- This topic has 55 replies, 21 voices, and was last updated 20 years, 7 months ago by
.
- Posts
Some very interesting posts on what is an important topic for any discerning punter.
I think that perceived top Trainers are probably the most overbet factor with examples almost every day. You can easily check this by looking at the profitability of backing leading trainers over the course of a season. During the last jump season, Pipe and Nicholls were locked in a battle for the trainer’s title, but despite both having decent strike rates(Pipe 17%, Nicholls 22%), they lost £224.47(Pipe) and £86.43(Nicholls) to a £1 level stake on each runner. I think that if you looked at the top three trainers under both codes for the last twenty years you wouldn’t find many that produced a level stake profit betting on every runner.
Most people do look a little deeper into a trainer’s record and try to seek out stats that are profitable such as ‘trainer X at course Y’, but these historical facts are often based on quite a small numbers of runners and are likely to be unreliable as a guide to the future.
Top jockeys are also overbet, but surprisingly less so than trainers. Following the top three jockeys under both codes every season will inevitably lead to significant losses. I believe these losses will be less than those sustained following trainers because the jockeys can pick and choose their mounts whereas trainers have to give everything a run.
Jockeys, the less fashionable the better, are often underbet when riding horses they have ridden before.<br>I usually look at the last three races for such information but I’m sure I would do better if I had the time to look further back.
As Cormack rightly said, the most difficult part of this game is assigning a price to a horse’s chance in a race.<br>I believe that assessing factors can only take you so far in this process and that intuition plays a major part. I’m rather sceptical of any systematic approach that can churn out a recommended price for a horse in a race, even Wallace’s 50 variables(although I do admire this type of approach to winner finding).
Artemis, valid points anbout jockey and trainer stats. the headline stats are over used by lazy people and there is some useful data if you dig a bit deeper.
For years, I was very sceptical of any systematic approach until I discovered some of the work of Peter May and others. Having read most of Peter’s books I started about developing my approach. There is no way I could go back to traditional methods. As Alan Potts suggests, don’t follow the crowd.
The raw output from my database is a good guide to assessing probable odds/chances and having this early in the afternnon gives me a head start in forming opinions for the next days racing.
Artemis
Fair point regarding trainers although im not totally sure that following top stables is such a bad idea. There will almost certainly be losses posted over a season (although I am sure I recall Stoute, Charlton and Godolphin being in profit once or twice) if betting blind on their runners simply because thats the way the market is framed over a period of time. In fact £87 down for Nichols on 709 runners equates to a 12% deficit. Not bad really…
I was looking at the very middling Mark Tompkins stats last week and his % loss was around 30% and there are plenty of trainers down there with catastrophic figures
I think though that trainers can be habitually seasonal and sometimes the month to month trends are interesting. More significant still are the first time out and debutant trends too. They are often very reliable
Wallace,
It would be interesting to add the opinion of the market to your other variables and see if that made any difference to profitability. The point of doing this is to factor in to the model some of the things that are really unquantifiable and only manifest themselves in bets made by those with inside knowledge. The great practical problem is that the market usually doesn’t reflect the inside knowledge until close to the off except in high profile races where the market is quite strong.
There is a precedent for the above described in Mordin’s book, "Winning Without Thinking". He describes the activities of the ‘Hong Kong’ syndicate who used a complex betting model on the pari mutuel. I believe the model had 22 variables and used multiple regression analysis to arrive at selections. Despite its complexity, the model couldn’t achieve profitability until the market forces were factored in.
I know this is following the money to some extent, but the strength of the ‘market correction factor’ may only be small in relation to the other variables.
Over-rated:
Most of the stats such as, ‘in the last 20 years 6 favourites have won this race.’ I think that each race should be taken on its individual merit.
The form of jockeys also strikes me as being over rated. It is the horse that we should be concerned with.
Under-rated
One of the most important things I try to assess when analysing a race is how the race is going to ‘pan out.'<br>So many times you see a race where the main contenders need to come from off the pace but there are only a couple of front runners. You then get a falsely run race and the result can be misleading for future reference.<br>
Artemis, you are correct, adding market data into the model is currently being tested.
We are capturing exchange data at three time slots prior to the off using the Betfair API. One problem we have seen with this is the distribution of money matched on individual horses on Betfair is far from what you would expect. Its quite common to see a 3/1 favourite with around 70% of the total matched. This is due to the large proportion of arbing and trading done at the front-end of the market.
There is still scope for good old fashioned judgement when setting odds and as I have been compiling tissue odds for years, its hard to give this up!
With good opportinities for trading on BF I have moved more to trying to identify market movers and getting a bet-to-nothing. The 3:45 at Brighton tomorrow looks a good example of this; Goldhill Prince 12/1 in the RP 14.0 on BF for small change should provide a good trading edge.
One of the most important features (for me) is the going. Yet this appears to be one on the hardest things to find out easily – good, accurate going descriptions. <br>Keith
Where? I really would like to know a place where I can easily visit and find a proper, informative descripton of the going.<br>Keith
Wallace,
The likely market movers are often easy to spot but there’s not enough liquidity in the market in early exchanges to make it worthwhile for me. It would be a full-time job monitoring the market.
I wish you the very best of good fortune with your methods, although using your approach, I don’t think you will need it.
I’m off on holiday for a week so I’m unable to add any more to this very interesting thread.
Artemis, yes poor liquidity is restricting opportunities and things seem to be getting worse. I tried to get at total of £800 on Goldhill Prince at rates from 16.0 to 24.0 overnight and ended up with £292 at average odds of 19.7. Still a good trade but not for much in this global market. Interesting the well known service published ratings at 9:30am and this horse was also top rated with them
Going Assessment; EC, I do a very similar thing to you in arriving at my own estimation of the going allowance for every meeting. We have built a module on our system which automatically calls up the relevant course/track/distance data with the forecast going and the actual GA based on the past meetings. A few minutes work watching the first race every day does pay off.
I have considered setting up a network of people to cover all courses in the country to take going measurements using a going stick and also to measure actual race distances when it’s obvious the running rails have been moved. Relaying this back to a web site. This may still happen but I wont be spamming the service on here!<br>
Don’t turftrax do the going stuff already?
Where did you get your £292 on Wallace?
EC
It’s strange that you’re trying to be smug.
After all, a week ago, I gave you the time and conditions of a race and all you could come up with by way of going was excuses.
So much for "takes about 5 minutes" and "You can even do it with 14f races"
The race was Musselburgh, 14f, all age 0-70 hcap. 3m7.82s.
You did have a crack at a 5 maiden at the same course but your opinion differed from the reported opinions of the jockeys.
Personally, I’m tempted to believe the jockeys were right about the ground.
Their opinions were more consistent with the weather prior to the meeting and your calculations totally ignored the strong wind.
Nor have you, in the last 5 days, mustered an answer to "do any of the speed figure advocates seriously suggest they know the distance of any of the races at Chester today [last Friday]?"
Unless they were 30+ lengths apart, I can say with 100% certainty how far one horse finished behind the horse in front.
You can’t even say, within 3 lengths how far horses were racing at Chester last Friday.
So, feel free to feel smug all you like, but I’m going to stick with my belief that, with inaccurate data, you can’t get accurate figures.
Steve
EC
I think if you read back it was Empty wallet that had a go at it..not me.
You’re right. My mistake. You never managed to have a crack at that one either.
I also said I didn’t make speed figures at 14f.
And so much for "You can even do it with 14f races".
If you want to test me again then throw up a race time..but I know you will deleberately pick a very slow run race
No s**t
, Einstein.So, slow run races make you guess the going wrongly.
I bet races run over further than the advertised distance (like last Friday) have the same effect.
Hmmm.
Wasn’t that the point I was making?
Steve
EC,
I am with you all the way on this and have been doing same for over 20 years. The going correction calculated contains a measure of correction for wind speed and errors in the average distance the race is run at. You are only trying to establish whether the going forecast is about right or way off. The more way off the better. Jockeys are not reliable pundits on going or pace; going sticks and Turftrax are not reliable (the latter’s circular opinions I was told by them are based on clerk of course opinions) – time and race observation are very reliable.
Robert
Robert or any Speedfreak :biggrin:
Hypothetical
Say if you had past race times between 73.0 and 77.5 (sample size say 12) on Good to Soft
would using  Standard Deviation of those racetimes give a more accurate Going assessment ?
(Edited by empty wallet at 12:02 am on May 12, 2005)
EC
Points scoring?
You’re the one who brought my name up on this thread.
Despite the fact you couldn’t answer two simple questions on that previous thread.
And, again, failed to address the questions here.
You want to use a real life example?
Tell us how you worked out the going adjustment after the first at Chester last Friday.
You can call it point scoring all you like.
I just call it the reason I’ve never thought that UK turf speed figures are worth taking the time to calculate – unless you do all the times and the distances yourself, the data’s unreliable.
I would have thought your error with Kyllachy would have shown you that.
It’s interesting that Robert made a point of mentioning "times and race observation".
I’ve been saying all along that, without watching the race and applying common sense, the times don’t tell you much.
Steve
- You must be logged in to reply to this topic.