods make money. I used to spend hours on a race, bogged myself down for a couple of years. Now i Can go through a card in 5 mins and make money. But thinking about it, if i didn't bog myself down and work hard i wouldn't have Found Them.

Duncan,

I am a betting market analyst, not a gambler.

I've heard a lot of people say that studying past results/stats is absolutely no guide to future results.

This is true regarding the outcome of a subset of races, it is untrue regarding the outcome of an overall set of races.

Bookmakers and sports pricers make up their markets from statistics and have built in an overround which has served them well for hundreds of years. The statistics will tell you what you know, market movement will tell you what is not in the statistics, or what you don't know.

Statistics will tell you the chances of an event occurring or the likelihood of the horse saluting, but it won't tell you the winner.

Think about this scenario, where a man is accused of murder and all evidence points to him being the guilty party, a jury of his peers unanimously find him guilty.
He is somewhere between a price of $1.01 and $1.05 of being guilty.
It's important to note, that he may not actually be guilty.
If you can accept that, then it will make a whole lot more sense.

There are people who have been put to death and on death row, who were or are innocent. There are people walking amongst us who are guilty and got let off.

Evidence and a jury of only 12 people can be wrong.
That's why TV ratings and Political opinion polls are not akways right, there is not a large enough sample of varied opinion types to make it accurate enough.
In the case of a jury, it's even worse, because they are often directly influenced by emotion, the judge, the lawyers, and eachother. Peer pressure plays a huge part. Very few jurors want to be the odd man out, except in rare circumstances. It's human nature.

I am a betting market analyst, not a gambler.

I.quit.my.country, agree that the more available information is, the less point/edge there is in using it (i've always thought that re. Timeform comments and ratings)...however, not so sure that bookies spend hours running queries about first-time blinkers on a L-H track etc on the RI database! I could be wrong tho.

Interesting when you say 'if i didn't bog myself down and work hard i wouldn't have found them.' Is that because it took that long for you to realise simple was the way to go? or was it simply  a case of 2 yrs trial and error to unearth something worthwhile?

The value thing is a given...however, while i can spot value from form study, i havent reached the point when i can be sure the stats are telling me something is value, prob cos i've not enough experience of them to know which stats to trust.

odds on shots always get beat is
best approach in the right races
no point in laying them .
real value is with the winner.
you have to find it.

Beat The Overround, yes, i'd imagine the stats would always only be a steer in the right direction, possibly a way of narrowing down the options with some validity, rather than being able to tell you the winner.

I've never been able to do better than a liberal reading of form, combined with an understanding of factors that aren't evident without figures extrapolated from form.

These will be based on an appreciation of the characteristics of particular horses (e.g. like to do better for a stronger pace, a rail draw, a turning track, a step up in trip, better ground, with normal improvements etc.), of particular trainer methods (strengthened by market movements in many cases) and of certain races (e.g. one side of the track was overwatered, the time was fast in relation to deteriorating ground, the race became a sprint, the race was a competitive maiden or hcap and will produce winners etc.) The more utterly specific the perception, the better the chance of making money.

For this reason, I've never been able to make stats work for me other than (a minor exception) for big-race trends.

*or evident in the brute figs. taken from form

"There are lies, damn lies, statistics, and then there is form!"

From The Sayings of Dr Slicer, available at all not very good bookshops nowhere in the country.

I hope this is of assistance.

Bit surprised there havent been more replies from people strongly in either the For or Against camps regards stats. Expected some to say that they had spent years number crunching and eventually given up cos they could never find a reliable edge, and others to say that anyone who rubbishes stats is looking at the wrong stats (or interpreting the stats wrongly).

Chiefly started the thread cos General Betting seems to have gone pretty flat lately and i thought this would be a subject that would create debate...but obviously not!

"stats" can be used to 'narrow' your search for a play on an event.

Example:  When I first watched racing, Frankie Dettori was at the top of his game.  However, on horses at odds of 8/1 or better, he was rarely winning the race, often finishing out of the frame.  Punters following FD were pushing the price in.  This was not saying that the horse was a lay but probably meant the winner was elsewhere and FD's horse was not a threat.

Recently I have been preparing my horse ratings for 2012 flat season.  I've also been looking at stats for draw.  Even listening to pundits, no one seems to be taking enough of an in depth look at the draw stats to make it worth while.  I've worked the 'grid' a bit more on this which factors in the draw of all the all the horses in the race and how it weighs on the outcome.  Simply "counting" how many times a horse wins from draw 1-2-3 is at Lingfield over X furlongs is too general "a stat".

A simple example would be Football O/U 2.5.  You need to create your own criteria and not shy away from playing odds against when your approach says that is the way to go. I find it best to research the plays WITHOUT knowing the odds and betting, create my own expectation and if the odds are equal or better, put the bet on.

Stats can help you find a value play.  On football, I have five bets I look at:

1X2
Asian Handicap
HT/FT
and
U/O 2.5

I'll build a foundation using stats, then recent form and RECENT history between teams and their mutual competitors in recent matches.  I'll play one or more of the above once I have drawn my own picture of how I think the match will progress and finish.

Simple example: if Home team is 4/10 and extremely confident they will win and score, the bet can be 50% on the 1X2 and 50% U/O.  Or if the visitors have been giving away goals in the first 30 minutes, HT/FT can be factored in (though I find the value in that type of bet is usually your confidence that one team will hold the match to a draw at HT but an outcome other than draw is predicted).
 

Stats will narrow your search but cannot be the end all.  There is no value in that.

I'm very much in the "for" camp.

Cooee, don't take this the wrong way - it's only my view. But

"What you really want to unearth are those initially hidden statistics that seem to go against common wisdom - the average side that virtually never loses at home, the highly rated side that tends not to score until the 2nd half, the mid-table side that's got some of the best current form in the league etc.  If you can catch those trends before everybody else wakes up to them, you might have something."

I think is a horrible way to approach it simply because if nobody else has "woken up to them" it's normally because the trend is so short-term and means nothing. You can convince yourself you've found a "trend" for most things if you look hard enough. Sometimes you'll pat yourself on the back because the trend continues, often you'll be wondering why you bothered.

That said, there is certainly plenty of common wisdom which is far from wise.

Duncan in response to your question; some would say two years of trial and error but i Call it a two year Education. You are right about the forum now, it offers nothing. I only posted because i had some time to kill in a hotel room and was trying to help someone on Another thread which didn't work by the looks of it.

Shapeshifter dó you make money predicting future events? You May say yes, But here's the big question; dó you get enough turnover (bets) a year to make thousands of pounds? Lets be honest it's very hard to get big money on unless you bet favs and are prépared to take All prices. But then you have no margin for error and roi is minimal.

Only after recording 100s of results using identical filters Can you get an understanding of what Will happen in the future. Then you don't know if the next 100 Will produce the same results, but with a bit of logic and experience you fathom which are Golden systems and will last a long time and which are just random. Anything under 6 is a small price for mé nowadays, the long losing runs are common but i have a bigger margin of error and when they hit, they hit. I don't Think i could achieve the same roi using my mystic powers even if i spent a week on a race. I certainly wouldn't get the turnover which half a dozen systems give mé or the time to dó other things.

The reason anything under 6 is a small price is because most of my systems go completely the opposite of what i set out to dó. So Many would discard the system when it Becomes obvious it doesn't work, but if you keep recording and look deeper, you find something that Will work. All Info you collect is valuable somewhere down the line.

Remember there are hundreds of variables in a race or match, so unexpected results happen All the time. This is how bookmakers used to make a living, knowing these freak results Will happen regular and that humans were incredibly bad at predicting future events. But whereas they might offer a fair price on a favourite; they never used to offer a fair price on an outsider. Today it is different.

Horse Racing; some 16-1 shots should be 33-1 shots, but some should be 8-1 and often you get 20+ on here. Same with the 25-1 shots, 40+ on here, 7-1 shots, sometimes 10 on here. I play Mostly on the aw where there are less variables and it's a more level playing field. In horse Racing (low grade) alot of the races, recent form goes out the window As the trainers play a cat and mouse game with the handicappers. They know the horse is not going to be a World-beater so its All about getting enough prize money just to cover the fees and if you Can get the owners a Nice little plot at a big price- even better.

Football: it's amazing how accurate the match prices are. You Can record a thousand matches and you Will break even ( backing) and lose laying with commission. But it's a different story in other markets.

Dog Racing: great prices but low liquidity. But every penny counts.

General betting folk can't get there head around backing outsiders and i must admit it took mé a long time.

Statistics show that in horse racing a third of all faves win year in year out. Yet I remember a few years ago there was a run of 46 consecutive losing faves. Statistics eh!!!!

"There are lies, damn lies, statistics, and then there is form!"

Statistics don't lie, it is the reliance on small data, dirty data or the interpretation of them which leads to lies.
There are more lies in form than in statistics imho.

Statistics show that in horse racing a third of all faves win year in year out. Yet I remember a few years ago there was a run of 46 consecutive losing faves. Statistics eh!!!!

It doesn't say anything about statistics.
I also remember probably 5 years ago, favourites won everything everywhere for almost two weeks.
It's called variation, nothing more.

"It's called variation, nothing more."

Actually it's called knowing which of the third of all faves are the ones that are going to win!

Hope this helps.

Statistics are an extremely valuable tool in my opinion, but it all comes down to how you interpret them. You can find stats to argue almost any selection if you look hard enough.

I've often seen data which has led a friend and I to draw completely opposing conclusions. Identifying the significant information is the difficult skill.

I don't include horse racing, but as far as sports betting goes, I have 6 months of stats-based selections which have produced a 67 percent winning strike rate at odds of Evens, and including commission.

I should add though that some personal experience and knowledge are used to further sift what the data throws up, before making final selections, so in that sense it's not purely and solely stats based.

67 percent seems pretty good to me though. I don't know how that compares with others experiences?

interesting thread.

HORSE RACING -

APPROACH - First factor to consider is....what is GOING?

Going varies and this is the main factor IMO .......You then check out other factors that are relevant to the race.

The AW racing varies re GOING......so you can't ignore checking GOING.

During racing again the GOING can change drastically, so if it does change you should review your initial selection.

Weather factors can compound the GOING position......eg Strong winds ; cold weather etc.....Seaside courses can dry out as racing progresses.

Certain horses may run better later in day , and some horses will not run as well later in the day/night. May well be linked to temperature changes ...especially in summer months...I know this is not strictly to do with GOING....but it is part of the puzzle.

It is this complexity that makes it difficult to go by NUMBER CRUNCHING alone. You need to look further to find edges that the average punter just can't see. A horse's race record alone is not enough.

If you learn to understand why some horses run well on certain types of GOING, but not on other types of GOING, then you then will have a greater understanding when weighing up horses that have many factors in their favour....but say a strong wind exists that will suit other horse types etc.

It is an extremly complex puzzle, but past results can provide valuable information that can assist you in rating a race. If you ignore past results, you are a fool IMO......

Start with "simple" race types ( Maiden Stakes) to get the "simple factors.......then gradually weigh up more complex race types ( H/Cs etc where same age horses compete against each other). Eventually move on to mixed age type races, it is then you will maybe find edges that the average punter is just not aware of.

If you keep trying to puzzle out why a horse won a race....and why your selection did not, then this is probably the "best" way to improve your knowledge of the game.

GL with bets.

all weather racing.
never check the going.
getting a value price
is more important.
usually dont back the
favs most of the time.

Brilliant thread, well done Duncan on the topic and all posters contributing and I agree completely with everyones general consensus that data and databases are very useful in the
practice of rating horses in races to arrive at more accurate predictions.

From my experience one should be most careful when trying to apply these trends and factors
for a number of reasons, firstly to study a profile or set of factors that have been profitable in the past is no guarantee that this will be reproduced in the future as the profit could just be random clustering and not robust co-efficients that can be relied upon, secondly the fact one has then to find rare combinations of form profile means one arrives at a smaller set of examples possible over time thus limiting the effectiveness of the selection process !

For me after years of research though not constant work, more many mini projects the logical conclusion that I have arrived at is to firstly rate all the runners in a race using points system that has the points as relative as possible to some form of measure of the runners
recent runs, beit lengths, parts of time or offial points ratings, then one can focus on the top third of the ratings that you arrive at, then you can search the market results for this top third for each race also taking into account the spread patterns of points and then it can be seen what parts of the markets are profitable on betfair and what is historically loss making.

This may seem quite simple a method for rating but dont forget you can add any points you want in for anything including the going so it does trully work and will arrive at a means to an end, that being profit!

I think is a horrible way to approach it simply because if nobody else has "woken up to them" it's normally because the trend is so short-term and means nothing.

Silverback, I think you just have more respect for the betting community than I do. I believe I actually used the phrase 'everybody has woken up' rather than 'nobody has woken up'. And by that I mean that I wasn't implying that I was the only person who might have uncovered a particular angle - just as long as the typical gambler isn't wise to it, you may well have something you can profit from. Most gamblers take the majority of stats at face value rather than testing them to see if the stats really mean what they appear to mean. And when the majority thinks one thing, there are often gaps there you can exploit.

deep study or overall awareness?

I found out that I do much better if I only have a very limited knowledge of what the stats are, it keeps my mind open and prevents what I call, the dreaded "surely/they must" sindrome to kick in, thats also why I use an exchange, it allows me to give stats the relative value they have and bet according to what I'm seeing
Thousands of bets made over the years and a lot of incredible situations witnessed over those same years gives me a pretty good notion of where I am when faced with the ever changing conditions of whatever I'm betting on, but I never take nothing for granted and leave all possibilities on the table
I think I've became a pretty good judge of those situations (finding value), there is only one problem... me, myself and I
Unfortunately I´m not always in that same optimum frame of mind and the time it takes me to acknowlegde that...maybe...I'm not in the right frame of mind and actually making the decision to leave it alone, it´s when the swings start
Whatching my graph over a big period of time, it becomes quite noticeable that those optimum frame of mind periods deviations aren´t a day to day affair but many times weeks if not a month periods where I just should've taken a longer break and regroup
When I started I obviously studied really hard and made countless excel sheets with ever increasing number of stats but at some point those became less important and what became of utter importance was/is where am I

So stats are important but to me are far from being the holy grail of betting (at least in my experience of betting on soccer)

Having said that I've been betting more and more on the horses and it seems a lot more dependable on the stats because it has much more variables to take into account than soccer

I find once you take into account that you are:

(a) betting on something that eats while you are asleep and can stumble at any time
(b) betting a school yard game where men are paid millions to kick a ball
(c) betting on a U.S. sport where the ball is not round and is thrown through the air
(d) betting on a sport that involves a hard rubber disc that bounces wildly around being hit by sticks while men skate around
(e)
(f)
etc

It makes everything clearer

From my experience one should be most careful when trying to apply these trends and factors
for a number of reasons, firstly to study a profile or set of factors that have been profitable in the past is no guarantee that this will be reproduced in the future as the profit could just be random clustering and not robust co-efficients that can be relied upon, secondly the fact one has then to find rare combinations of form profile means one arrives at a smaller set of examples possible over time thus limiting the effectiveness of the selection process !

Thats where common sense comes into play, some are Golden eggs, some go in the dustbin.

From my experience one should be most careful when trying to apply these trends and factors
for a number of reasons, firstly to study a profile or set of factors that have been profitable in the past is no guarantee that this will be reproduced in the future as the profit could just be random clustering and not robust co-efficients that can be relied upon, secondly the fact one has then to find rare combinations of form profile means one arrives at a smaller set of examples possible over time thus limiting the effectiveness of the selection process !

Thats where common sense comes into play, some are Golden eggs, some go in the dustbin.

Why is everyone so obsessed with the Best stats and the Best rated? Long way it continue for my sake. Like shapeshifter just Said there are so Many variables in sport that we just Cannot measure. All i know for a fact is that these variables (freak results) Will ALWAYS happen in sport, it's what makes it so unpredictable and it's how the bookmakers used to make there money. Cos they wouldn't give you value on an outsider.

Coeee's last post hit the nail on the head.

Dfc you are getting too bogged down. There are so Many variables but some are more important than others. GOING i would factor in 10% of my overall rating. Its an important factor for sure. I would class AW, flat and jumps As three diff sports. 75% of my betting on horses is in AW hcaps where there is tons of course and recent form. When the flat starts it Becomes a lot more complex, with horses running off diff breaks, up/down in distances and over diff courses and going. I haven't watched a jump race for 2 years, not interested in it. Some of my Best methods are simple, but effective. When you start looking for chip bags blowing across the track, 1f out at great yarmouth, you know you are going too far man.

IQMC -

All depends how high your standards are. If you believe that a STRONG WIND will not indicate what type of horse can run well, then so be it.

YARMOUTH is a seaside course, and GOOD TO SOFT going can change by latter races to GOOD or GOOD(GTFIP) etc , depending on strength of wind +sun shining.

If you believe that GOING is only a 10% factor for AW racing .......generally over course of year that might be OK for STANDARD going. Though SLOW / FAST AW tracks will effect the outcome of races, and 10% is far too low then...IMO.

However, with TURF and NH racing GOING it is extremely important, and understanding how changes in going will effect outcomes is well worth pursuing.

You are "not interested" in "jumps" racing......so perhaps the GOING FACTOR has not been fully considered by you. As for TURF racing, I suspect you do most of betting when better weather is on......GOOD TO FIRM.....therefor as there is some CONSISTENCY in going when you do your betting, the impact of CHANGES IN GOING perhaps has not been fully considered. Perhaps you use other STATS in your appraisal.

Anyway , as long as you are "happy" rating races the way you do, then fine........I'll continue to rate using factors relevant to conditions.

GL with bets.

The weather plays a big part in all sports that are outdoors and must effect horse racing the greatest as one can plainly see when the grand national is run on a seaside course and when its windy and wet few horses finish the race compared to good going on an over-cast day, also bright sunlight must effect horses vision when jumping or am I going to far

If you keep trying to puzzle out why a horse won a race....and why your selection did not, then this is probably the "best" way to improve your knowledge of the game.



DFC, totally agree with that...that's what i tell people when they ask how to get started studying horse racing.


However....! having punted successfully for 25 yrs, it's ironic therefore that the one factor which i believe most people overstate the importance of is the GOING! But, as this forum illustrates, there are many, many ways to crack a nut when it comes to betting and whatever works for each individual is what matters.

In terms of all the variables there are to take into account, i read this in my notes last night (i hadnt noted where it originates from):-

Research has shown that the more handicapping information a person has, the less acurate his decision making. When weighing the relative value of numerous horse racing handicapping information variables, the decision maing process begins to break down when more than 5 of those variables are considered.

(That's using 'handicapping' in the US sense). It's a little hard to go with cos there are clearly more than 5 things that have to be taken into account when making a selection, but i can agree that it's wise to limit the factors that you put great weight behind.

There're obviously dozens of factors that could be important in horse racing, but you can't think deeply about all of those things or you'll go mad.   I have five or six important factors that I concentrate on. Quite often those will be enough for me to decide whether or not to play something in that race.   

However, at the same time, my mind will be unconsciously wandering about considering other aspects of the form. Sometimes I'll have an uneasy feeling in my gut, and that'll be a sign that my subconscious mind has picked up on something that I've missed in my main analysis. In that case I'll start going through the more obscure factors to see if I can work out what the problem is. I'll know if I find the cause of that unease, as I'll stop feeling uncomfortable the second I've noticed it..

A lot of novice punters can't understand why somebody with experience can't just explain their system to them. Even if that person wanted to explain it, they usually can't. I think that's because an important chunk of our analysis is in the unconscious - the sum total of all the thousands of races and race-cards that we've looked over in our time. Much of that isn't available to us to access as such, but it's still there in the backs of our minds, should we need to recall it. I suppose that's our mind's way of handling the thousands of experiences and memories that we form throughout our lives - whether for gambling purposes or not.

the sum total of all the thousands of races and race-cards that we've looked over in our time. Much of that isn't available to us to access as such, but it's still there in the backs of our minds, should we need to recall it. I suppose that's our mind's way of handling the thousands of experiences and memories that we form throughout our lives -
very true

A couple of points to be wary about.

Sometimes, in fact quite often the back of our mind is wrong from a statistical point. We remember the big loss, the big win, the bad day, the great day and the subconcious distorts reality, because we don't remember the day in day out variations, long term trends etc.
Just look at some of the punters golden rules which everyone thought would be a path to gold, now we could lay those horses.
Because much of it wasn't backed up with data, it soon became apparent that a lot of those rules were not correct after all.
They were perception rules which Joe Punter and a few of his mates perpetuated after a bad run.

With regard to statistics, most make the mistake of working with errored data in the first place.
One cannot ever hope to come up with anything reflecting reality if the data is incorrect to start with.
A lot of my time is spent crosschecking data from two or three independant avenues, and the the astonishment at how much data contains errors. Without exception every single day I find errors in data. I also spend investigation time determining the correct information.

Without this, you're pretty much buggered.

Similar views to Beat The OverRound, I find that you need to look at HOW a result came about.

Simple example is looking at a 2-1 football result.  If two of these goals were off penalties, should the performance of the teams be considered an "UNDER 2.5"?  If all three goals came down in the last fifteen minutes, what does it say about the defensive strength of each squad?

Horses;  One pundit:  "On his last three outings, he has placed each time".  Yes, but what were the distances?  There is a big difference on placing a length and ten lengths.  Also, in the case of hurdles, the 2F and 3F fell on one occasion.

And similar to what Cooee said, you need to find that balance between sitting on the fact and going a bit deeper.  Too deep and everything gets muddled.  To shallow and you might as well play roulette.

Surely making your own book from your ratings are their only use. I had a spell of  compiling ratings based on 10 main factors, backing All horses in a race at 120% and laying All at 80% of my price. After hundreds of races, i broke even, meaning i wasn't far away.

Now i don't make ratings, i have quicker ways of selection process. It's very hard to make money from ratings on their own but you learn alot. This exercise gave mé the confidence to back the horses lower down As this is where the value generally lies.

I was watching the snooker yesterday and i thought how competetive the standard is now. No player is dominating. It's 7-1 bar co favs every tournament. I also looked at how the format has changed now with 5 out of the 8 World ranking events this season taking place abroad (4 in the far east and australia). Now i know players are not used to this and the old guard don't like being away from home.

how Can i use this to my advantage? A morning spent going through the last 2 seasons WR events results and i have Another 50 or so bets a year to add to the turnover. I'm not sure how much liquidity there is but it should bring in An extra few hundred pounds a year. I ask the question: is the system sustainable for genuine reasons? And the answer is yes; As long As the players remain so closely matched and they are jet setting around the World.

This is an example of strategy, never stop looking for more angles. The more turnover you get, the more money you make.

Too deep and everything gets muddled.  To shallow and you might as well play roulette.

This is gold, agree 100%

I.quit.my.country 29 Feb 12 10:52 
Surely making your own book from your ratings are their only use. I had a spell of  compiling ratings based on 10 main factors, backing All horses in a race at 120% and laying All at 80% of my price. After hundreds of races, i broke even, meaning i wasn't far away.

Now i don't make ratings, i have quicker ways of selection process. It's very hard to make money from ratings on their own but you learn alot. This exercise gave mé the confidence to back the horses lower down As this is where the value generally lies.
===================================================================


IQMC - Ratings are difficult to fully appreciate how to use them. Certainly has taken me long enough to understand how to use them after rating races in 1998.

IMO - you can only see how to use them after carrying out DETAILED analysis work on them. Such "work" has taken me about 2 years working on them on and off. It is such boring "work" , but is essential to do this work before you can check out strategies using ratings.

Doing a BOOK is not what I see as the use of RATINGS........but they are an essenetial tool to find VALUE bets. The results of BACKING and LAYING on the strategies appears to give a chance of high ROI, whether it is attacking the short odds end of market or top of ratings or high odds end of market. No LOSSES indicated on ANY of the chosen strategies.....only "god" profit. This is not by using a BOOK approach.....

I always knew I had devised a system to help get VALUE BACKS.........but, until I had carried out the detailed "work" checking out how the system fared on rating races, I just did not fully appreciate how "good" the ratings were. Few would recognise just how "good" the ratings were just looking at them and results alone.


So you gotta do the work to progress with ratings......

Gl with bets.

1998 = 2008

Ratings: I turned Them upside down, inside out - Found very little. luckily i Found other, simple ways to make a profit. I Went down the systematic route.

I asked myself; are my ratings unique? I answered emphatically NO, every Tom **** and Harry must be also factoring in my 10 factors. So i moved on, but learnt a lot.

the astonishment at how much data contains errors. Without exception every single day I find errors in data. 


sounds like many of you fellas have your own databases....i'm not technically-minded enough to have done that...did you find the commercially-available query databases eg Raceform Interactive, FlatStats, Horseracebase etc to be too limiting for your purposes?

I created a system of filters that used three racing sources, one being Adrian Massey's site. 

Most of the time, the data would 'align' itself but I had a 'flag' that went up if something didn't fall within a tolerance.  At that point, I would do the research and there were a number of times a reputable site was offering up 'mis-information' that I would correct for my database.

I also have my own ratings that will be entering into its third flat season of use.  I've trimmed it down to 10 minutes per meeting to load in the data by outsourcing to a student in Nepal who does the screen scraping for me.  Results handled as well then a matter of cut and paste into the data base.

This frees up a lot of time to analyze the figures that come up.

A note for those playing horses:  racing has changed since I moved here in 2000.  If you are creating data, I find a good window now is one year of racing by a horse.  Less than that, you don't get enough of a view on different goings.  More than that, it just seems to repeat itself.

- - -

For football and tennis, you can build criteria from results but, again, your own ratings are needed.

I'm really proud of the algorithms I built for Tennis.  Not only does it seed the players but also offers up how well players are doing through a tournament offering up three or four plays a week.

Football is okay.

A couple of points to share on football;

- criteria you build for one league in the world will not necessarily work on another.  In my opinion, it comes down to the quality of play.

- Look at your ratings from a different angle.  One league I was using ratings to find 1X2 results offered no value BUT when I divided the home teams rating by the visiting team, if it fell between two figures, the match would be UNDER 2.5 enough times to offer an ROI of around 30% on the season.

Duncan,
It has been my observation that because their sole income does not rely on the integrity of the data, nearly all providers have some errors in the data, otherwise it would be very expensive.
My sole income relies on data integrity, so I devote a major part of my time to it. They tend to mass gather information and then pass it on, nobody sits there with flags and errorchecking formulae datamining every part of the data.
It's a sad reflection that the worst data integrity comes from Betfair.
In 12 months of historical data, there were over 10,000 individual but blatant errors. When the server has any hiccup whatsoever, it seems large chunks of data is missing, but not just missing, wrong data is generated into the files.
Late withdrawals are still in there, runners duplicated with different prices, odds that never existed etc etc etc.

Betfair / Timeform do seem to be particularly bad at historical data integrity, especially the in-play figures which are out by up to 50% multiple times per day - e.g. you successfully back a horse at 12 in-play but Timeform shows the in-play high as 8 !!!

This obviously results in various betting systems being profitable on paper when in fact they would not have been in the real world, so be careful.

BTOR - you should be pleased that there are errors in the data sources, it creates a barrier to entry to stop everyone else following the same strategy and eating up your edge.

All of my trading is systematic, not based on any human decisions. I would say that the most important thing about the statistics is to know the limitations of the model and the level of uncertainty in your estimates.

There are two ways I would recommend using statistics that are not mentioned above: firstly in analysing your own profit/loss track record and getting some objective measure of performance.

Secondly, instead of using statistics to predict the outcome of an event they can be used to predict relative prices for different markets in the same event, e.g. win vs place, or draw vs 0-0.

The only way to make consistent profits in racing other than using stable information or trading is to be as maverick as possible in estimating the chances of individual runners in any race.

Those that follow hot trainers and jockeys or official ratings will see that the prices will always reflect any increase in strike rate obtainable so the trick is to jump to the front of the queue and beat the handicapper, bookmakers and betting public to finding hidden value!

lets take a recent race at leopardstown that occurred a few weeks prior as an example.

2:10 »  Leopardstown (IRE)
SUN 12 FEB 2012 

Deloitte Novice Hurdle (Grade 1) (5yo+) Winner €52,000 RTE1
Yielding To Soft 2m2f Number of runners: 7

In this race we have quite a few good horses in the mix and many have been running in
grade 1 races upto £44,000 and have official ratings up to 131!

Now the favorite looks a dandy win recent form results of 2-113 Sous Les Cieux
and looks like it could well win the race without taking a second look at the form!

But this is exactly as little work as mr bookmaker stroke form compiler ever puts into anything and we must beat him somehow, but how?

Well lets think about the race for a minutes, its not a normal 2 mile race for a start its
2 mile 2 furlong, now the extra 2 furlong may not seem like much but whats the going, well it was heavy that day so that extra 2 furlong is going to feel like much further and most of these horses have been running at 16 furlong in there recent outings, the favorite has run at 20 furlongs in a 5K race at fairyhouse and beat a field of 13 by 6 1/2 lengths in soft going so this must put it high in the staying stakes for this race, you would agree.....

But hang on a quick look through the card and we find that there is only one other horse that has attempted this trip and been succesful, 20 furlong that is and that is Benefficient


This horse ran in a similar type event 20 furlong, was rode by the same jockey as the favourite
ruby walsh and won a 20f 10 runner race in heavy going by 8 length, well on that race comparison
this puts this horse a few lengths better than the favourite, but wait Benefficient
is over 180 on betfair and 50/1 in the offices so surely it cant win because odds compilers know everything.

Well its up to you how much effort you want to put into making your own ratings but rest assured
if you put the work in you can make big rewards!

G.L. all

it is quite common for trainers to run horses at longer/shorter trips to either build up speed or stamina and sometimes to lower or even hide the clear rating of the horse so as to keep it on a handy weight, if you add this data with trainers placing the runners next into weaker or stronger races you can bet a job is on!

Sometimes it pays to think like a trainer instead of a dyed in the wool punter!

Think like a trainer. That's good advice for inexperienced punters. In many low-level races you can tell that several of the trainers are doing little more than going through the motions with their horse. You need to try to work out what the trainer's plans are, and whether that horse is 'intended' to win that day.

I agree with the "think like a trainer".  It hilights some of the points trying to be made on this thread that the 'facts' are just one aspect of finding a play.

Also, last year I started looking at how many horses were going to meetings.  The cost of petrol made meant never heading down to the track with an empty slot in the box.  From there, a matter of looking at the horses and you can usually put a line through at least one that, as mentioned, is there for the run.

Hmmm... Yeah, that's an interesting way of thinking about it. Get some idea of how many horses a particular trainer will be able to take down, and then look to see which of their runners appear to be the ones 'filling up the empty slots'. That could work particularly well in the case of big name trainers whose horses all become 'fancied' in the eyes of the betting public.