Absolutely loving this thread. Learnt a huge amount from it.
ARAZI91 on the basis there is no such thing as a stupid question, can you humour me...
I assume the syndicates are betting in sufficient volume (did you say 40% of the volume in some football markets?) to move the odds around, whether that is on a public exchange or in a private Asian market or wherever. Therefore, are they:
a) either manipulating the market up and down (or at least taking advantage of market movements), making high volumes of tiny profits from arbitraging between a *lot* of back and lay bets, or
b) making money by being smarter than everyone else in the market (Joe Punter, owners, trainers etc.) and hoovering up the other 60% of the market volume by placing small numbers of very high $$$ bets & being correct more often than everyone else, or
c) doing something else?
The bit I don't really get is if they are the smartest guys in the room and have better predictions than anyone else, then there isn't any value bet, i.e. if they spot some value and pile in, they effectively kill the value before they can get a fraction of their bet on.
Thanks in advance.
First of all your confusing these groups or consortiums with the type of betting we are now used to in the UK.
The five main football syndicates probably haven't placed a football bet within a UK market since they started - Bloom who is chairman of Brighton started StarLizard (his data and research HQ off the back of poker winnings and Asian Hcap football betting profits - Harry Findlay was a shareholder in StarLizard for a while but the two had a fallout but Findlay saw the potential Asian Hcap betting had) - Benham worked for him but broke away with a 3mill payout (profitshare) and started Smartodds (another data and research HQ) and now owns Brentford - both use the "Moneyball" approach("value orientated" using data and analytics but to a more advanced level than what you saw in the film with Brad Pitt as Billy Beane of the Oakland A's). There is a third operation in the UK ,we shall call "Mr X" - he's slightly below Bloom and Benham's level but is a very paranoid Irish chap who pays a team to monitor the internet looking for mentions of his name. First thing you do when you obtain employment in one of these HQ's is sign a "non-disclosure" agreement basically telling you to never "spill the beans" on how they operate. There are also two very large groups based in Asia that operate in the very same way. Biggest one there is called "Chon" - They all have algorithms that estimate global betting volume and are probably by now calibrated very accurately so know how much money to place without affecting the odds . They bet with Asian bet brokers who take a small fee per x amount staked but because they have such good data and analytics most of their bets are on big margins , ie a generally available 5-4 shot when it should be 4-6 -in football that is a big difference so they can get seven figures down placed over 3-4 brokers in a matter of minutes. They analyse
everything from how much attendances affect games relative to a league/club to individual referees historical data. They create their own custom metrics and numerate everything as if you can do that , then it can be measured, if it can be measured then it can be calibrated to it's maximum advantage so R & D is a big thing. Regarding your a), b) and c)'s they do all of that (as do the horse racing syndicates)
The 40% thing was to do with the current situation in US horse racing right now where rebates between 5 to 10% are available (they used to be higher if a syndicate could guarantee x amount of dollars per annum ) - for example when the Zeljko/Walsh team moved in on US horse racing in around 2010 and were eventually hounded out by some self righteous US Senators, mostly from the Deep South and surrounding areas (how dare these clever Aussies bet into our pools and take an average 12 million dollar profit per annum) - so in a 3 year period they cleared 36 million dollars - now this group is
the largest in the world -if there is a strong Pari-Mutuel somewhere then Z and co will be in there like France right now where they are cleaning up - yes they have algorithms that can manipulate markets but they are not doing anything illegal there - all of this came out in a famous court case in Aus in 2014 where the topic was should "Pro-Punters" be taxed - people assume they win because of the rebates - the Z-man produced a 1640 page account of every bet made during that 3 year US stint and only 21% of gross profits were from rebates which means they were vastly +EV in the first place and always have been - if somebody is going to offer these teams rebates they are going to take them but that does not mean that they rely on them to win - so lets clear that
myth up first. So back to the 40% thing , it's estimated that for every 10 dollar's staked in the US right now , 4 of those dollars are from "computer teams" - because the ADW's cannot process their batch bets fast enough , a horse could be 5-2(Pari-Mutuel) going in the gate and be 4-5 at the first bend - the syndicates may have it at 4-6 or 4-7 on their line.
They don't all use the same modelling techniques but "clean data" is a must have so they have groups checking times, call positions, etc. Another big factor in the States is "track takeout including brokerage" - Saratoga for example has a track take out on average of around 18.5% where as somewhere like Tampa Bay could be as much as 25%, Finger Lakes could be in the low 30%'s - that just means the syndicates have to be more selective and require a bigger advantage at the high take-out tracks than the lower ones.
As for the UK i have consulted for a couple of medium sized syndicates - one in Ireland who actually specialised in "shop-runs" - this was at least 7-8 years ago when there were plenty more shops and still plenty of Best Odds Guaranteed especially in Ireland - their modus operandi was "value betting" permed multis - now there are good multis and absolute trash multis, they honed in on the good multis and had a large team of "runners" - i provided a tissue for each days cards along with a notable amount of other knowledgeable people - i actually advised them on how to run each day's tissue (and they had all the tissue data backdated from when they started) through some open source software to score them on accuracy - very easy to do with something like mean squared error or mean squared deviation - after doing this they dropped a few of the odds compilers and started weighting the remaining ones , hence they had built a "Weighted Wisdom Of the Crowds" which was far more accurate but as the shops declined so did BOG and the business and they had no other option but to fold and shut up shop.
The other more recent one i have consulted for on a remote basis using Zoom and private servers are based in London and were quite large players in the sports markets such as Golf, Rugby, Cricket etc - their Golf Model built on simulation from Google Earth was spectacular and was their main earner. As a sideline they provided certain bookmaking chains with in-play Golf odds and also tried to build a Pinnacle type model as their odds were very accurate but these sidelines were soon dropped due to various reasons and they went back to what they were best at - exchange trading, they also have access to Asian based brokers. They moved into horse racing a couple of years ago and their head modeller admitted to me that they did not have a clue how to get it off the ground and up and running - fair to say they have been trading in that time on horse racing and are going great guns. Payment is based on profit-share.
I will say that most of the really large syndicates all used logistic regression as their model of choice (not compound probability!!!!) - this was due to the Benter/Woods effect where those two teams used that to win absolute fortunes but as these groups progressed they added Bayesian inference, Custom built algorithms etc- Benter moved on to Probit regression in the 90's which gave him a bigger edge than the other HK teams at the time and hired the best Chinese mathematicians - even Woods has admitted before he died that Benter is the greatest living advantage gambler ever. Both Zeljko and Walsh worked under Woods for a while before they realised they could do it on their own and Zeljko has probably surpassed Benter now as Walsh is an absolute probability animal but it all goes back to Benter and his rudimentary algorithms at first and also Woods idea of taking the card-counting "team" approach and applying it to horse racing - something he should have been given more credit for imo. As the syndicates have progressed so have their modelling techniques and data analysis - years ago they would recruit from forums - i have a friend who was contracted to do some work for Woods, an Australian chap in the field of creating time figures for HK , Japan , Kranji , Singapore etc all the places Woods and his team bet into - he got that gig through an Australian racing forum. Nowadays your more likely to see Harvard and Yale graduates in their tax-haven Headquarters and they go by the term "quants"
:max_bytes(150000):strip_icc()/contrarian-5c6e6cb6c9e77c0001cda26b.jpg)
I had some of the papers already, but most not. Reassuringly I already do some of what is described in your last couple of posts. This is just in a limited, simplistic way to filter out the bulk of the field and leave me with some "probables" from which I then make a more subjective selection.
