Smartgambler
Pro-Punter

Go Back   OZmium Sports Betting and Horse Racing Forums > Public Forums > General Topics
User Name
Password
Register FAQ Search Today's Posts Mark all topics as read

To advertise on these
forums, e-mail us.

Reply
 
Thread Tools Search this Thread Display Modes
  #11  
Old 18th May 2016, 08:44 PM
UselessBettor UselessBettor is offline
Member
 
Join Date: Sep 2011
Posts: 3,097
Default

Quote:
Originally Posted by Mark
And after all that, in 99% of cases you'll end up with the favourites in market order. So what's the point?


Agreed. I still think final odds are the result of a swarm. A very large swarm which is much more accurate then a smaller swarm.
Reply With Quote
  #12  
Old 19th May 2016, 01:28 PM
walkermac walkermac is offline
Member
 
Join Date: Nov 2013
Posts: 901
Default

Quote:
Originally Posted by Mark
And after all that, in 99% of cases you'll end up with the favourites in market order. So what's the point?
What's the point of knowing the market order a week before everyone else? (as was the case with the Kentucky Derby example). You'd bet the swarm favourite before its odds shortened and became the tote favourite, I guess. If you were making a book you'd take on risk with regard to what the betting public will resolve for themselves days later.


Quote:
Originally Posted by UselessBettor
Agreed. I still think final odds are the result of a swarm. A very large swarm which is much more accurate then a smaller swarm.
If you can remove your money from the pool and re-bet, then it's a swarm. You cannot so it is not.

I understand you can cancel unmatched bets in Betfair, so that's slightly closer to a swarm (but still isn't one). For markets with plenty of liquidity, is Betfair more accurate regarding a horse's actual chances than the tote SP?

Would you consider it likely more accurate if the following were allowed to occur: Bob's bet is matched at $3, the market eventually moves to a new price of $4 for that runner, Bob cancels his initial matched bet and proposes instead a new bet at odds of $3.80 - Bob improves on the odds that he was initially getting; a new layer risks a smaller payout on Bob's bet than if they were risking the same betsize at market odds). Every participant has this ability and the market remains open until all parties agree that they are happy with the risk of their standing bets. Surely it would be more accurate...it's how the sharemarket/currency trading/etc works after all, only with those markets we don't actually know the end result (is the company really worth $10mill?, is the Aussie dollar actually worth $0.75?).

Think back to the cow weight example: were it a betting market the poor sap who went first and guessed that a cow weighs 2000lbs gets to watch everyone else have their turn and see how very different from average - the best known long-term predictor in this market type - the guess was, and can do nothing about it. In a swarm they can think 'well, I still think a cow is heavy, but maybe not *that* heavy' and change their answer to one that's closer to being correct.

There are also several examples in this thread which demonstrate a small swarm performing better than larger groups not working collaboratively. If a swarm beats all but 4 of 469 competitors actively trying to individually do better, (the Superbowl example), does it sound like it's only picking favourites?

Alternatively you can read the conclusion of this paper that determined, under the area of study, that "large crowds make worse predictions than smaller, more expert crowds". (http://www.dangoldstein.com/papers/...owds_ec2014.pdf)
Reply With Quote
  #13  
Old 19th May 2016, 02:03 PM
walkermac walkermac is offline
Member
 
Join Date: Nov 2013
Posts: 901
Default

Debating myself, here's some interesting points from an article at: http://phenomena.nationalgeographic...sdom-of-crowds/

Quote:
So what happens to the wisdom of the crowd when the crowd talks to one another?

Andrew King from the Royal Veterinary College found that it falls apart, but only in certain circumstances. At his university open day, he asked 82 people to guess the number of sweets in a jar. If they made their guesses without any extra information, the wisdom of the crowd prevailed. The crowd’s median guess was 751.* The actual number of sweets was… 752.

This collective accuracy collapsed if King told different groups of volunteers about what their peers had guessed. If they knew about the previous guess, a random earlier guess or the average of all the earlier guesses, they overestimated the number of sweets. Their median guesses ranged from 882 to 1109. King likens this effect to real-world situations where people collectively drive the prices of items above their value and create economic bubbles. It’s what happened to create the recent US/British housing market crash or, more historically, the tulip mania of 17th century Holland.

Jan Lorenz recently found the same thing. Swiss college students can form a wise crowd when answering questions independently, but once they could find out what their peers had guessed, their answers became more inaccurate. In his summary of the study, Jonah Lehrer wrote, “The range of guesses dramatically narrowed; people were mindlessly imitating each other. Instead of canceling out their errors, they ended up magnifying their biases, which is why each round led to worse guesses.”

Is the crowd doomed to groupthink? Not quite. King found that he could steer them back towards a wiser guess by giving them the current best guess. When this happened, the median returned to a respectable 795. So the crowd loses its wisdom when it gets random pieces of information about what its members think, but it regains its wisdom if it finds out what the most successful individual said.

King says that this mirrors what happens in real life. The crowd may be a social beast, but it isn’t an indiscriminate one. Certain individuals wield disproportionate influence, and groups of soldiers, employees, players and even animals often rely on leaders when they make decisions.

There’s a reason for this. When King provided his volunteers with the best previous guess, their range of answers was narrower with fewer extreme predictions. Their collective answers were also about as accurate in small groups of 10 people as they were in larger ones of 70. King writes, “Copying successful individuals can enable accuracy at both the individual and group level, even at small group sizes.”
Reply With Quote
  #14  
Old 19th May 2016, 02:08 PM
blackdog1 blackdog1 is offline
Member
 
Join Date: Aug 2013
Posts: 427
Default

Quote:
Originally Posted by walkermac
"large crowds make worse predictions than smaller, more expert crowds"
I don't think anyone would dispute that observation!

One Einstein v a million illiterates jumps to mind.
Reply With Quote
  #15  
Old 19th May 2016, 03:12 PM
walkermac walkermac is offline
Member
 
Join Date: Nov 2013
Posts: 901
Default

"If they knew about the previous guess, a random earlier guess or the average of all the earlier guesses, they overestimated the number of sweets"

On reading further, the guesses form more of a log-normal distribution (the majority of estimates are low and a minority of estimates are scattered in a fat right tail; like the cow data). The average that King presumably passed on to his students to inform their later guesses was found to be a TERRIBLE estimate on distributions of these type in the Swiss testing. (I say presumably, as arithmetic mean is usually meant when talking about mean/average and I couldn't find him saying otherwise).

(From http://www.pnas.org/content/108/22/....expansion.html)
1. Population density of Switzerland
Actual density = 184
Arithmetic mean of data = 2,644 (+1,337.2%)
Geometric mean of data = 132 (−28.1%)
Median of data = 130 (−29.3%)

In every case of the Swiss data the arithmetic mean was WAY above the actual answer, so it's natural that King's participants would bump their estimates towards it and away from the true answer. If they were given the geometric mean, or median, they would've been sweet.

(From wikipedia: the geometric mean is a type of mean or average, which indicates the central tendency or typical value of a set of numbers by using the product of their values (as opposed to the arithmetic mean which uses their sum) ).

I've read Lorenz's paper here: http://www.pnas.org/content/108/22/9020.full#sec-1. ...but I obviously don't understand it as, to me, the pretty pictures show that the more information about the other participants' guesses participants have, the lower the error and the quicker the mean (both of geometric and arithmetic as the disribution presumably changes) converges on the correct answer, naturally losing diversity in opinion on the way *shrug*



(to briefly explain the image: the 1, 2, 3, 4, 5 refers to the times that a participant guessed the answer. Leftmost: no info - just guess again, middle: here's the terrible arithmetic mean - now guess again, rightmost: have what everybody has guessed so far in this picture format - now guess again).
Reply With Quote
  #16  
Old 19th May 2016, 11:06 PM
blackdog1 blackdog1 is offline
Member
 
Join Date: Aug 2013
Posts: 427
Default

I think this sort of thing is that keeps academics in bread and butter.
Google will take you to millions of sites like this.

In the end, if the participants know nothing about horse racing, how do you expect an intelligent outcome?

And that goes for any range of human activity GIGO.
Reply With Quote
  #17  
Old 19th May 2016, 11:22 PM
blackdog1 blackdog1 is offline
Member
 
Join Date: Aug 2013
Posts: 427
Default

Quote:
Originally Posted by blackdog1
I think this sort of thing is that keeps academics in bread and butter.
Google will take you to millions of sites like this.

In the end, if the participants know nothing about horse racing, how do you expect an intelligent outcome?

And that goes for any range of human activity GIGO.
Was too late to edit.
I was going to add that if the participants know nothing about horse racing and you have to guide them with external input then where is the 'swarm' effect? after all they follow your instructions and after that each others' input.

Sorry don't buy it unless it comes down to e x p erts in the field are involved in the swarm in which case we are back to crowd opinions AKA polling.
Reply With Quote
  #18  
Old 21st May 2016, 12:59 AM
walkermac walkermac is offline
Member
 
Join Date: Nov 2013
Posts: 901
Default

Quote:
Originally Posted by blackdog1
In the end, if the participants know nothing about horse racing, how do you expect an intelligent outcome?
The estimates that the subjects were asked for in the Swiss study:

1. Population density of Switzerland
2. Border length, Switzerland/Italy
3. New immigrants to Zurich
4. Murders, 2006, Switzerland
5. Rapes, 2006, Switzerland
6. Assaults, 2006, Switzerland

Unless you were an avid reader of past censuses and atlases with a photographic memory, you would have little confidence in knowing the answers to these questions. But the study participants, particularly when receiving information regarding all the past estimates of their fellows (i.e. swarming), got remarkably close.

From the Oscars example from earlier on, none of the 7 swarm members had seen any of the nominees for best documentary and best foreign film, yet picked the winners in both categories. They were able to pool their little knowledge and experience to formulate an opinion, likely each internally considering questions like:
- is it a film I've even heard of? (given the obscurity of most nominees, they could presume if a film's name has become part of the cultural zeitgist it must be in with a shot)
- does it sound like an Oscar-worthy subject?
- do I recognise any of the people involved in the film?
If one member didn't have a strong opinion then the group consensus would be influenced more by the members who did.

As far as horse racing goes, there's nothing magical about it that makes it different. No-one is suggesting that an inexpert swarm would be right all the time, but they would be able to make some decisions (particularly in bigger races):
- have I heard of this horse?
- have I heard of the jockey?
- have I heard of the trainer?

Under those circumstances they likely would steer towards the favourite. But as the research also shows, small, more expert crowds - using the same tools - can make better decisions. Better than larger, inexpert crowds. And often better than they do individually.

The Kentucky Derby swarm consisted of people who considered themselves knowledgeable. As a group they picked the first four. None of them did individually. Yes, the four were in market order at post. But they were not in market order a week earlier when they made their selection.


Are the most successful gamblers in horse racing toiling away in isolation? To my understanding: no. They're working in groups. Zeljko Ranogajec employs somewhere between 30 and a 100 people, per a newspaper article.

A swarm is just another proven means for a group to collaborate together effectively.


...I think people are getting confused that I'm suggesting you can usher people in off the street, give them a list of names and get them to pick the winner in the 1st at Burrandowan...

The Kentucky Derby swarm were knowledgeable players.

The Swiss students lived in Switzerland, had seen maps, travelled on the trains, driven on the roads, lived in the same society that reported crimes in the media, read articles about one immigration crises or another, etc. They'd never heard the figures they were being asked to estimate, but they lived there and could make a guess. I would suspect they would have done far worse had they been queried about Azerbaijan.

The people who guessed the weight of a cow had seen cows, eaten portions of them, knew how much they themselves weighed, how much their car weighed, could determine comparitive size differences between objects, etc.

The Oscars swarm had seen films, (some perhaps even up for nomination), knew actors, existed in a society where the industry was reported on, and so on. If they were asked who was going to win at the annual Nigerian Film Awards Ceremony (Nollywood) instead, then good luck...

This is as unsurprising as Zeljko's 30-100 staff members being clever and skilled at their jobs.
Reply With Quote
  #19  
Old 21st May 2016, 05:19 PM
blackdog1 blackdog1 is offline
Member
 
Join Date: Aug 2013
Posts: 427
Default

The Kentucky Derby swarm were knowledgeable players.

Look I don't want to be contrary about this idea, not going to influence me or my punting one way or other it's simply the principle of the thing.

If what you said there is correct, spec. about that punter with a huge staff, then why isn't what Mark and UB said also correct?
Reply With Quote
  #20  
Old 22nd May 2016, 12:01 AM
walkermac walkermac is offline
Member
 
Join Date: Nov 2013
Posts: 901
Default

Quote:
Originally Posted by blackdog1
If what you said there is correct, spec. about that punter with a huge staff, then why isn't what Mark and UB said also correct?
An ex-ATO official on Lateline claimed that Zeljko actually has/had 300 staff, and started 3 associated companies to service his business, which was turning over an estimated 2.4 billion dollars each year.

I don't understand the link you're drawing between the above and Mark and UB's opinion on using swarm intelligence in picking horse races. I mentioned it simply to note that the largest and most successful punters understand that there are advantages to working in groups. They aren't mad geniuses doing it all on their own. Don Scott was a hobby punter compared to these guys, but imagine how much bigger he could've been if 30 people were helping him with the gruntwork of watching races and trackwork, finding the best prices, etc. People can collaborate with effort, why so surprising that they can collaborate with knowledge? Swarm intelligence is just an effective means for it to occur.


What Mark and UB said/agreed upon: you're going to get favourites 99% of the time

What I said then: the only documented case thus far of someone using a swarm in an effort to pick the first four finishers in a horse race, did not pick the runners in the order that the market had them in at the time (a week later, the market eventually agreed with the swarm).

Nobody knows what would happen in a larger sample of races as, if anyone has ever investigated it, they haven't reported the results anywhere.

In swarm studies from other realms however, the group gets a 'brain boost' from collaborating together, performing better than those working separately and combining their results. The latter is what determines the market favourite. Whether the former means that groups select a winner more often, or can select the favourite faster than the market does (or neither of these things) is uncertain.
Reply With Quote
Reply


Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is On
Forum Jump



All times are GMT +10. The time now is 01:28 AM.


Powered by: vBulletin Version 3.0.3
Copyright ©2000 - 2018, Jelsoft Enterprises Ltd.
©2008 OZmium Pty. Ltd. All rights reserved . ACN 091184655