Sunday 25 November 2018

Ignore The Rest, Sometimes

The importance of rest is well known in sports, and is more important in some sports than others. In baseball, the starting pitcher typically has four of five days between starts, while the NFL schedule means teams always have at least three days rest between games. NFL scheduling these days makes sure that no 'short rested' teams plays against a longer rested team - only one game on a Thursday in early December 1997 (Cincinnati Bengals v Tennessee Oilers) was an exception, when the Bengals played on the previous Sunday while the Oilers enjoyed a full week's rest after playing on Thanksgiving Day. Not that it did them any good, trailing 0-38 after three quarters before losing 14-41. They were drilled, you might say.  

The two sports where rest does matter are the sports (excluding baseball) where teams go on 'road trips'. For the most part, the relative differences in rest days between two teams is factored into the lines. As I said, the importance of rest is well known. What is considered and often discussed is the impact of a game going to overtime, but I suspect something that isn't generally considered is when the previous game was a comfortable win or an uncomfortable loss, but in either case, players are rested in the fourth quarter and thus while technically the game is a back-to-back, in practical terms it is not.

In other words, an NBA team playing triple overtime in a close fought game should be considered differently in their next game from a team that had a 20+ point lead early in the third quarter and coasted to victory, allowing the bench players to finish off the game.

In the 2017-18 season, this system had a 12-9 record, and backing such teams against the spread this season now sits at 4-1. All time the record is 135-109-5 (55.3%). 

Including home teams, the above numbers become 9-2, 3-0 and 200-170-8 (54.1%) respectively.

There are a couple of other parameters that can boost the historic win percentage to 63%, but selections become few and far between and not many people have the discipline to wait for all the planets to align.

There's also a definite edge on Unders in these games, 56.2% overall an as high as 62.4% with other parameters.

I'm not sure how excited Christopher Haley of Winnipeg was when he saw this Tweet yesterday given that the last time anyone scored five goals in one NHL game was on 2nd February 2011.

But you can guess the rest. Finland's 20 year old Patrik Laine, defied the odds by scoring five goals from five shots in the Jets 8-4 victory and win Mr Haley a nice Can$1,100,000, payable over 20 years.
This game was a selection for the NHL system, but my profit was a little less than Mr Haley's, although I don't have to wait 20 years for it. So that's good.    

Friday 23 November 2018


Some of you may have caught a discussion on Twitter yesterday regarding the use of statistics in betting, specifically in football and horse racing. 

Because Twitter's character limit can be a little constraining, here's a summary, though you can read the full conversation here.

The basic claim of A Lucky A  Day was that because soft books "seem to" use early prices as loss leaders, betting on horses pre-race "can be profitable for someone in their living room". I begged to disagree.

A Lucky wrote that:
Opening horse racing markets are generally less efficient than opening soccer markets. 
While pointing out that one can't compare a ternary event with a horse race that may have over thirty possible outcomes, I think this is generally true, the reason being that privately held inside information is what makes a market inefficient, and as I have written before, horse racing is a sport which will always have insiders. It's far easier to make sure a horse loses than a football team.

Large price moves alone don't mean a market is inefficient. In fact it could well mean the opposite, and that the market is simply reacting quickly to smart money coming in to the event.

A Lucky continued:
The more inefficient a market is the more opportunity there is to profit from knowledge. 
This statement is essentially correct. An inefficient market is one "in which an asset's market price doesn't always accurately reflect its true value. In an inefficient market, some investors can make excess returns while others can lose more than expected". 

The problem is that the knowledge required to make sure you are on the "making excess returns" side of the coin, isn't available to the living room punter. It's the insider who has this edge. The armchair punter is essentially guessing. He may win sometimes, he may lose sometimes, but ultimately in the long-run he will lose out to commission / transactions costs. 

Lucky seems to think that an armchair punter can use public data to gain an edge:
Everything I've read says in an inefficient market all public information is not factored in, so the edge can come from interpretation of public information, In an efficient market the edge can only come from private information.
The source of his confusion perhaps lay in the false assumption that an inefficient market "was the opposite of Eugene Fama's definition of an efficient market. The papers that I have read, that test betting markets for inefficiency, do it by trying to beat that market using public information".

I'd love to see those studies! Of course all public information is factored in to markets, wherever they lie on the scale of 'efficiency'. (An inefficient market is not the 'opposite' of an efficient market any more than red is the opposite of violet in the visible spectrum). 

The argument that "we don't all interpret data in the same way" doesn't mean that the data isn't factored in. 

Nor does the suggestion of Daniel Górka make too much sense, that:
because [the] majority of gamblers don't know at first place how to compile their own win probabilities and then how to to use/incorporate those information into their pricing process
In which case, these would hardly be people in a position to gain an edge in any market.

The truth is you're not going to gain a edge in horse-racing markets by looking at data available to the public, and which has been pored over by the thousands of people who are looking for a bet. 

To imagine that you have a unique ability to spot something thousands of others have missed is naive at best, and frankly more than a little arrogant.

If someone does have a unique ability to read markets and thus has an edge generating a long-term profit, some thoughts.

One, they would not be writing trading manuals, producing videos or offering trading courses to give away their edge.

Two, someone else will find the same edge one day, and it will disappear.

Three, with a skill like this, they should move into the far more liquid and lucrative world of financial  transactions.

Thursday 22 November 2018

Overs Keeps Giving - Thanks

With 250 games in the record books, i.e. 20% of the regular season, it's a good time to look at how the NBA Overs System is faring in this 2018-19 season.

At the 216 points entry level, the Overs currently have an ROI of 9.6%.

The 220 points entry level last season had an ROI of 12.2%, the best for 200 selections or more, but this season is currently falling a little short of that number.

I thought the 220 point level would average about two bets a day, but we are averaging almost exactly four a day, which is OK if they're winning I guess.

Regular readers will have noticed that some strategies carry over from one sport to another, and while backing Overs in MLB games when the total is on the high side has previously been discussed, I'm not sure I've mentioned this strategy in the NFL. Below are the results since 2010.
A reminder that today is Thanksgiving, and there are three NFL games in play. 

Monday 19 November 2018

The Big 6 Premium Charge, Pinny Style

I postulated yesterday that when a Big 6 team are playing, Pinnacle know there will be a lot of interest, and have started to increase the over-round.

A Lucky A Day had another idea, asking:
This did seem a reasonable observation, given that many of the games with the higher over-rounds had one team at short-odds, but having researched further, this doesn't explain why this season is different to previous seasons. Short priced odds-on favourites are nothing new, and a match between Tottenham Hotspur and AFC Bournemouth in October 2017 actually has the lowest recent  over-round 101.2%, so that's no excuse.

Going back to the 2015-16 season, only 40 matches (of 1,260) had an over-round above 102.5%, and 35 of these featured a Big 6 team playing a Little 14 opponent. Four were all-Little 14 games and one was a Big 6 match.

Here's the thing - all but five of these 40 matches were played since April 2018. 

Something has changed, and the evidence would suggest that it's deliberate.

A season finale game in May 2016 was the first of the five exceptions, and the second was also an end of season game in May 2017. Both featured Manchester United. 

The other three were in August, September and October of 2017, and then none until April this year; the lull before the storm.

Again, Pinnacle's decision to elevate the over-rounds in certain matches is fair enough. Non-monopoly businesses can charge what they want, and can choose which customers to do business with for that matter, and if the customer doesn't like the terms, then they are free to take their business elsewhere. 

The impact of over-rounds is not understood by many bettors, and books tend not to shout about their margins, so it's unlikely that too many people have noticed this yet. 

That books got away with over-rounds in excess of 112% in football for many years, and still do in some horse races, is a good indicator that the average customer is not sharp. They either see betting as a hobby and care little that they lose money, or they are irrationally exuberant. Caveat emptor.

There is good news though. Of the 40 matches discussed above, Bet365 had a lower over-round in 15 of the highest 17, and in 25 of the 40. It's good to have options.

Sunday 18 November 2018

Pinnacle's Big Seven

In my recent post on the EPL Draw, I mentioned the troubling trend whereby Pinnacle's over-round this season has crept up from last season's 102.11% (itself up from 2016-17's 102.05%) to 102.24%. 

We're still early in the season, but a closer look reveals that the larger over-rounds are in matches where there is a Big 6 team playing a Little 14 team. 

Of the 17 matches where the over-round is 102.5% or greater, 16 fit this description. The one exception is Crystal Palace v Newcastle United and there's a legitimate argument for including Palace in any list of 'Big' teams. Two FA Cup Finals since 1990, a Zenith Data Systems Cup win in 1991, back-to-back FA Youth Cup wins in the 1970s, first ever winners of Division Three in 1921... the list goes on. 

Conversely, where the over-round is less than 102%, 23 of the 28 games are all Little 14 matches. 

It seems likely that when a 'Big' team (or Crystal Palace) are playing, Pinnacle know there will be a lot of interest and push the over-round up, while for the matches between 'smaller' teams such as below (no sign of Palace here, which rather backs up my theory about them), the odds on offer are a little more generous. 
It makes perfect sense from a business perspective, but it's something to be aware of as the season resumes next week.

Unofficial Sixteen

Pleasing to know that at least one reader made some money yesterday on the College Football selections. 

Not all of the 16 selections may end up as 'official' selections, (Northwestern moved from 'dog to pick 'em), but it was a good day. At a little after midnight, the state of play was:
Iowa State ultimately lost, as did Ole Miss who went to overtime, but a 10-6 day is excellent, even if it looked at one time like an even bigger day. Frederic J commented:
Attention turns to the NFL today, with four probable 'official' selections for the Small Road 'Dogs with a small bet on the Tampa Bay Buccaneers at 2.38. 

EPL - 18 Years Of The Draw

I mentioned in a recent post that I would have more on the English Premier League (EPL) Draw, and as a man of my word, here it is.

Thanks to Joseph Buchdahl’s website, it is possible to look at prices for matches going back to the 2000-01 season, i.e. we have 18 full seasons of price data. Unfortunately back in the olden days, the over-rounds were a lot higher than they are since Pinnacle arrived on the scene, so making comparisons of earlier seasons with more recent ones isn't fair.

It's worth mentioning that Pinnacle's over-round appears to be creeping back up this season, currently at 102.24%. Perhaps not significant to most readers, but if you are taking your betting seriously, it's something to be aware of.

Many readers will be familiar with David Sumpter's book “Soccermatics: Mathematical Adventures in the Beautiful Game” and in other articles he has promoted the idea that backing the Draw in Big 6 games is a profitable strategy:
When two big-six teams meet then all the focus is on one of the two teams winning and the draw is forgotten by the punters and the bookies. These circumstances are somewhat unique to the Premier League, because there are so many ‘just for fun’ gamblers involved in the market and six teams with a worldwide following.
There hasn't always been a Big 6 of course. The EPL began competition with the 1992-93 season, and is now in its 27th season.

In the early seasons, Manchester United dominated, winning four of the first five Championships, before Arsenal, Chelsea and Liverpool joined United in forming a ‘Big 4’.

Of the twelve seasons from 1997 to 2009, these four teams filled the top four places six times, with three of the four in the top four the other six seasons. 

Only Leeds United, Newcastle United and Everton were able to break into the top four places in the final table during this time.

The nine seasons since have seen the emergence of a Big 6, with Manchester City and Tottenham Hotspur being added. These six have taken the top six spots in four of those seasons, five of the top six four times, and only in 2015-16 were two of the six displaced, famously by Leicester City who won the Championship, and less famously by Southampton who finished sixth.

As an aside here, since the formation of the EPL, only three non-Big 6 teams have won the FA Cup, namely Everton, Portsmouth and Wigan Athletic, and Tottenham are the only Big 6 team with no Cup wins in this time.

Back to the League, and I thought it might be interesting to compare the returns from the Big 4 matches (2000-09) and Big 6 matches (2009-date) with the over-round adjusted to the 102% of recent seasons. 

First, here are the results of backing the Draw in every EPL game from the 2000-01 season: 
As I've written before, there are many matches where backing the Draw is akin to throwing your money away.

For Big 4/6 matches over the same period:
Clearly, blindly backing the Draw in all Big games isn't a profitable strategy, and why should it be? Big 4 teams have finished as low as 7th, while Big 6 teams have finished as low as 10th. A 'Big' match isn't necessarily a 'close' one. 

For the unwashed masses, "Back the Draw in Big 6 matches" is a nice and simple system. However, David Sumpter has also written:
It turns out that when two well-matched teams meet (i.e. the probability of a home win is only slightly bigger than the probability of away win) then draws are under-priced. When matches are skewed so there is a strong a favourite (i.e. the probability of one team or the other winning is larger than the other) then draws are over-priced.
While the definition of well-matched teams is vague, there are a couple of measures I like to use. One is to exclude any matches where a team has a 'true' win probability greater than 0.5, and where the 'true' probability of the Draw is 0.25 or less.

Using these filters, backing the Draw in 'Big' matches from 2000 would have resulted in this:
The other method is to calculate the difference between the 'true' win probabilities for both teams. For teams within 40% of each other the profit is 22.81 points from 232 matches, an ROI of  9.8%. A more selective rule of less than 25% and this climbs to 41.78 points from 155 matches, an ROI of 27%

Expanding this criteria to include all EPL matches, and the profit is 137.67 points from 1,431 matches, an ROI of 9.6%

ROIs around 10% over 18 years should catch everyone's attention.

Saturday 17 November 2018

Power To The Group Of Five

The College Football season is in the final weeks of its season, with another bumper crop of selections for the Small Road 'Dogs system likely today - 16 at the time of writing. 

The number of selections this season already comfortably exceeds the previous high (113 in 2004), confirmation perhaps that home advantage in this sport, as in others, is not what it was. As this informative 2016 article noted:

Coaches and analysts churn out the same old talking points about the rigors of winning road games. They're becoming harder to believe. College football is changing. Home-field advantage has nearly disappeared.
Well, not exactly, as I explain further below.

However, if you've been following this basic system this year, you're in a good position to record another profit.
If you're being a little selective and just playing the Conference games, you're even further ahead. 
It's a little late for this year, and by next August everyone will have forgotten, but it seems a few UK readers aren't too well informed about the structure of College Football. This article from the UK's Online Betting Guide does a good job in explaining how it all works. 

Be aware that not all conferences are the same. There are ten major conferences, in two groups of five. One is called the Power Five, and these are the Premier League equivalent, with the Championship comprised of the not very imaginatively named Group of Five.

Since I'm in a generous mood this morning, below are the overall records for home teams in these ten conferences since 2014, along with the results for the Small Road 'Dogs by Conference in that time.
Home teams won far more games in Group of Five matches, but the Road 'Dogs had a better return there than from Power Five matches where the home team won just 52.3% of games. With four of the Group of Five teams, and two of the Power Five with a double digit ROI, a little filtering can make a big difference.

As the article I referenced above explains about home advantage, nothing stays the same in sports. Here is how home advantage has changed in the last three seasons (2016-today):
While the overall home winning percentage has increased for the Power Five, it has decreased for the Group of Five, and every Road 'Dog in those conferences has a double digit ROI in that time. 
I first wrote about key numbers in football betting a little over six years ago, and Road 'Dogs getting exactly 3 points are another way to boost your ROI. 

Going back to 1996, (arbitrary starting points tend to distort statistics, but one has to start somewhere), and 39% of the profits have come from road teams receiving exactly 3 points, although the percentage drops to 16% when seasons are weighted (bottom three rows) to give more value to recent seasons.
The +3.5 line is the number where 47% of recent weighted returns are garnered, and backing just these two lines over the last five seasons has an ROI of 24%. Over the past five seasons the +3.5 line has a ridiculous 51-21 record. 

Friday 9 November 2018

October, Tesla and Totals

I'm back. My three and a half hour scheduled surgery last Friday is now a distant memory, although the scars are there as proof, lest I forget. Still a little banged up, but the surgeon was pleased with the outcome, and my habit of visiting the A&E every couple of months should now be behind me.

It wasn't the best timing though. I'll spare you the details, but on the Thursday my old Mum had a fall, and is now in hospital while my even older  Dad, who can't be left alone, has moved into a care home, at least for a month. Not the best way for them to celebrate their 63rd wedding anniversary.

In other words, there's been a lot going on, and likely will be a lot going on over the next few weeks or months, and sports investing hasn't been a top priority. 

Overall, October was at one time looking likely to be a record monthly loss, but an end of month recovery meant the final amount lost was just 3.8%. I should make it clear that this is all investments, only a small percentage of which are sports investments.

Perspective is always good after a disappointing month, and this was only the second down month since February of 2016, if exclude May 2016's 0.01% loss which was basically a push. 

My Tesla tip is looking good, with a new chairwoman replacing the always interesting, but maverick, Elon Musk this week.

In the more exciting world of sports, the NHL system now has an ROI of 7%, the NBA's East in the West is at 10.1% while the NBA Overs at last season's 215.5 entry point has a 62-59-1 record, basically a push. 

As I cautioned in this post last month, the lines have moved higher and it wasn't a bad time for me to be away from the action. 

In the entire 2017 season, the Totals were 230 or more only 31 times. In 2016 30 times, and on five occasions in 2015. You have to go back to April 2010 for the previous time.

Contrast this with 2018 and 35 games so far have the Totals in this range, and the season is less than a month old!

In American Football, the College Road Dogs are now headed to their eighth consecutive, and 14th of the last 15 seasons in, profit: 
In the NFL, the basic Road Dogs are just behind:

With four of the last five EPL Big 6 matches ending as Draws, the basic Big 6 Draw strategy now has an ROI of 65.6% this season.  

Per Blogger, my most popular all-time post is January's Probability, Difference and The EPL Draw and I'll have some more fascinating insights into the EPL Draw in the next few days, as time and recovery permit of course.