Debunking Parrando's Paradox
 

While there are a few intelligent, well-meaning and worth-reading contributors at some forums the lunatics inevitably seem to destroy the debate with their spamming of path to ruin systems.

Amid the current incessant promotion of loss chasing, negative expectation insanity is Parrondo's Paradox.

So here is my personally conceived demonstration of why it is a sham.

Imagine a class of games where you have to traverse a disparate region.

Game A: Use a bicycle

Game B: Use a canoe

Unfortunately the region is so bad that you will inevitably bog down using just one conveyance. So neither game can be won.

But if you wave a wand to magically randomly transform the vehicle into EITHER a bicycle or canoe you will eventually win.

This is the flaw in Parrondo's Paradox.

You are NOT combining 2 negative Games A and B to produce a positive Game.

You are actually combining 3 games A, B and M - where M is the magic wand capable of morphing between A and B.

M is clearly a positive game and that contradicts the claim of turning negatives into positives.

There is no magic wand in racing.

There is winning and losing, there is no third element except breaking even, but that adds no element to the end result.

If M could only talk. The tales it would tell.

While we wait for the Parrondo Proponents to rear their delusional heads, you might find this link (spotted elsewhere) interesting:

http://www.eleceng.adelaide.edu.au/...er/games/ma.htm

Presumably it is trying to describe the underlying principles of the paradox.

I've been working through it and it has more than its fair share of holes. Unfortunate since formulae with typos defeat the purpose.

But you don't need to do too much work to note the obvious flaw in the hoax.

Game B is supposed to be negative. And it plays out according to its own current capital.

But the moment you combine it with Game A you alter Game B and its probabilities. Because you introduce random external variations to its capital thus altering the way it plays out.

Yet the author has no qualms about squandering taxpayers' money writing papers with deceptive titles like:

"Losing strategies can win by Parrondo's paradox"

What about the theory 'Ockham's Razor' [for those that don't know what that is, goggle it] . could that be applied to horse Racing in some way ?

Nearly everybody knows the principle but probably not by the mathematical name Occam's Razor.

In essence it is "when there are a number of different correct solutions to a problem choose the simplest".

The best racing applications that I can come up with right now are things like barrier tables.

Nearly every public analysis I've seen (including Don Scott) produces complicated tables which turn out to be wrong. That's because they only count from one side.

Barrier 6 might sometimes be an inner gate, and at other times the outer.

If instead you counted from the edges using -1, -2, -3 for the widest, 2nd widest etc. then you would spot important patterns that the others miss.

In particular you might spot cases where the widest gate is actually an advantage, and sometimes wide gates in general.

That info would simply be buried in the original more complicated tables.

The same technique also bears fruit for order of favouritism. The 6th favourite might be well in the money or actually the bolter.


Occam's Razor is actually used in real life for pattern recognition as illustrated here:


http://cgm.cs.mcgill.ca/~soss/cs644...attern-rec.html

Not sure that I agree with your sentiments jfc. I personally know of one person who claims to have had some success with Parrondo's Paradox & know of another who did a fair bit of work on the issue, claiming that used correctly it was possible to turn two negatives into positive.

There is a certain forum where what has been written by both can be located but I dare say if I mentioned that forum here the thought police would quickely have it removed. Pity about that, as this forum has lost a lot due to those damned thought police, but nevertheless, it is Propun's loss not mine.

But if you know who Chiron or Peter Jamieson are then perhaps you can work out which forum it is.

It's quite simple jfc, they didn't account for the number of starters in a race.

But with 6 barriers over 2000m does anyone seriously think barrier 1 is any better than barrier 6?

The further the distance, the less the barrier impact.
The shorter the distance the more the speed impact.

Barrier positions are extremely overrated and exaggerated.

It's what happens or what a horse can do in the straight that matters, the early part is up to the jockey, not the barrier.

Barriers have very little to do with winning chances in fact.

Just my findings from a research I did two years ago.

However, here is an instance of the "Paradox".....
(negative barrier - negative chance = overlay)
Horses in "unfavoured" barriers mean a higher average price, therefore profit can be made, as horses overcome this obstacle more often than the price they are.

Horses in public favoured barriers are given too much of a discount by the public, so are great lay opportunities.

While in agree,emt with your sentiments Chrome, not sure if it's a case of too much a discount for horses in favoured barriers or more a case of those drawn out wide ignored way beyond what their real chances thus creating pverlays.

Know of one professional punter (who you know about) that has profited from well over the last dozen years or so from horses drawn out wide.

To get any info worthwhile from barriers, one would have to delve a lot deeper into such things as the class of horse, the quality of jockey , as well as the horses running style just to name a few. Most average punters assume that if there is a large enough sample, that each barrier would get an equal distribution of runners of all styles, jockey capabilities, and favouritism. This may well be the case but I haven't seen any info that detailed as yet and I dont think I'm likely to. Like most other factors in racing, there would be millions of different combinations for each.

Hi whiteycat :-)

Anyone who states barrier position are not important knows very little about horses. Barrier position can be crucial to the running style of the horse. Put a good leader in the outside barrier and on most tracks at most distances and you can forget he is going around. Put a good back-marker in barrier 1 and you can forget he is going around. Put him in the outside barrier and you will see a different ball game.

Ice Chariot was a very good example of how easy it was for a good back-marker to win and overturn his crappy performances previously [inside barriers] to his 2 big group wins at the Brisbane carnival from barrier 17 in both events.

The problem for most punters is they wouldn't have a clue as to the running style of the horses in a race they are betting on and so say 'oh barrier positions aren't important or otherwise, inside barrier 9 is good'. Track layout, distance, Jockey ability AND barrier position are all vitally important info, to those who don't like to bet blind [or blinkered].

Find a race with a good horse who is a leader in an inside barrier with no other leaders in the race and you have a horse who controls the race and is money for jam. Find a good back-marker in an outsidish barrier in a race with more than 1 leader to set up a bit of a good pace and you also have money for jam.

I really should do an article here about horse's running styles and how important barriers are. Well overdue I think. Kingston Town was one of the very few horses who could be a leader, an on-pacer, a mid-pacer or a back-marker and win from any barrier taking up the running style the barrier suited.

Not too many Kingston Towns around nowadays so it's worth taking note of a horse's running style and what barrier it's in.

Hi Crash, just wondering if you could share some of your methods on establishing a horses preferred running style.

It really needs an article young buck. I tell you what though, this game is all about 'edge' and from anywhere there is one the public doesn't get their head around except for a few, it's worth understanding. I'm still a bit of a learner, but my punting outcomes have improved heaps since I sat up and took notice of understanding the horse itself, a creature of habit in more areas than we think and each one an 'edge'.

It's all about the edge. I do look at barriers and their winning strike rates at the particular distances, but it is really just another element of many to consider.

Believe me, it's the main one and it's taken me over 30yrs of punting to get around to really looking at it closely.

Crash, not sure if your first sentence was referring to me or not? If so I didnt say barriers wern't important , just that using raw strike rates from an individual barrier on its own wont do the job. If you were not referring to me then please disregard this post.:-)

La Mer,

I'm not sure if you've actually browsed the Markov chain link I posted, but I've now actually figured how to do the key calculations with a spreadsheet. Too much effort to post here, but if someone mathematically inclined genuinely wants to see it I'll try to eventually oblige.

Anyway it's more obvious than ever to me where the hoax is - Game B cannot be combined without irreperably altering its essence.

But it is inconceivable that the Parrondo model could be applied to racing.

Essentially it involves a Game B - a stair climb - where every (say) 3rd step is very unlucky.

But in racing if you know that certain races are very unlucky why don't you just skip them!

So I suspect your associate is deluding himself.

I'm sure I've vented how underwhelming I find arguments involving negative games and/or loss chasing.

But are other plans based on the notion that racing is different to roulette. That is true - there are definite long runs which defy conventional statistics. The trouble is you only find out about them AFTERWARDS. I don't know of anyone who succeeds in predicting runs early on. So these plans appear extremely suspect.

The method I use is to try and withstand even the worst possible run and always keep hanging in. Which means at some point I am forced to reduce. But a horrendous run usually wipes out a lot of competitors so if I survive I end up picking up far more juicy collects.

Barriers and other things
 

Having worked extensively in HK and being able to compile many studies of the racing product, one such study was on horses working out on their own or in pairs (time based) in the vast majority of cases these horses INHERENTLY have a "normal" probability distribution (which one would expect or be very likely) introduce these "normal" horses to real competition (Class Ladder) and he has a "Different" distribution, the difference in those two distributions is almost entirely due to the physical artifacts of the Class Ladder and some other racing conditions , and NOT due to the talent component of the Horse.

Further studies confirmed that Barriers / Racing style produced a very distinct range of Distributions .... so in the wash up of all this mumbo jumbo is that Barrier / Racing styles might not produce absolute differences but they do hold the key in being able to provide a very good "performance" distribution

I'll have to just let it slide then, but the jockey is 10 times more important than the barrier.

Let's say we have a leader in barrier 20.

A good jockey will not try to outrace them, he'll slot in midfield and let them settle, rev up to the lead once they settle and work his way to the lead, an inexperience jockey will hunt up and try to outrace them burning all chance.

There is no difference in expended energy in drawing barrier 1 and hunting up to hold off all challengers and drawing barrier 20 settling midfield and going to the lead when they settle....if the jockey is skilled enough.

A bad jockey will get pratted 4 and five wide from barrier 20, or in the first instance race a no hoper in barrier 2 just to hold the lead at all costs.

Nope. Was referring to the average punter generally. No one in particular Dr ron. Sorry if I gave [anyone] that impression.

Chrome, Even the Beadmens are often forced to 'ride for luck' when they get a bad barrier for the horses they are on [why back them in those situations?].
Where good Jockeys are important, is like you point out being able to 'rate' a race and take the breaks and cracks and turn them to advantage. Mid-pacers often need a good jock to stay in a handy position and sure, sometimes a good jock is 6 times more important than barriers and at other times 4kg claiming newbies can look like champions and bring home the bacon.

The best thing for punters who don't know the ins and outs of a horse's running style [check over their past races and where they were at the turn for a good hint as to their running style], is to only back horses with good jockeys on them. That's an edge thats better than nothing. For those punters, jockeys are 10 times more important than barriers because they really are betting for luck.

A Markov disrespect
 

Some may have noticed a request for a layman's explanation of that Parrondo paper. Which no one was prepared to attempt.

http://www.eleceng.adelaide.edu.au/...er/games/ma.htm

However I'll produce my effort here in the hope that a few might actually bother looking at the link, learn something and stop insulting others' intelligence.

Game B is supposed to be negative.

The win chances depend on your position:

0 10%
1 75%
2 75%

Now intuitively that looks positive.

If you assume the 3 slots are equally likely then your win expectation is

(10+75+75)/3% ~= 53.33%

But if you actually use a spreadsheet to repeatedly play the game you find your chances of being in a particular spot end up as:

0 38.46%
1 15.38%
2 46.15%

And that win expectation is precisely 50% = Break Even.

Note that you are spending far more than 1/3rd of your time in ZERO - the Slot of near-Death.

But if you alter Game B by merging it with a negative coin toss with only 49% win chance and repeat the spreadsheet operation then the chance of being on pesky ZERO reduces from 38.46% to 34.51%.

That in turn changes Game B to 52.57%.

Now the average of 49% and 52.57% is clearly better than 50%.

And if you are worried that your taxpayer-funded grants will be taken away because everybody else is bored by such pointless calculations which have absolutely no real-life use, then you title your effort with an eye-catching lie like:

"Losing strategies can win"

I'll take your word for your conclusions jfc. I've got one of those windows 'math' programs with all the maths symbols .....but I still don't know how to use it!