Professor And The Peons

Tage is becoming quite a regular commenter on this blog. It may simply be that Danish doesn't translate easily to English, or it could be just me and that I am getting sensitive in my old age, but Tage's commenting style comes across as a little brusque at times, if not downright passive-aggressive. That he is an intelligent man is not in question, but the manner in which he treats lesser mortals (i.e. pretty much everyone) is.

Quoting this sentence from my post earlier:

"For me, successful betting starts with finding an edge, and this is the hard part. Once you can generate selections with a high degree of confidence in their long-term profitability, then you can play around with staking methods to increase your profits."

Tage has some comments:

Most bettors subscribe to services, they are unable to find an edge by themselves.

I beg to disagree here. In my opinion, only a very small percentage of bettors subscribe to services. 

Most tipsters use level staking, they are unable to find a better staking method.

While it may be true that most tipsters, although certainly not all, use level staking in recording results, one good reason they might choose to do this (rather than be "unable" to) is that the merit, or lack thereof, of the picks is readily apparent with level stakes. Variable stakes tend to obfuscate results. It's one reason why the FTL mandates one point bet per selection. Besides, one could make the argument that the tipster's task is to give the true prices for matches and let the subscriber decide if there is enough value, and what staking method to apply.

There are a few good services that use f.ex. (fractional) Kelly staking.

Good for them. Kelly is the King in theory. Fractional Kelly is the King in practice.

Staking is actually 'only' applied math (Utah paper) whereas building a profitable model requires math, statistics and more.

A closing comment which rather supports my assertion that finding selections with an edge is the hard part of the selection / staking process. I am well aware, as are all readers of this blog, that staking strategies can improve systems that are 'profitable at level stakes'. What they can't do is make a 'losing at level stakes' system profitable.

The fact that most tipsters don't use better staking methods is for me evidence of a lack of necessary education.

A rather elitist comment. Since when has the betting business required a 'necessary education'? It's patently obvious that many so-called 'betting experts' are not well educated - a quick look at how poorly many blog entries are written is a big clue! The lack of any regulation and the ease in which anyone can set themselves up as a tipster is one of the reasons why it is important to be diligent before subscribing.

But it is not up to the tipster to baby-sit subscribers. Only the tipster knows what staking method they use for their own selections, and as said previously, recording official results to level stakes is good for comparison reasons. A service might suggest a stake for subscribers, but ultimately is up to them to stake how they wish. I'm sure that at the end of a season, it would be rare for any two subscribers to a service to have exactly the same returns.  

That a service doesn't send out recommended stakes isn't necessarily because the tipster isn't as well educated as the brilliant Aarhus Professor Emeritus Tage. Some tipsters want to avoid the 'nanny' approach.

Many tipsters are just 'monkeys throwing darts'
Joe Buchdahl has showed this in his books and on his website.

Correct. The problem with a hugely popular sport like football is that finding a long-term edge is very difficult, and a season is a short-term window. It behoves the potential subscriber to be diligent and research a tipster before parting with their money. Even more so than the financial industry, services all come with an "As with any investment, a services past performance is no guarantee of its future success". Questions such as how are results recorded are key.

Longevity is always a plus of course, and the likes of FTL entries Football Elite, Football Investor, Skeeve and The Football Analyst have all been around for a while.

Tage also had a comment on the FTL itself, specifically in regard to the formula used for settling lay bets. Although I should probably ignore it, given his petulant demand to "Please comment on your blog", (he must think I am one of his students), I will simply say that the rules were laid out from the beginning of the season, and following constructive input from Jamie A, amended slightly in September. Updates were communicated on the blog at the time.

The relevant rules are:

Selections will be recorded against the Pinnacle Sports Match Odds prices. For back bets, this is self-explanatory. For lay selections, the price used will be Pinnacle’s +0.025 (for odds-on or evens selections) and +0.05 for others. Thus for example, a winning lay at 2.00 will be recorded as a profit of 0.98 points.

Following Jamie A's advice, the following adjustment for odds-against lays was introduced in September:

The rather naïve assumption on my part was that lays would be mostly at odds-on and shorter odds-against, but the fixed formula needs to be changed so that higher lays (2.5+) are being made into a market that mirrors that for back bets.

The same rules apply to everyone, and the implication that they give me an advantage that others do not have access to, is frankly very irritating. Perhaps Tage missed the September update, but regardless the rules are the same for everyone.

"In fact the way you settle odds on bets gives you an advantage of at least 5-6% compared to Pinnacle prices which are used for all other bet types.

All odds-on and even lays use the first formula; all odds against use the second. Using two formulae admittedly isn't ideal, but as the September post makes clear, the decision was based on a naïve pre-season assumption and making wholesale changes in mid-flow was not a solution I wanted to apply.

Given his current position in the FTL table, despite using Backs and Lays himself at various times, might I politely suggest that Tage (Sjosta) would be better served spending some of his apparently plentiful spare time and incredibly superior intellect in working on finding profitable selections rather than finding fault with how hard-working tipsters (notably TFA and FI) run their services and casting aspersions on the integrity of the FTL which I am running as equitably as my feeble brain will allow?

It takes enough time as it is each week to make the updates without having to respond to spurious accusations. I'm two hours (all unpaid by the way) behind schedule now!

Rob 'n' Banks

I wrote in my last post that in my opinion, the idea of a betting bank is conceptual and of limited use. In my experience, most bettors in practice do not ring-fence their betting bank from other funds. They over stake, blow their initial bank, and come back the next day with a new bank- which is not really the idea of a betting bank. Or perhaps despite the fact that there are funds remaining, the bettor or trader convinces himself that the system isn't working and abandons the strategy. As I mentioned before, discipline and handling losing runs is certainly one of the biggest challenges a serious bettor faces. 

For Tage "a betting bank is very real and at the heart of my staking strategy". Tage is something of a rarity though, coming from a background in statistics and taking a very statistical approach to his investing. Not exactly your common or garden bettor.

Tage's email continued with:

Consider a casino game where the setup is fixed, all probabilities are unchanged throughout.

One issue here is “The Gamblers Ruin” problem.

Given a bank B there is a certain probability P to go bankrupt placing one unit per play.

Unfortunately no one has yet found the function B = f(P)

Anyone playing a negative expected value game will over the long-term, lose. The occasional casino visit for entertainment purposes is something I am not averse to - indeed a few hours at the craps table with friends and drinks for tips is a lot of fun and by sticking with the least negative expectancy bets plus the odds (fair odds) bets, you have a reasonable chance of walking away a winner. Long-term though, it's not a viable money-spinner. No staking strategy can turn a negative expected value game into a profitable one. Betting to level stakes is the true test, and while Kelly is proven to be the optimal strategy for maximising profits for a positive expected value game, no staking strategy in the world can help you turn a negative expectancy positive in the long run.

Tage provided a link to an academic paper from the University of Utah (a state not known for its love of gambling) and while these studies are always interesting, or at least the conclusions are, I don't believe the mathematics within them is understood by too many gamblers.

Of course this is well known!

For me, successful betting starts with finding an edge, and this is the hard part. Once you can generate selections with a high degree of confidence in their long-term profitability, then you can play around with staking methods to increase your profits.

The FTL this season once again shows the difficulty in finding a source for profitable selections. While the majority of names in the table are, or at least were, relatively unknown, there are 11 services on display out of 33 entries. At the midway point of the weekend, and thus the numbers are estimates in most cases, just four of those are in profit. The usual disclaimer here that the numbers are generated from Pinnacle Sports prices and can usually be beaten. The 11 are down a combined 112.22 points and the entry with the highest ROI% is the formulaic Bundeslayga whose selections are based on the market's prices and not generated from a model. In other words, a system that should not work, but which has for some time now.

The XX Draws struggled last season due to the draw drought, (23.38% strike rate for draws), but this season is benefiting from the improved Strike Rate of 26.93% in the Big Five leagues. Football Elite are currently up, though not hugely, while other respected services are struggling. The Cassini Value selections are doing the opposite of the XX Draws, and following a great season (winning the FTL last season) with an awful season. The end of season inquest is likely to point the finger, at least in part, to the high percentage of draws in Serie A this season, but it shows the importance of not having all your eggs in one basket. In Ian Erskine's defence, his selections are mostly intended for trading, lay the draw for example, but the overall picture isn't pretty.

In regard to Betslayer's claim to know that a certain professional UK bettor uses goal times, he had this to say:

I don't need to back it up, I'm a professional bettor and I use goal times.

I know of at least two others who use it as part of their analysis. BUT the key is, its not the soft data that Jonny uses such as in the last 10 Liverpool games with a goal in 0-20 10/10 BTTS etc etc Thats no better than formlabs or other bookmaker backed data suppliers.

Of course there's no need to back the claim up, unless you want the claim to be taken seriously. Saying that you 'think' and then 'know' something with absolutely no credible evidence to support the claim means it is at best going to be taken with a pinch of salt. As the Sagan standard says "Extraordinary claims require extraordinary evidence".

The claim to be a professional bettor could do with some clarification also. While the intent is no doubt to imply a hugely successful individual raking in thousands each week, of which there are certainly a few, when I've heard this claim before, the truth is usually that the 'professional' is unemployed, (perhaps by choice, perhaps not), and is collecting dole money, which is more likely to be funding their 'professional' gambling career than supplementing it. Which is not the definition of 'professional'.

Betslayer is Rob, one of the people behind the BetOnTarget site:

If you want to look at the markets when the score is 1-2 at half time you can do that too, or how about finding out what the probabilities are when the score was 1-2 at half time and went to 2-2 between the 70th and 80th minute on a match in Portugal’s Primeira Liga ? It’s all there at the push of a button… You ask our engine questions and it gives you answers, it’s up to you what questions you ask and how you use those answers. It is not designed around a certain strategy, it is designed to give it’s users access to an expanding and potentially limitless number of strategies.

This is a constantly evolving project and has always been a joint venture between a programmer, myself, and a trader, Rob.

Once again, how is historical data from independent events useful in any way as a predictor for the future? It all sounds very scientific, but so can astrology to a gullible audience. I'd love to be told how independent events can magically become dependent events, but it'll never happen however much the project evolves. What isn't immediately clear is what makes BetOnTarget any different from "formlabs or other bookmaker backed data suppliers". Data is data, and goal times are in the public domain. And useless.

Independent Events

When two events are said to be independent of each other, what this means is that the probability that one event occurs in no way affects the probability of the other event occurring. An example of two independent events is as follows; say Crystal Palace scored the opening goal between the 21st and 30th minutes of an FA Cup game at Dover and you're interested in betting on the Bromley v Sutton United game in three months time. The probability of goals in this Conference South game is in no way influenced by the goal at Dover. 

Dependent Events

When two events are said to be dependent, the probability of one event occurring influences the likelihood of the other event. For example, you open a deck of cards and remove an ace. You then draw a card, and the probability of drawing an ace is less than any other rank because you removed an ace earlier. Now, if you were to put that pack aside, and open a brand new pack, the probability of drawing an ace here is not altered by the first ace being removed. It really is that logical. 

I'm also skeptical that a 'professional' would use historical match data between two clubs as shown in the example above, and waffle about 'some good value'. What happened 8 matches (four years?) ago between the two clubs has no relevance to a game today.

Since when has anyone HAD to bet? "Worth a look". "There could be some value in backing Sunderland".

These wishy-washy comments suggest opinions that are not supported by any data. They might make good filler for a match preview, but from a successful full-time bettor, I think we would see some meaningful data.

Speaking of nonsense, my attention was drawn to some more Police activity with our old friend. As usual, the comment doesn't appear to be written in English, and makes little sense, and for the record, I have been waiting over a year for the dreaded knock on my door:

footnote: On tweeter very recently I received tweets implying that I was the cause of the entity who had tweeted me to lose them £10 000 pounds via betting.

This is obviously a very serious allegation and I am unaware that I have ever “engaged” with anyone for the purpose of a betting “venture”

They refused to provide further details so I have a meeting lined up with the Police in an attempt to find out who is behind the claim in order to find out what the motive of the tweets are

To come full circle and return to Tage, he has long advocated that tipsters simply send out their prices on matches as early as possible, and let the subscriber worry about finding the value. Sending out the selection with a suggested or 'should be attainable' price without mentioning the 'true' price is a practice that could be easily improved upon.

Finally, on a much longer post than originally intended, back to the Utah paper, and this caught my eye:

Also, the odds bet can be made not just for β units, but for an amount sufficient to win β units. An unusual feature of this bet is that m3 < 0 for all β ≥ 0, that is, its distribution is skewed to the left. Gamblers do not usually like to make bets in which the potential loss exceeds the potential win, which may help to explain why this bet is relatively unpopular.

That last sentence seems poorly researched, although 'gamblers' probably refers here to those less sophisticated than your typical Green All Over reader who falls closer to the upper sophisticated investor end of the gambler-investor spectrum than the more naïve Jonny / Betslayer end.

The observed Favourite-Longshot Bias shows that gamblers would be better off backing at shorter (including odds-on) prices.

Except in the Bundesliga of course!  

Woo, Strike Rate, M/P/H LR and Roulette

While I didn't expect Betslayer to come back with any evidence to back up his assertion that a professional bettor would use goal times, he did return with a comment, or rather a request:

Please do another post on Jonny Grossman he is a massive inspiration (if that means tool).

When a team has played with 4 players called Henry and a goal went in off someones backside, 100% home wins, back at anything over 1.01.

His new 'strategy' of looking at individual teams and their performance in unrelated games after a goal in a time zone is laughable....no account taken of team strength, league averages or any other betting sense.

In my opinion he is on the pathway to the looney bin.

I think Betslayer means 'loony bin' and from what I read, there are many who are in no doubt that Jonny is already there, but I'm a more charitable fellow and of the opinion that Jonny is just a little different to most of us.

Data is essential, but it has to be relevant. One of Jonny's problems is that he is a slave to the data, trying to find patterns where none exist, and uses data that has no value. The approach Jonny has more in common with astrology than with astronomy, or with Deepak Chopra type woo woo rather than evidence based science. Compiling birth charts might be fun to the practitioner but no one, with the exception perhaps of Bosworth MP David Tredinnick, really believes, with good reason, that they have any merit. I'm an Aries, so I should know!

Astrology is a useful diagnostic tool enabling us to see strengths and weaknesses via the birth chart.

“And, yes, I have helped fellow MPs. I do foresee that one day astrology will have a role to play in healthcare.”

Mr Tredinnick, 65, added: “Astrology offers self-understanding to people. People who oppose what I say are usually bullies who have never studied astrology.

“Astrology was until modern times part of the tradition of medicine ... People such as Professor Brian Cox, who called astrology ‘rubbish’, have simply not studied the subject.

“The BBC is quite dismissive of astrology and seeks to promote the science perspective and seems always keen to broadcast criticisms of astrology.”

Tredinnick claimed £755 in expenses for software that used astrology to diagnose medical conditions, and rightly had to repay this amount.

Back to Jonny, and it may be fascinating to him that a Conference South team scoring first between the 11th and 20th minute goes on to win the match 73% of the time, but it has no practical value. It's as useful as knowing that Jupiter has entered Leo and will transit until the 30th.

There's no liquidity for one thing, which means there's almost zero chance of finding value, and who wants to spend close to two hours in front of a screen chasing pennies? For most of us there's a considerable opportunity cost to be considered when betting, but apparently not for all of us.

The topic of how useful is the Strike Rate from a couple of posts ago, triggered a mini conversation between Tage and Marty.

Tage:

Some players use strike rate (SR) to determine a system's betting bank. SR determines Maximum Losing Run (MLR). For SR = 50% is MLR = 10 and for SR = 25% MLR = 24

When MLR is known betting bank can be calculated.

For a DNB system, it therefore matters how SR is calculated.

Let me add that I think there are better methods to determine betting banks.

Marty asked:

How can MLR be calculated without knowing the odds bet at?

If I have a SR of 50% but bet at 10 my MLR is going to be a lot longer than if I bet at 1.5

Tage:

Your SR would probably be higher for bets at 1.5 than for bets at 10.

But if you actually have SR=50% for both types your MLR would be the same.

To save Cassini for some work I should add that my numbers are only examples. Don’t take them too literally. It’s the issue I want to focus on, not the math.

Marty:

Oh, so MLR is the NUMBER of matches you can lose in a row then (I guess assuming some worst case probability).

Seems equally as dubious as SR then.

The idea of a betting bank to me is conceptual as I mentioned here and of limited use.

In practice, bettors will call it a day on a system when they hit a poor run, and save what is left of their bank for another system. They may also simply start again with another bank if they are of the opinion that they just hit a bad run of luck. Either action renders a betting bank pointless. The discipline required to work to a bank simply isn't there for many bettors.

The longer you play a system for, the greater the probability that you will hit a long losing run. Tage makes it clear that his numbers are examples, but the MLR can't be calculated without knowing the number of bets to be made, and even then, the name MLR is misleading.

If your Strike Rate is 50%, then over 100 bets, you are likely (a greater than 0.5 probability) to hit 6 consecutive losers, but in theory, the MLR would be 100 since you COULD lose all 100. So to call it a Maximum is, in my opinion, wrong. PLR for Probable Losing Run is my preference, or HLR (Historical Losing Run) for record keeping purposes.

Just to illustrate how the PLR increases with trials, over 1,000 bets you are likely to hit a losing run of 9, and over 10,000, a run of 12.

For a 25% Strike Rate, the probable losing run over 100 bets is 12, for 1,000 it is 20, and 10,000 it is 28.

Discipline and handling losing runs is one of the biggest challenges bettors face and while many of us probably spent hours of our youth poring over sequences and back-fitting staking systems, not so many of us have been able to stick with one in practice.

I write about a gullible young Cassini and his Roulette adventures a couple of years back. On the off-chance you missed that post, and shame on you if you did, the relevant portion is this:

It was in the late 70s when, anxious to try out one such brilliant system, I wrangled an invitation through a colleague to an illegal Chinese run gambling den somewhere close to Piccadilly Circus. I was a little surprised that not only did they let me record the outcomes of spins and take notes, but they provided the pencil and cards for it. This was all too easy I thought.

I didn't win, and had to go in to work next day.

My partner in crime that night stopped coming in to work soon after. He was working the graveyard shift, and thought it would be amusing to call the department manager up at 3am and report a problem with the computer system. In those days such a call necessitated getting dressed and driving in to the office, and he was none too pleased to arrive at the office and find my erstwhile friend somewhat the worse for wear and slurring "April Fool" to him. I never saw him again. Perhaps he found the Roulette Holy Grail?

FTL: New Leader

Six entrants were in action on a quiet midweek, and there two winners. Mortimer surged into profit on the back of a 6.59 point profit, while the other five all wiped out with not a winner between them.

The other winner wasn't in action, but took over at the top of the table after BettingTools.co.uk picked three losers. The table shows Gecko looking down from his lofty perch, and shows those who saw either a change in points or a change in position in the table.

FTL 3k Winners

The FTL entries reached the 3,000 winners mark this weekend, but the overall losses increased, although by just a modest 10.74 points. There were three big winners. Jamie A was top dog on points and places making 10.35 of the former and climbing eight of the latter to move into the green. Gecko made 9.55 points to close the gap on leader BettingTools.co.uk (+3.01) but stayed in second place as these two pull away from the pack, and Mortimer made 8.54 points moving up seven places.

The only other profitable entries were Fairfranco, who will be pleased to turn a profit (1.75) after a tough four months, and Overgoalify who picked up 1.35 points.

Once again, it was a tough week for the more famous, or infamous, names in the table. The Bounty Boys all lost points again and their portfolio is now down by 14.04 points. More importantly for those above them in the table, their liabilities now total £1,725.

Football Elite dropped three places to 11th after a loss of 5.23 points and is now the only Bounty Boy in profit. Skeeve dropped six places to 20th after all five selections lost. Football Investor dropped four places to 22nd with a loss of 3.05 points and the TFA's Draws lost 0.30 points and dropped just one place.

It was another tough weekend for draw seekers, with just 11 from the 40 top league matches, and the Draw Portfolio also lost points. Six of those 11 were in Serie A, with none in France or the EPL, although with just the one game, that's perhaps not a huge surprise! The XX Draws lost 4.84 points, Draw picks lost 1.89, TFA Draws lost 0.30 as mentioned, Drawmaster lost 0.42 and TFA Euro Draws lost 3.31 points to remain in last place.

The biggest loser of the week was the XX Unders which followed a Cup week of 10 wins from 3 with a weekend of 11 losers from 15 and a loss of 7.68 points.

Here are the leaders with no changes in placings on last week. The Prize money includes the Bounty bonuses:

The chasing pack currently in profit is comprised of the following:

The 17 entries in the red are:

Jamie A takes the early lead in the monthly competition, his 11.13 points leading Gecko (+9.55), Mortimer (+7.37) and BettingTools.co.uk (+6.80). TFA_Raz props up the table down 8.65 on the month.

The net loss climbs to 115.89. Coincidentally, the Cassini wins and Bounty Boy wins are both at 307.

A few qualifying games in play this midweek as the Conference National has six games, plus a handful in the lower English Leagues as the FA Cup quarter-finalists make up their lost match.

Joan left a comment on the Push / Yield debate:

congratulations for your last post about Push bets and Yields. Personally I also include to my ROI the pushed bets that aren't postponed matches.

Another similar issue could be how to account the strike rate, which influence have a push bet on it, not consider the bet, some kind of strike rate reduction. The same issue will arise with half won and half lost bets, which part of them would have to take into account to increase or reduce the strike rate. I haven't found a simple solution for this yet.

I do track the strike rates by excluding any Pushes from the calculation, but I don't find it a very useful number. Abromo currently leads the table based on Strike Rate at 71.9%, but the strategy he uses sees him sitting on a small loss. Contrast this to Trend to end:

The lowest Strike Rates are those shown above, and Trend to end is currently in 5th place and up 16.88 points with the third lowest Strike Rate. "Lowest" is meaningless without knowing the context, i.e. what method is being used. Laying long-shots will give you a high Strike Rate, but if more win than implied by the odds, you won't make any money.  

Finally, Home wins this season are at 10 year lows in the EPL and Serie A and in the Big Five overall, while Draws are at a low in Ligue 1, League One and the Conference National.

Had you been lucky enough to decide to back the draw blindly in Serie A this season, you would be sitting on a pile of 68.07 points (Pinnacle's prices) as they are following their lowest draw strike rate for 10 years in 2013-14 with their highest this season, 34.11%.

For Aways, La Liga, Ligue 1, League One and Overall are at 10 years highs with only the Bundesliga and League Two seeing 2014-15 Aways below their 5 and 10 year averages.    

FTL Update 4.Mar

The midweek action was good for Bounty Boy Football Elite who shot up five places on the back of a 7.00 point profit.

Not so good for the other Bounty Boy in action, The Football Analyst Graeme, (aka the self-proclaimed Barcelona of the FTL I might remind him) whose draws lost another 3.63 points dropping him another place.

BettingTools.co.uk extended his lead at the top with a gain of 3.79 points thanks to Wigan Athletic's win at Norwich City. Skeeve and Football Investor kept their powder dry for this round, and the Bounty Boy combined portfolio is now down by just 0.46 points after 919 bets.

@ValueBankFooty was next best picking up 6.45 points, and moving ahead of Sjosta who lost 4.57 points. Also up five places in a congested middle of the pack (14 entries are up or down by less than 10 points) was Online Trader making up 4.27 points and moving into the green.

TFA_Raz was the big loser with his loss of 6.09 points just beating the 6.03 points lost by Rubicon.  The entries with changes either in points or places this midweek are shown below. Losses on the round moved the total losses to -105.15 points.

A Ton Of Luck

I mentioned in the last post that taking an ROI seriously after just 100 bets probably isn't the best of ideas.

The FTL this season illustrates quite neatly how different entries have gone on good and bad runs. The six monthly prizes have each been won by a different entry, and there have been five 'worst of the month' losers. For those wondering, it is Fairfranco - he managed to have the best record in November, followed by the worst record in both November and December.

Peter Norsted's Drawmaster is another good example. After 100 bets this season, Peter's ROI was an impressive 18.9%. 

However, a prospective investor carrying out his (or her) due diligence, would have looked at Drawmaster's record in 2013-14 (-16.45 points from 183 selections, ROI -8.99%) and suspected that a regression to the mean was likely.

Sure enough, after 222 bets so far this season, Drawmaster's ROI is -8.5%, and over the two seasons combined, the record is 405 selections, -35.40 points and an ROI of -8.74%, which is not quite as impressive as that 'middle sequence' of 100. 

Peter writes about this, saying:

We are going through a bit of a torrid time with the draws and indeed we have only had a couple of reasonably successful weeks.

The draws should at least be hitting 30% however currently we are hitting around 27%

Last season's strike rate was actually 26.8%, so this seasons 27.5% is an improvement. but here the ROI is a more useful measure of success than strike rate. The torrid time may be relative to the lucky start, but 27% is close to the long-term strike rate.

I'm not sure what Peter means by 'should be hitting around 30%'. He includes some leagues that I don't touch, but in the top five European Leagues, the overall strike rate for draws is 27.2%.

In the Lower Four English leagues, the strike rate is 25.1%. League One and the Conference National are at 10 year lows.

Overall in these nine leagues, the draws are hitting at 26.0%, a number that is beaten by four of the Draw Portfolio members, but only two of whom are currently showing a profit. It's worth mentioning again that the prices used in the FTL are usually beatable, so these numbers are on the low side.

It would be interesting to know why Peter thinks his selections should be hitting at 30%. 

Pushing On

Tage comments further on the Push / Void discussion:

As my comment clearly shows I’m of course aware that I place two distinct bets in the situations I describe.

My answer was to the question posed by Tage which was "How to record this bet?"  This bet, not 'these bets'. My answer was that his examples were those of two bets.

And I agree with you that if ROI is the only purpose of a betting record your method is flawless.

I don't believe that I have ever suggested ROI is the only purpose of a betting record. I would suggest that ROI is something of a peripheral statistic to be used with extreme caution.

But I actually use my own and tipsters betting records for many other purposes. 

Presumably you know exactly how the ROI numbers are calculated. As the threads and comments have shown, there are different opinions on this.

And I see problems with your method.

Conceptually I do in fact regard a surebet as one bet.

How else can I place two 10k bets with a 2k betting bank?

A betting bank is also conceptual, and as I will come too later, 'sure bets' have a tendency to not be so sure.

Most exchanges are smart enough to calculate the amount available to bet. So it’s possible to trade a golf tournament for 100k starting with 5k.

I doubt you record the 150 single bets on a golf market separately, and if you do I can’t imagine how the profit/loss on each bet should be recorded.

How one records the 150 single bets is up to the person placing them, but each bet SHOULD be recorded.

Serious investors should record all their bets, although grouping say 50 losing bets with the same book into one for convenience isn't a big deal.  

For statistical purposes a betting record with arbitrages is dangerous using your method.

A tipster has 100 bets in Denmark Superliga with a ROI of 3%.

I don’t see evidence of this tipster having an edge betting the Superliga if all 100 bets are parts of surebets.

What is dangerous here is someone giving the ROI% any value after just 100 bets! As the FTL has shown this season, entries have gone on runs, both good and bad, and 100 is too small a sample.

If a tipster's strategy involves arbitrage, surely this would be made evident?

I (mistakenly) thought you would agree that the simple method you describe is only useful in simple situations.

I went up a level and described what I called ‘a negative arbitrage bet’.

Using the definition of a surebet there is a guaranteed profit. Not here. There will in most cases be a loss. The profit occurs in case of a PUSH or a VOID. As your article was about PUSH and VOID bets I offered these examples.

My post was written to cover binary or football bets, and how best to record them and have an accurate, if not necessarily very useful, ROI%.

For conceptual 'surebets' where the risk is zero, ROI is meaningless (although whether or not the risk is ever really zero is arguable given that you might not necessarily get your money back once invested. Sportsbooks folding is hardly a rare occurrence, and there are other risks such as losing out to the 'palpable error' rule).

If I were forced to come up with an ROI% for these, I would personally record the individual bets and combine the bottom lines, but make it clear to any third-parties how the figure is calculated.

Back to the usefulness of ROI, and it is something that Steve at Daily 25 keeps track of:

Long run, most of us would be ecstatic with an ROI of 5%, and more than happy with a 1 - 2% net return.

Place Your Bets (Plural)

A comment from Tage regarding the Push / Void debate, and to further the debate he has this to say:

There are some interesting variations on this theme.

On a tennis match you place a negative arbitrage bet at two different bookies.

The underdog is bet at a bookie who only requires 1 ball played.

The favourite is bet at a bookie who requires the match to finish 'normally'

Your profit will of course come when the favourite retires through the match.

This is not a void bet, but it is not a push bet either. How to record this bet?

The solution to this apparent conundrum is apparent when you see that Tage errs by considering this arbitrage to be one bet.

An arbitrage is created by two (or more, but we'll stick with two to keep it simple) discrete bets which when combined create the arbitrage, but there are two outcomes which both need to be recorded individually.

While one might want to consider a manufactured arbitrage as one bet, it clearly isn’t.

As Tage himself says “The underdog is bet at a bookie who only requires 1 ball played. The favourite is bet at a bookie who requires the match to finish 'normally' ”. "Bet at two different bookies".

Two bookies, two bets, two entries in the betting records. 

Arbitrage bet in sports betting is an activity where you simultaneously place bets on all possible outcomes of an event at odds that guarantee profit, whatever the result of the event will be. These bets are also known as surebets, miraclebets, surewins or just arbs.

In Tage’s example, he places two distinct bets, Bet One on the Underdog and Bet Two on the Favourite.

If the game never starts, then both bets are, of course, Void.

If the game plays out to its natural conclusion, then one bet is settled as a Winner, while the other is settled as a Loser.

If the game is abandoned due to a retirement, Bet One is settled as either a Winner or a Loser (depending on who retires), and Bet Two is settled as a Push because this bookmaker refunds the money if the game doesn’t end.

If the game is abandoned for reasons other than a player retiring, then unless there is some bizarre rule about paying out on the player who leads at the time, then the two bets would each be settled as a Push.

Tage goes on to offer a similar situation from Baseball:

Another bet type I had great success with 10 years ago.

A very well known UK bookie dipped their toes into MLB betting.

But they treated a baseball match exactly like a soccer match, that is, the odds stood even after a change of pitcher.

So place a bet on a big favourite at a bookie who voids the bet in case of a pitcher change and the arbitrage bet on the dog at the UK bookie.

This situation is a possible 'half void' bet. How to record this?

Again, the point here is that it is incorrect to refer to this situation as “the arbitrage bet”. The arbitrage opportunity is the result of two discrete bets, and the outcomes of each should be recorded separately.

Betting Tools Brian had this to say:

Interesting one Cassini. I originally included void bets in the ROI calculation on my TipsterTable but I changed when I read what you said on your site.

My thinking before then was that they should be included because the money was 'invested' without a return. Even if the money was never at risk due to a postponement, another potentially profitable investment may have been missed.

In most cases it will make little difference however and as long as the methods used are clear and consistent then I think either is acceptable.

It’s certainly true technically that once the money has been handed over for a bet, that money is not available to invest elsewhere, but because the bet requires the event to take place before it is ‘live’, I think there’s a valid argument to differentiate between a bet that is never activated, and a bet that is.

What is a little more awkward is how to handle an abandoned match. I lean towards recording this as a Push since the bet was live, and money was thus at risk so it more closely resembles a Draw No Bet, than a pre-game postponement, but I can see the argument for simply pretending the bet never happened and voiding it. A game abandoned in the opening few minutes bears more of a resemblance to a game called off before kick-off than does a game abandoned after 89 minutes. 

  
However, abandoned games are so rare that it doesn’t make a significant difference - of the 7,674 bets so far this season in the FTL, I can’t recall a single abandonment - but handling Draw No Bets as Voids rather than Pushes can make a difference.

Randolph has had slightly more than 10% of his selections end as "No Bets" and the difference between recording these as Void or Push moves the ROI% from 9.13% to 8.18%. While Randolph isn't a service, or at least not that I know of, Football Elite is, and his ROI% is either 1.16% or 1.07% depending on how the bets are recorded. 

Football betting may not be as mature as the financial markets yet, but it's growing as a serious option, and as it does, it will be important to potential serious investors what their expected ROI% really is.   

FTL Update 2.Mar And Speciousness

Another tough week for the FTL entrants, with an overall loss of 41.33 points. It was particularly bad for the Bounty Boys portfolio which lost a combined 20.47 points, and whose combined bounty liabilities have now increased to £1,525. On the season the overall net total is now down by almost 100 points with 148.19 points lost in January and February.

The potential (still a long way to go yet) additional bounty money is good news for anyone in the top 20 though, and the total league pot up for grabs currently stands at £3,175.

It was also a very bad round for the Draw Portfolio which lost a combined 18.80 points. The top five European leagues had just 11 draws (from 46 matches) while the lower four English leagues (Championship, League One, League two and Conference National) had just 9 (also from 46 matches).

Peter Nordsted's Drawmaster selections lost another 9.27 points, TFA Draws lost 5.62 points, Draw Picks lost 2.87 points, TFA Euro Draws lost 2.27 points and only XX Draws showed a profit, albeit just a whopping 1.23 points but on a tough weekend, I'll happily take it.  

It wasn't all doom and gloom though, with the big winners being Gecko and TFA_Raz and of course four entrants advanced to the semi-finals of the Erskine Cup.

Gecko made an impressive 12.49 points to close within six points of leader BettingTools.co.uk (-2.21), and as a bonus threw in that Chelsea would win the League Cup (although this match wasn't a league game and was thus ineligible). Gecko moved up five places, and with a gain of 9.44 points, TFA_Raz moved up nine places.

The leaders in double figures, with the new and old ROI%, are:

The four Erskine Cup Quarter-Finals were played, and in the first tie there was no treble joy for @ValueBankFooty this time around and his loss of 6.45 points combined with a 2.90 point profit for XX Unders, meant a comfortable win for the latter.

Randolph graced the league with his presence for the first time since November and picked up a small profit from two short-priced selections, but he could have remained idle since Jamie A had a bad weekend losing a round worst 12.26 points.

Abromo was a no-show which meant that all Talkies Tips needed to advance was a profit, and he did that with some ease making 3.97 points.

The fourth tie was between Club Havana and Gecko, and while both made profits, it was Gecko's huge weekend that trumped Club Havana's more modest profit of 1.17 points. Club Havana does have the consolation of being the only entrant so far to win a monthly price and a cup prize.  

The semi-finals to be played over the weekend of 20-23 March will be:

Back to the league, and the number of entries in profit dropped to 15 after both Football Investor, Fulltimebettingblog and Jamie A all dropped into the red. The others in profit are:

In the red zone we have:

The totals and Portfolios now look like this

The February monthly prize goes to BettingTools.co.uk although in the end it was closer than expected. Only six entries were profitable last month:

No rest for the League Commissioner as there are plenty of games this midweek in England, and some elsewhere, and the entries are already coming in.

For those of you into Golf betting, I came across the Golf Betting Form web site which had this to say about this blog:

Green All Over – possibly the best betting related blog around. Mainly taken from the angle of Betting Exchanges and focusing on Football and US Team sports. Well written (and sometimes combative in style!), there is plenty that can be gleaned from the blog if you want to bet successfully.

Possibly? If there is a better one, I'd live to know about it so that I can add it to my blog roll. I'm also not sure what is meant by "sometimes combative in style".

Moving on, and Betslayer left a comment on my recent excellent Ignoring The Result post writing:

You are getting a little obsessed with goal times....I imagine and in fact know that Benham and his operations as applied to football include goal times in their analysis!

I'm hardly obsessed with goal times. I do have a habit of tracking how close to the end any draw-breaker goal comes, (just so I can torture myself with "if only"), but as I have explained before, while goal times in football are interesting, they are not useful for predictive purposes. The suggestion that Matthew Benham would base his betting decisions on basic historical goal times is no more credible than that he would look at the result of a game in 1930 between two teams before making a decision:

Previously the two clubs met in the 1930 Semi-Final with Arsenal advancing after a replay winning 1:0 at Villa Park after a 2:2 draw in the first match at Elland Road. Hull led 2:0 with 20 minutes to go in the first match, and I therefore perceive value in laying Arsenal in the half-time market. These matches are particularly relevant given that both were played on neutral grounds, as of course the FA Cup Final will be. Interestingly 1930 saw the first World Cup finals and with this year also being a World Cup year, I perceive there is value to be had in backing Uruguay to win the World Cup.

Yeah, makes sense! I'll leave readers to make up their own minds about Betslayer's claim that he "imagines and in fact knows" anything at all about Matthew Benham and his strategies, but if you know something for a fact, you don't preface your remark with "I imagine and..." !  

Such a claim should be easy enough to back up with some evidence, but I have my doubts that any will be forthcoming. Somehow, I can't see the rather naïve Biggie Betslayer (last blog post May 2013) as an insider on the hugely successful Benham's "operations".

Admittedly, "including goal times in their analysis" is a rather wishy-washy statement (one would hope that every detail available would be included if you are betting big money) but the implication that raw goal times are given much weight when taking serious betting decisions is specious.

I really wish successful football betting were as easy as looking at historical results or historical goal times. We'd all be rich! But of course it isn't, and making money consistently from football is not easy.

I imagine, with a high degree of certainty, that Betslayer is talking nonsense, and not for the first time. I'll let you all know should any evidence be forthcoming, but don't hold your collective breath.

Yields Down When Push Comes To Void

Threads on the Betfair and Blogabet Forums made me realise that some of the the ROI percentages in the FTL are slightly inflated.

I have a faint recollection of this topic coming up early in the season, with the infamous Jonny commenting that:

When ROI is calculated by betting syndicates it is normal practice to exclude void bets

This is correct in my opinion, but bets resulting in a Push are not the same as Void bets.

When a bet is a push, that does not mean that money was not risked – it only means that one gets the stake back after the bet is decided because the match ended in a certain way. It means that this amount of money has 0% yield (or interest) between the time of betting and the time when the stake is returned (it is not available to be invested elsewhere).

The fact that money was at risk is key, and is why Push outcomes should be included in the ROI percentages. When a match is Void, for example due to a postponement, the money was never at risk. Bets on the outcome of an event are conditional upon the event actually taking place!


The next update of the FTL will therefore incorporate this small change. Only seven of the 33 entrants have Push outcomes (Draw No Bet) and the highest percentage is Randolph's, who has just over 10% of his selections ending this way. His ROI% before this weekend was 8.93% under the old method, 7.99% using the new.

I'll actually list both ROI% calculations for a couple of weeks and if anyone has a different opinion about the above change, I'd be interested to hear it, but I think I have it right, albeit a little late.

Advanced Poisson Guide For Sale

Courtesy of Twitter and Mark, it was brought to my attention last week that Hejik is selling an “Advanced Poisson Guide”. I was asked my opinion of this, but contrary to another of Jonny Grossmark’s unfounded beliefs, (he has several), Hejik and I are in not actually in partnership to take over the world, (evil laugh), and the mundane truth is that I don’t know him (Hejik) from Adam, although Hejik does actually exist, which is likely a plus. I therefore have no knowledge of such a guide.

I was a little surprised given Hejik's thoughts on Poisson less than twelve months ago (5th April 2014) where he clearly has major doubts about Possion being a good fit for football, and further admits that the perfect model does not exist:

There seems to have been much debate about Poisson on various forums of late and the danger would appear to be the perfectly natural beginners obsession with building the perfect model.

Such a thing does not exist as you'll well be aware and 'best guess' is a reasonably fair assessment of what professional bettors are doing.

As long as your 'guess' is extremely well educated you have half a chance at the books who are also, apparently surprisingly to some, employing 'best guess' pricing models, not psychic witches with crystal balls.

For what it's worth, I think it's important that we question how efficiently Poisson fits to football and observation would suggest it's far from perfect.

However, even without that information it's easy enough to arrive at the same conclusion when we consider that some of the biggest bookmakers in the world are actively promoting it as a method of beating them. I'm not sure they'd be so inclined to do that were it even near as effective as they suggest.

Note the comment that Poisson is a far from perfect fit for football and the suggestion that bookies promote Posisson because it is ineffective, so we have something of a voltafaccia, as we say in Italy.

  

As many readers will know, I use a modified form of Poisson myself, (unmodified Poisson is not a good fit for football) so this is an area of interest, and I am only too well aware that the quality of the output (i.e. probabilities and thus prices) generated is only as good as the input, which is the goal expectancies for both teams.

What is not clear at all from Hejik's sales pitch is where these goal expectancies come from. If you want a tool that accepts the two goal expectancies and generates match odds, you can find one for free at BettingTools.co.uk web site. I’m not sure if this calculator adjusts for the infamous zero-inflation problem, but turning input into output is the easy part.

Where any value would lie is in the calculation of the input parameters, i.e. accurate goal expectancy figures, and Hejik is very much aware of this fact, so my concern is around the maintainability of meaningful numbers. 

Hejik’s guide supports the top leagues in England, France, Germany, Italy, Spain, Scotland and The Netherlands, as well four lower English leagues down to Conference National level, and claims that they “require minimal input from the user to maintain them – 20 minutes per week is generous”.

From my own experience in maintaining numbers for the top five leagues in Europe, it takes a lot more than 20 minutes per league, so either the data Hejik is maintaining is relatively unsophisticated or I am extremely slow.

I like to think it’s the former, in which case the output will be similarly unsophisticated. I’m speculating here, but 20 minutes to enter match data for 11 leagues would barely be enough time to enter in the match results, and as has been much discussed previously here, match results are often misleading and need to be placed into context.

For a start, as is the case with brains, goals are not created equal, which means that even for this basic starting point, a certain amount of effort is required to make them useful data. One would hope that few would argue that the first goal in a match is more relevant than say the fifth goal of a 5:0 thrashing, although having read some of the nonsense that is spouted about goal times or historical head-to-head results having predictive value, it may in fact be more than a few!

The other concern I have regarding this guide is the lack of any evidence that it actually works, and somewhat related is that if indeed it does work, why would Hejik be selling it? Selling eggs from your tried and tested Golden Goose is one thing, but replicating and selling the geese themselves for just 47 each would seem to be an odd business move, aka shooting yourself in the foot. 

If anyone has some first-hand knowledge about the quality of this product, I would be interested to hear from them (anonymity guaranteed of course). £47 is not exactly a fortune, and while the guide may well be worth that amount for entertainment value alone, or for provoking ideas for your own model, I have my doubts that such a ‘one size fits all’ tool is profitable.