Confronting some basic assumptions

[Lenny]

Here's a question for all you clever analytical types out there. The common assumption is that a selection rated as a 33% chance will win one race in three, all conditions being equal in each (repeated) race.

My question is this: is this assumption really true? Isn't the percentages we assign to a selection just a way of us estimating a selections chance, rather than reflecting what will happen in reality? Surely the same winner, in equal circumstances, will win again? Thus isn't the reality of any race really - winners chance:100% all others: 0% - rather than the market we frame for the race? Isn't the frame a way of adjusting for the unknown variables we can't predict?

I ask this because it is having a bearing on how I consider the viability and practical analysis of systems in general. Remember, stats is just a way of representing many numbers with just one "meaningful" one.

I'd be very interested to hear your considered opinions.

~Lenny

[hermes]

You can only say that the winner had 100% chance of winning the race AFTER the race. Before the race it had 33% chance (or whatever). If the same race with the same runners in the same conditions was run again it would then have a 100% chance of winning because, if all things are equal, it will win again. Which can't happen.

It is true than in *reality* the winner was going to win so in that sense it was a 100% thing, but the "chance" we ascribe to it is a human contrivance for human convenience, i.e. the chance is relative to the human ability to predict that it was going to win. What happened was never not going to happen, but human beings give themselves only a 33% chance of predicting it. So its the human ability to predict, not the reality of the outcome, to which the percentage refers. Or have I misunderstood you Lenny?

Hermes

[becareful]

On 2004-01-07 23:58, Lenny wrote:

Surely the same winner, in equal circumstances, will win again? Thus isn't the reality of any race really - winners chance:100% all others: 0%

No! Even if you put the same horses in the same barriers with the same conditions (track, wind, etc) you will not get exactly the same result. Horses (and jockeys) are not machines so they will not repeat the same performance twice.

If I take you to an indoor pool on 10 different days but with the same water temperature, etc, and get you to swim 100 metres will your time be exactly the same on all 10 occassions? Of course it wont - sometimes you will stuff up the start, sometimes your technique will be off, maybe you will mess up the turn, one day you will put it all together just right.

Exactly the same applies to a each horse in a race (and the jockey on its back). The "best" horse (the one we rate the best chance) will not always win for any number of reasons (bad start, blocked for a run, bad choice by the jockey, the horse is not running up to its best on the day or simply another runner puts it all together perfectly) but if you ran the race again any one of these could be different and the horse may win.

[Lenny]

Hermes, I think you understood this situation perfectly. It is interesting that you ascribe the 33% chance not to the horse, but on the handicapper's ability to pick it! This is food for thought and I will indeed consider this idea further.

Becareful, well done! You picked up on a point I had hoped someone would have twigged onto. You said,

"...but if you ran the race again any one of these (conditions) could be different and the horse may win."

Rerunning becomes outside the parameters of this problem, as we want to consider the selections chances if the race was REPEATED, not rerun. In this case, I would say the winner is still likely to win again close to 100% of the time. The free will and choices of the contenders are likely to throw up different circumstances; but the chance of the winner winning again this time, in my opinion, is only a few percent off 100%. This, though, may vary depending on how clear his likely win is. The clearer the win, the closer to 100% of the time the rerunning winner will win again. Hard to get across, but a valid argument I think.

In short, there seems to be two bases (or philosophies) we all tend to bet by - finding the "definite" winner (direct approach) or finding the probable winner based on a statistical approach. I'm sure both approaches have their validity, but how they relate to each other is what interests me here.

Thanks guys,

~Lenny


<font size=-1>[ This Message was edited by: Lenny on 2004-01-11 02:54 ]</font>

[crash]

WOW !!! Run all that by me again in layman's language would you ?

Cheers.

[darkydog2002]

ECHO THAT SENTIMENT CRASH.

[becareful]

I think I see where you are going but I don't think there is a "lone winner" for a race. I believe that if you were able to repeat the race it is more than likely you would get a different winner.

For each race there are literally millions of different ways the race could be run, some result in horse A winning, some in horse B, some in Horse C, etc. When we are doing ratings we are trying to work out what the chance of a scenario where horse A winning occurs, the chance of a scenario where horse B winning occurs, etc.

If you were able to repeat the race you would get a different scenario as the chances of the race being run in EXACTLY the same way are millions to one - of course the different scenario may still give the same winner but that is where the probabilities come into it.

How many times have you watched a race and seen a runner lose the race because the jockey made the decision to try to go around the leaders but if he had waited a split second longer he would have seen the gap opening for an easy run through? Or a horse lose the race by missing the start? Or an outsider win because the leader went wide on the home bend taking 3 or 4 horses with it and giving the outsider the run through on the rail? If you repeated the race it is probable that these factors would change and therefore the race result would change.

[puntz]

<font size=-1>[ This Message was edited by: puntz on 2004-01-31 00:32 ]</font>

[Rock Steady]

That is precisely why I bet on greyhounds and give the horses and trots a miss. Dogs are VERY VERY consistent. They are creatures of habit - a wide runner will always be a wide runner. A slow beginner will begin slowly 9 times out of ten. A railer will always look for the rails.
Therefore, times are everything in greyhound racing. As I said on another thread, I have been averaging a POT of 15% on the dogs for years now - to do that on the horses would take me probably three times the work and even then, you have the dreaded "bad ride" to contend with.

[stebbo]

And of course, the dogs *NEVER* stop to scratch themselves either, do they????????

Or, they never get knocked for six by some mad railer in box 8 cutting them off...

Or, they never draw a middle box and get crunched from either side....

Or, they never get their legs knocked out from underneath them by some cat that runs up their backside....

Or.... there are just as many things that can go wrong in a dog race that there are in a thorougbred race, maybe even more.... And there's no protests either!

[Rock Steady]

Hi stebbo,
You are spot on!
All those things happen - that is precisely why I do the form. If there is a wide runner in Box 2 that begins well, it will set the race up for Box 1 providing Box 1 is a railer. If you bet on dogs without studying lots of video replays, you will get slaughtered.

[kenchar]

Hi Rocksteady,
Just one thing I'm curious about and I'm not trying to be smart as I've had enough of that from others here myself.

I have a friend who is very very good backing dishlickers but he ALWAYS goes to the track to get the weight, which basically restricts to local Sydney races.

He will NEVER back a dog that is up 1/2 a kilo or down 1 kilo on it's last run.

Do you bring this into your equations and if so how do you get the weight on the net.

Also he will NEVER back a small bitch from barrier 1 as will shy away from bigger dogs on her outside.

Cheers

<font size=-1>[ This Message was edited by: kenchar on 2004-01-10 19:58 ]</font>

<font size=-1>[ This Message was edited by: kenchar on 2004-01-10 19:59 ]</font>

[Lenny]

Hi guys,

Getting back on track here (not that I'm not interesting in the dogs - interesting punting option!). The purpose of the discussion was to clarify some mathematical assumptions (as this forum group is about systems). I believe what my ramblings boil down to is this: a horse that is super fit and bolts in in 6 lengths is likely to repeat the act in the SAME race, if it could be repeated (which it can't - hence the earlier conceptual difficulty). The less the estimated win length, the more the danger of inrunning influences.

After considering the likelyhood of a 6L winning bolter to repeat the performance to be near 100% and further considering what Hermes wrote earlier, I now think (my) handicapping is a matter looking for the likely percentage variation in the honest running of all horses in the race.

If I get a good estimate of the theoretical running of the field (by using for example Price Predictor Pro), then I can estimate what influence sll contenders may have in influencing the winners run by considering running style, jockey and barrier position.

Thus I guess I believe that getting a clear estimate of how a race will be run can be used to reduce or increase the likelyhood that a selection will perform to its rated best. Perhaps this is just repeating a simpler philosophy, but I like reinventing the wheel my way :wink:

~Lenny

[woof43]

Hi,
I can answer a few questions regarding greyhounds, as being the owner of possibly one of the largest form databases of greyhounds nationally
Fitness=Weight in most cases, If I find the avg racing weight of say every dog in my databes that have more than 50 career starts, the lighter racing weight half will be quicker than the heavy half, the frequency this occurs is quite remarkable.
So your friend is heading in the right direction, the only thing that will conflict with weight is when they are young 18 months etc, they will add weight as they grow an become more powerful(barring injury).

In regards to light weight bitches not favouring Box 1, well think of a heavy dog as a cement truck with the same acceleration, an the bitch as a sports coupe, so its a no brainer that on a two turn track with an 80 metre run from the boxes to the first turn, on avg a greyhound weighing less than 28 kilos will run faster than a greyhounds weighing over 33 kilos, but things do get a little more interesting on one turn tracks.



<font size=-1>[ This Message was edited by: woof43 on 2004-01-10 20:59 ]</font>

[kenchar]

Lenny,

I'm VERY sorry to get off the subject, but if you want to back a horse that won by 6 lengths at it's last start go ahead it will be $1.20 and if it wins who cares at those odds but will probably be knocked off by a bolter.

Happens too many times for my liking.

Cheers

[stebbo]

Hi Woof,

I'm interested in your form database for the dogs... Is this something you've compiled yourself or something you subscribe to?

As to "one turn" or "two turns"... how do you define a one turn track as opposed a two turn track?

Many moons ago I subscribed to the greyhoundform website, and developed what I thought was a nice little system on the dogs.. Part system, part form analysis, it had a strike rate of about 75% on paper.... and lost the first 15 bets I put real money on...

My neighbor and I punt on the dogs as relaxation... $2 and $3 bets and have a ball.... I wouldn't mind winning a little more frequently than we do tho!

Cheers,
Chris.

[puntz]

<font size=-1>[ This Message was edited by: puntz on 2004-01-31 00:36 ]</font>

[Lenny]

On 2004-01-10 22:27, kenchar wrote:

Lenny,

I'm VERY sorry to get off the subject



No worries, Kenchar. :smile:

To me (and many others I'm sure), there are three key notions to winning at the races. The right horse, the right race and the right price. We all know the right horse is not always the one that is in form and rated the highest, it can suffer due to its running style, the track shape, direction of running, jockey, BP, etc.

I'm trying to quantify what influence the actual race has on a top form runner, and find out to what extent in-race circumstances will affect its run. If I can convert the best and worse possible race for a selection into kilograms or lengths, then perhaps I have a way of handicapping that can bypass the multitude of variables that keeps guarenteed winners elusive. If the worst case rating of a selection is beyond the best case rating of all other runners, its a darned good bet.

~Lenny


<font size=-1>[ This Message was edited by: Lenny on 2004-01-11 04:33 ]</font>

[kenchar]

Lenny,
I really think becareful answered your question.

10 events run under the same conditions 10 times and you can get a different overall outcome each time.

Cheers

[crash]

Kenchar,

I feel what Lenny has been trying to say regarding repeatability is that old handicapping adage : 'Horses are creatures of habit, and repeat what they have done before'. If only it was true as they have a habit of suffering a bout of amnesia the moment your money goes on !!

Lenny,

I think what you are discovering [ 're-inventing' as you put it ] and trying to come to terms with above is basically, 'handicapping ' in general. If not, it sounds very much like it [?].

Regarding Greyhound weight as a handicapping tool. In the UK they weigh the horses for the same reason. It would be great to see that done here as performance peaks can then be matched to past preps.

Judging a Horses fitness used to be a very commonly acquired art form from years of plodding over to the ring to 'look at fitness'. As a kid, almost every Saturday used to be an off to the races day with my Grandfather. Nowadays with Sky Channel, TAB's and Pub or Puter punting which encourages stay at home punting, most Punters wouldn't know a winning fit horse from a hot dog so knowing a Horse's weight would be a boon to our handicapping efforts.

Cheers,


<font size=-1>[ This Message was edited by: crash on 2004-01-11 08:50 ]</font>

[Rock Steady]

Hate to be party pooper stebbo BUT.... anyone who seriously thinks they can win in the long run betting on dogs WITHOUT doing hours of videos and form study is kidding themselves. Maybe you would have avoided backing the dog that got into trouble if you knew what its racing style was.
Seriously, forget about "systems", put in the hard yards and long hours and you might be surprised how many value winners you back.

[pickle_punter]

Lenny,
I think what you are saying is that if a race were repeated (rather than rerun) with all the same options taken, the same abilities on the line and the same tactics employed...the same winner would eventuate.
Unfortunatley this is not the case. The reason lies within the horse itself. A horse can build to a peak performance. Most will then fall off this peak at the next run, even in identical circumstances. If however the horse can hold it's peak. It thereby provides a platform to kick from at it's next preperation. Few horse hold more than two peaks in succession. If they do, they will probably find a lot higher peak later in their career.
Maybe if you reran the race in identical circumsatnces with all runners at the same point in their preperations it would be a similar result but this senario can never happen. The runners will be older and a million other variables that can never be recreated will stuff things up.

[woof43]

Hi Stebbo,
The database is private, an not open for subscription ( i need an edge).
A one turn track is basically a horse shoe layout of the race distance. ie they start in the back straight an finish in the front straight, most one turn tracks have a 60 - 65 metre radius turn whereas two turn tracks are around the 50 metre radius

I think the first goal if your producing an Odds Line is to develop an Oddsline that is within a % point or two of the General Publics Odds Line.an is smooth in its transition thru the rankings, when tested

Then you produce a performance envelope for each runner, that is, what is it's Mean and its Standard Deviation, the STD is maybe one of the most important formulas in racing.This then allows you to produce an OddsLine.
Then you would be able to compare, both your Public Generated Line an your own to at least give you an idea, as too what sort of investmnent you may make on this race.

I personally use Monte carlo simulations, using a variable that will outperform the General Publics Line's best variable for that particular race.

[Lenny]

Hi Crash,
/
I think what you are discovering [ 're-inventing' as you put it ] and trying to come to terms with above is basically, 'handicapping ' in general. If not, it sounds very much like it [?].
/

Perhaps - systems are my thing (in regard to punting) and I've always tried to avoid actually looking at the form if possible (click and win is my ultimate goal). In this thread, I am just trying to quantify the variation that exists outside of h'capping's standard parameters, such as BP, fitness, etc. That way I can know that if my software throws up a rating of "40", it may actually mean "36-44", or whatever. In short, I am trying to quatify a horses rating variation, but further, I'm trying to quatify what degree of variation is due to understood variables, and what degree of variation is in the "lap of the gods".

It is interesting Woof mentions standard deviation in his ratings for the dogs. This is exactly my thinking for my ratings of the ponies.

Hi Pickle,

Thanks for your reply. I appreciate the insight.

~Lenny



<font size=-1>[ This Message was edited by: Lenny on 2004-01-13 04:44 ]</font>

[crash]

I think that my posts over the months have made people think I look down on systems. Nothing could be further from the truth as
I know there are System Pro's and Handicapping Pro's, both doing very nicely thank you !!
I just love to rubbish you system junkies [even though I do use them myself during the year ] because you get so defensive !!!

Cheers.

[puntz]

I think a system has to come to some sort of closure. One can go on and on, to polish the wheel, but then after the polishing,it may have a limit. I guess ya look for another wheel ?
But not having to re-invent.
Just use other options where it did not fit on the rim of the first wheel.
I think there was a post somewhere, where he/she uses 100 or so different systems.
( wonder if they own a tyre company!)

[Lenny]

Hi Puntz,

I think it was Bahgwan (apologies for any spelling errors) that used ~100 systems. For what its worth, my two best systems are extremely light on the rules, and this is quite unexpected to me.

~Lenny