Friday, 5 July 2013

Serious Questions

A few more comments on the past couple of posts, and first this clarification from Secret Betting Club:
I would like to clarify the comment that claims that the Secret Betting Club endorses GYTO.
SBC have never reviewed GYTO, let alone endorsed them. The person who left this comment clearly is not an SBC member and is woefully mis-informed.
I think that's clear enough.

Danny then left a second comment on the Most Bookmakers Don't Like Winners post which made me think that perhaps he is not well, but a later comment we shall get to did reveal that this wasn't a serious. Danny has too much time on his hands. He should try maintaining Elo based ratings on a few sports - that's fill up his spare time!
While we're on the subject of limited possibilities, what news of our old friend the ELO ratings? I do believe it is some months since they graced these pages and they appear to have faded like some old Dire Straits chord.
As I understand it they started with chess - if I beat you I 'own you' as the kids round here say and I get your points. Something like that. Then throw in some shots on target, shots and goals and weight it (I once read that hitting the post is not a shot on target!). Then if you are still awake it spits out a WDL number and you rush off to Betfair in your case to bet accordingly.
So how did the ELO ratings perform? Is it a case of 'this time this year we're millionaires Rodney!' Or did they hit a bum note and you quietly Jonjo Shelved them?
Cassini I'm going to steal your thunder. The only explanation must he this - Pinnacle, on hearing of your fabled 1980's exploits against the London bookmaking trade, have LIMITED YOU!
The Elo (not ELO) ratings are alive and well and being used for a number of sports, but I have no idea what a WDL number is unless it's a poor way of saying that you can use Elo ratings to calculate probabilities for various outcomes. The latest Elo effort is baseball, and still a work in progress, but you enter the team's last few results, the opponents, the starting pitcher's ERA and it spits out an estimated price which can then be compared to Betfair or wherever you place your baseball bets to identify value. (The advantage of Betfair is that you can bet bigger and trade out either to lock in profits or reduce your losses). Here's one from a game last night which suggested the Los Angeles Dodgers were 10% value at 2.21.
Unfortunately the Dodgers lost on this occasion, but not before taking a lead and allowing me to lock in a profit on both sides.
With the importance of the starting pitcher, baseball is a hard game to use Elo ratings for, but a winning streak like the Dodgers were on means that the 'body' of the team is performing well, with the rotating starting pitcher perhaps being seen as the 'head', and as I mentioned in a couple of posts in April, odds can be out of line once the starting pitcher is out of the game

As Danny well knows, the football Elo ratings are very much alive and well and are the heart of the XX Draws which generated 158 winners (draws) from 517 bets last season (implied odds of 3.27, for an ROI of 8.57% and 44.3 points.

Someone asked me for a breakdown by month of the 2012-13 season, and while I haven't drilled down by league, here they are:
The May / June losses may well be due to the end of season games not being played as truly as they might have been, but this wasn't seen in 2011-12. One might also expect August to be a tricky month with the ratings yet to settle down, but while the profit wasn't much, at least it was a profit. December (the busiest month) was the worst month; in fact December through March showed a loss in total, so the profits came in the three months of October, November and April. 

Back to comments now, and here's yet another one from Danny:
Serious questions for once:
1) How are Nate Silver's ELO Baseball ratings performing?
I have no idea - I have my own baseball ratings. How are Nate Silver's Elo (not ELO) Baseball ratings performing? Do you have Internet access?
2) Looking at Chelsea QPR last season - with the benefit of hindsight the result was a total outlier and there were clues at the time such as QPR having no shots and little possession. Do results such as this throw the ELO ratings right out as I believe the weak team would get a lot of points from the strong team?
I'm not sure what you man by 'clues at the time'. In the game in question, Queens Park Rangers had eight shots, five of which were on target. Possession in itself is meaningless without knowing a lot more about the type of possession. Minds far sharper than mine have looked at this, but here's a summary on this topic from the excellent The Power of Goals blog:
However, it is increasingly becoming apparent that the relationship between possession and wins is far from straightforward. In this post I outlined an earlier view that possession merely tells you how long teams spent trying to do certain things on the pitch. It doesn't tell you what those actions were, it doesn't tell you how successfully the actions were translated into really important things, such as goals and it doesn't tell you how effectively each side carried out those tasks. Game states, tactical approaches and relative skill levels between the sides and of course randomness, decide match outcomes and how these are played out on the day decide the largely secondary statistical measure that is recorded as possession.
Results like this do not throw a good Elo (not ELO) rating system out because they use more than just the final result to determine the post-game ratings adjustment, so along with the margin of victory, various weightings are assigned to the data. Chelsea had almost three times as many shots as QPR (23), had more shots on target than QPR (11) and out-cornered (I made that word up) QPR by 14-2. While the relationship between corners and goals is unproven, they likely reflect 'pressure'. Chelsea did have a 58:42 edge on possession for the record.

I use an even more lopsided example from the EPL in 2010-2011 in my 'prospectus' - the match between Birmingham City and Chelsea.
Chelsea had 24 shots to Birmingham City’s 1. Chelsea had 9 shots on target to Birmingham’s 1. Chelsea were awarded 14 corners to Birmingham’s 2, yet the result was a 1-0 win for Birmingham. It is my opinion that by including more statistics than simply goals, more accurate ratings are achieved. Birmingham did not deserve to receive a big boost to their ratings that day as the overall statistics did not merit that.
The low scoring nature of football means that such results will happen from time to time, but by
including more data than simply goals scored, the impact of a freak result is less dramatic, and I would suggest, more relevant to the ratings.
I have seen nothing since I wrote those words to change my mind.
3) What are your thoughts about the STL interview with "professional gambler" Ian Erskine? I was wondering if you could run that clever number checking technique you wrote about once that checks the integrity of numbers (you see I do occasionally pay attention). His declared earnings are £1,423,678, quite an unusual figure in itself! 4) For £595 he will give me the "Millionaire Mindset" and six cups of coffee. Is this VALUE???
I personally know little about Ian Erskine other than that he seems to be a genuine guy who has done very well for himself, in part due to his 'Lay The Draw' system which I have commented on many times before, and even Ian concedes that the glory days have now gone, and that "the margins are tighter", but also from successfully marketing himself and his product. What I do agree with Ian on is that attitude / mindset is all important. One of the things I believe he has said, is that you can teach a system to a group of people, but when they go to apply it, some fail and others succeed. He says in the interview mentioned:
Most are looking to get rich quick and think it is easy.
And:
I think discipline and mindset are my greatest assets. I accept the fact you have losing days since I keep records and play the game long term. Most people give things a week or less and then move on. I cover this with an example in the seminars of where I carried on this year using a particular method and it has netted me in excess of 40k in 6 months, someone I know had exactly the same information and gave up after 2 weeks in January when the weather was iffy so racing was a bit inconsistent. Most people have no discipline, can’t ride out losses and manage their money abominably when they do bet.
I think this is very true. I don't think many people start investing in the traditional financial markets and expect to get rich overnight, but sports investors have a tendency to be impatient, and give up on a system after a few losses, and jump on to something else. For some reason a betting method generating "only" 5% ROI is considered a waste of time, yet a return of that size in interest on a bank account would be jumped at.

If someone has made £1.4 million in 7 years from betting, then they clearly have done something right.

The integrity of numbers check that Danny mentions is Benford numbers which I wrote about back in March 2009. Unfortunately they don't work on one total, but you need a number of individual and independent totals such as daily, weekly or monthly to determine their legitimacy. 1423678 has no repeating digits, (6% probability), and only one digit lower than the preceding one, but it's hardly enough evidence to question its accuracy. My grand total at the end of June 2008 had all seven digits unique, although sadly for me, two of the seven were behind the decimal point! 7821936 to be precise so it's not that unlikely. Since I went to eight digits in late 2008, this has yet to occur, with the probability 1.8%. In the post linked to above, I ran my own daily P/L numbers in 2009, and then again in 2011 and they came out pretty close to those expected.
Idle moments are rare these days, but it might be time for another update on these.

Would I pay £595 to attend the FTS Roadshow? No, the value is not there for me. While not in Ian's professional league, I don't think I do so badly as an amateur with a professional approach, i.e. the right mindset.

£595 is a lot of money for me, but it may well be money very well spent for someone who needs an example of what is possible with a disciplined approach. My concern though is that as soon as the show is over, many people will again give up at the first setback and revert to their old, losing habits.

Ian certainly isn't perfect though - he supports Tottenham Hotspur for one thing, (prepare for an opening day loss Ian) and overuses the word 'gubbed' in his Tweets!

5 comments:

Jill said...

£595 a lot? I thought your ELO ratings were making you fortunes?

Have you considered the horses, I see Jack's doing a 2 day course for a bargain £695.

http://www.tradingforfreedom.com/sp/21145-trading-workshop


Anonymous said...

How much money do you think tipsters make from running seminars?

Danny Murphy said...

Cassini I'm glad I'm not chained to a set of Elo ratings and do leave myself freedom and time to think.

If the Elo powered XXDraws is yielding 8.57% then why not volume it up at Pinnacle and SBO next season then? Why pay 60% of your winnings back to Betfair, it makes no sense.

Elo ratings don't work for the obvious reason - if they did the bookies would all use them, only better versions. Infact they may use them among other things already. You are playing right into their strength ie big data and statistical modelling.

If Elo ratings are giving a number for team strength then why does your version not perform on the win market?

If your ratings really were producing 8% yields on draws then others will have noticed (how could they not given the ceaseless publicity on here) and the market will simply correct itself next season.

If Elo ratings work and you have a proven edge of even 4% then why sell for £99? Infact I remember Peter Webb blogged about exactly this situation, I think it was called 'What's an edge worth?'

Why would you do that???

Danny Murphy said...

Cassini (or perhaps Ian) - why should I pay for Ian's course when everything needed for the 'Millionaire Mindset' is available for free on this triffic blog?

If you do delve on here there are a lot of nuggets to be had (not to be confused with the Nugget Crew). Eighty percent is complete bull**** ofcourse but the remaining twenty is gold.

I wonder what Thomas Bayes would make of Ian's course? What is the probabilistic chance Ian can give me the 'Millionaire Mindset' I so crave? I have to say it's a worry he's a Spurs fan, he may harbour the wholly unrealistic expectations that bedevil so many of his flock. And hopefully his seminar isn't one from off 'The Shelf'.

Decisions, decisions.

Therese said...

This is fantastic!