Friday 5 May 2017

Future Savvy

Tony Stephens had a fairly long comment, actually more of a question, mostly on the theme of ‘blip or trend’:

Sad to see people walk away from sports betting/trading (whatever you want to call it) but if something stops being fun or too costly then you must ask the question of why you want to continue.
It would be interesting (for me) to get a bit more understanding around what we should consider to be good or bad trends if such a thing exists. Trends exist all the time possibly over various dimensions/overlays of time periods.
In say greyhound racing we see the sp fav opened up as fav at 2.0 and SP'd at 1.5 with a straight line inbetween. The trend there was very much a downward trend and you would pat yourself on the back (or some other form of celebration that takes your fancy) if you got on near the 2.0 or even 1.75. Now, if we then looked at the trend of all the dogs that opened up in the market as a fav over the past year we might determine that 75% of them actually tended to drift more than they shortened up. So a trend that occurred within the 1 race for maybe thousands of seconds looked like a good trend initially but in fact may be more likely to be a bad trend to follow race to race.
So can a good trend be identified within an individual race / match / day or should a good trend generally always be over a much longer period? Does this make sense?
Just a quick comment first on his opening paragraph before addressing the main point.

I think statsbet’s decision to walk away from sports betting/trading is actually laudable and quite the opposite of sad. Spending 40 hours or so a week away from home working is bad enough, but if you are bringing in a salary or wage, i.e. providing for your family, and investing in your and their future, it's a necessary evil and most of us who are employable and want to better ourselves do it without too much complaint. 

However, spending your additional leisure time in the evenings or weekends and probably slowly giving some of that income back is not. If you have small children, as statsbet does, that time is irreplaceable.

What is sad is the amount of time and money that most gamblers waste pursuing unrealistic dreams. This pastime seems to attract more than its fair share of young and undereducated individuals, looking for a short-cut to success. Some have children that they no longer see on a daily basis. To me, that's what is sad. 

Betting / trading is for most of us, a hobby, and hopefully one that doesn't prove too expensive, either in financial terms or in time, which is arguably more important. Unfortunately it is not a victimless hobby as Joseph Buchdahl eloquently puts it:
Science is teaching us that the uncertainty of rewards, rather than the rewards themselves, is what drives a curious mind. Gambling presents itself as a kind of elixir of knowledge; ultimately it disappoints but has us coming back for more. This is the addictive power of maybe.
Credit to statsbet for understanding the realities of it all and where his priorities need to be, and a thank you to him for sending my hit count through the roof this month. The unrealistic dream of a 1,000 hit daily average might actually be possible. After four days of May:
Now to Tony's question about trends. In the context I use the term, a trend cannot be identified within an individual event, but over a period of time or number of events. 

It's important to understand that a trend is something that is usually long term in nature, for example there's a trend for people to marry later in life, or to have fewer children. You wouldn't identify a trend based on one month or one year of data, but after several years.

In a sporting context, a trend might be that there are more or fewer goals being scored on average. 

It is not a trend if Crewe Alexandra have scored one goal in each of their past three games, or Swindon Town scored early against a northern team starting with the letter B on a Tuesday night in the past two Novembers. That's noise. You can find patterns everywhere if you look hard enough.

However, what we're really interested in is finding a weakness in the market that we can profit from. 

The term “blip or trend” can be seen everywhere, e.g. in economics, property, currency, stocks, bonds, etc. here and for anyone who is interested in the topic, there’s an excellent book called Future Savvy: Identifying Trends to Make Better Decisions, Manage Uncertainty, and Profit From Change by Adam Gordon which defines a trend as:
…a sequential pattern of change in record data – a change evidenced by a rise or fall of variables when measured between two points over time. To be commonly considered a trend, rather than a fad or a blip, a pattern in the data must pass basic tests of significance.
Now the problem with sports markets is that they change, sometimes literally overnight. Even the most casual of users on the exchanges will tell you that, for example, horse racing markets are not the same today as they were five years ago, maybe not even one year ago.

They have evolved, track-siders may have moved in while Premium Charge payers have moved out for example, and will continue to evolve, so there is little to be gained from looking at markets from too far back.

You’ll read about people who claim to have huge databases with tens of thousands of records going back to the dawn of, if not time itself, the exchanges, and how certain patterns can be seen, but looking at markets from way back is of little relevance to today. It might impress someone unfamiliar with data processing, but once the parameters of a market change, old data is worthless. Garbage in, garbage out, I seem to remember from my programming days.

So the quote from Adam Gordon “when measured between two points over time” has to be modified for sports to ensure that the two points are comparable. Looking at football as an example, looking at matches in the two points for a win era, or prior to the change in the offside law, has no value.

Football is a good sport to look as an example in this debate. I’ve previously revealed that blindly backing Away teams in League Two has been profitable in recent seasons. 

Not only is there a trend showing that Away teams are winning more games (from 2000-2011 the average was 28.3%; for the six subsequent seasons, it was above 30% every time, with an average to date of 32.3%) but more importantly, the Away price is currently under-priced. It's useless hitting at 50% if the price on winners averages 2.0 or less.  
The data doesn’t lie, but the nature of sports betting means that by the time any statistical “basic tests of significance” have been passed, it’ll be too late. The edge will be gone.

Incidentally, while it would be unrealistic to expect to be able to always match the Maximum Odds, note the difference to your returns that a little shopping around can potentially make. Pinnacle may be the best in the business for most of us - generally competitive prices and a business model that doesn't close accounts - but they are not always the best on price.

Betting on League Two has a couple of advantages – markets are liquid, prices are widely available, and the standard of players is still high, but the amount of interest in the league is much less than that for the top European Leagues. The markets are less efficient.

If you go too far down the football pyramid, you run into problems not only of liquidity, but also the problem where the VORP (Value Over Replacement Player) of certain key players becomes critical. 

For an example of this, just follow Skeeve and his detailed analyses of games at the National League level (one down from League Two for any non-UK readers, the league where pro meets semi-pro).

At this level, you need to emulate Skeeve, and since not many of us have either the time or inclination to spend hours reading injury reports and tracking suspensions, back-four combinations, weather conditions etc., you’re better off just signing up to his service and watching the money roll in. And before anyone asks, no I am not in any way connected to Skeeve.

Back to trends, and if you can explain why your trend might be occurring, the trend driver, you will be more confident in its veracity. A drop in weight can be positive (you’re watching your diet and exercising regularly) or negative (you’ve got cancer) – having an explanation for your results is always a good thing.

I should also mention that a trend needn't necessarily be on the results side of the equation, but can be on the supply (price) side if they change as a result of different expectations from the market.  

In other words, the strike rate is important, but equally important are the prices available. It's as useful to us if there is an increase in Away wins while prices stay the same, as if the Away wins remain constant, but the market adjusts its prices in our favour. 

Football is a little different from many sports though, and changes more rapidly. 
Many sports are run with teams as franchises with no promotion or relegation, but in the Football League for example, every season sees a turnover of at least 25% of the teams. This is a significant disruption each year, bringing with it new managers, new ideas, new players and new venues.

This is good though - with change and disruption comes opportunity.

And things don’t stay constant within a season in football either. Aside from injuries, transfer windows and suspensions which affect the playing side of things, there were no less than 18 managerial changes for the 2015-16 League Two teams, either prior to the season or during it. Four were pre-season, seven were before the end of 2015, and seven more were before the end of season.

This was quite an increase from the eight changes during the 2013-14 season and consequently very disruptive.

I’m not claiming this as a reason for the Away bet showing value over the past five seasons, just giving an example of the thought process behind trying to determine a trend-driver.

League Two is also the most competitive division of the Football League. Over the last five completed seasons, the average point differential between 1st and last (24th) is 50. 

In League One, this number rises to 55.8 and in the Championship to 56.2. This difference will increase again after this season with League Two currently at 50, League One at 63 and The Championship at 70, with one round remaining.

As for when to start acting on what appears to be a trend, I take the approach of starting slowly and allocating a certain amount to the project. One losing round of results isn’t important, perhaps even one month of losses isn’t, but in the same way that you use trailing stops in the financial markets, there’s a line that once crossed says enough is enough.

Ideally once your profits start to build, you can increase your stakes, but adjust the line at which you are going to pull the plug as you go. All trends will end, and it would be folly to build up 100 points of profit over a couple of years, and then see it all disappear. How much you take off the table is a personal choice, but waiting for the trend to return can be lengthy and costly.

In the long run, we are all dead – John Maynard Keynes

Anyone verifying my numbers, all calculations are courtesy of Joseph Buchdahl's excellent Football Data web site, may have noticed that the system has slumped since the turn of the year, -13.87 points in 2017 using Pinnacle's Closing Odds, and up a slender 1.85 points at best odds. 

I’ll be updating the final numbers for the League Two Away system after the final round this weekend, and it'll be an interesting one. 

In my youth, Hartlepools United became Hartlepool AFC before the United was restored, although not the 's'. I rather like curiosities like that. The story of how Queen's Park Rangers lost their apostrophe is interesting too, but that can wait for another day. 

Hartlepool, whatever their name at the time, seemed to always be seeking re-election back then, and every time (a record 14) they were successful. 1924, 1929, 1939, 1960, 1961, 1962, 1963, 1964, 1970, 1971, 1977, 1978, 1983, and 1984. Phew! 

Their luck may have finally run out. If Newport County win, the Poolies are down. If Newport slip up, a win for Hartlepool United over already promoted Doncaster Rovers would save them. Doncaster might be less than fully committed in front of a packed Victoria Park if news filters through that Plymouth Argyle are winning at Grimsby Town and the title is out of reach, but it doesn't look good for Hartlepool to continue their Football League career.

1 comment:

Tony S said...
This comment has been removed by the author.