Thursday, 23 February 2017

Away Days, Data, and Detachment

Slightly concerned that I may have been in the midst of a senior moment, I was relieved to see Brian confirm my numbers on the goals per game in the EPL. 

Hi Cassini, Thanks for the update. I've just queried my database too which is populated using files and got exactly the same numbers as you, with an overall average of 2.65 since 1993/94.
The key takeaway from this is the importance of having quality data, as mentioned on page 16 of James Butler's seminal work "Betfair Trading Techniques". And don't rely on someone else's research. 2.4 goals per game! 

Brian has another post on this topic, going back (almost) to the start of the Premier League, with details on the numbers of matches going Over and Under 2.5 goals, along with the average goals per game.

Brian discusses possible reasons for the recent increase in goals per game, but in my opinion, it's all part of the decline in home field advantage that I have touched on before. Home teams are still scoring pretty much in line with long-term averages, but... the first 19 seasons of the EPL (1992-2011), Away teams averaged 1.1 goals per game. 

In the almost six seasons since (2011-17) this average has 'shot' up to 1.19 goals, and this season is currently at 1.23. 

The number of Home goals per Away goal declined from 1.39 to 1.29, and Away teams are now winning over 30% of matches (compared to 26.5% previously). 

Last season Home teams had their lowest winning percentage ever (41.3%) while the four best seasons ever for Away teams have all been in the past five seasons. 

It's not just an English phenomenon either - I wrote in May 2014:
Away wins were at record (10 year) highs in England, France, Italy and Spain. Only Germany failed to set a new high, although they were still 8 wins above their 10-year average of 89.
One other observation is that except for Spain, in every other league the Home Goals Per Game (HGPG) minus the Away (AGPG) figure, i.e. the home advantage, is down on its 10 year average.
In Germany it is at 0.34 from 0.40, in France 0.37 (from 0.41), in Italy 0.35 (from 0.39) and in England 0.38 (from 0.42). Spain was at 0.51 (up from 0.42, in part because of the frequent thrashings handed out at home by Barcelona, Real Madrid and this year Atletico Madrid).
In August 2014, I added:
One other trend worth drawing attention to is that of Away teams.
Historically, since 1947, Away teams have won 25% of matches. Home teams have won 49% and Draws the remaining 26%.
But averages can be deceiving. In the days of Division One, Away teams won 24% of matches.
Since the Premier League was formed, they have won 27%, and have steadily chipped away at the long established home advantage.
The last three seasons have seen an average Away win percentage of 30.3%. This is actually quite remarkable because it is a number not seen in a single season since 1947. Away teams over this three year period have averaged 1.22 goals per game, a number not seen since England won the World Cup.
Here are the win percentages over time – notice how the Away win averages are steadily gaining over the years, and in the last three seasons in particular:
Away Pct
I need to update that graphic.

The difficulty is always in determining whether observations like this are blips or a trend. Wait too long and any edge you may have found has long gone. Jump in too soon, and the trend becomes a blip and there never was an edge.

The thing with Unders and Overs betting is that your losing runs will be short and in low over-round markets, any losses should be small even if it's a blip you're following.

Brian concludes his post with these words:
At the risk of stating the obvious, if there’s one thing that is clear from all this, it’s that whether it’s the unders or overs that you are looking to back, you need to be finding odds of over 2.00 to show a profit long-term.
Not quite correct. If you limit yourselves to Overs where the implied probability is in the 50% to 54.99% range, your ROI is 12.4% (18.39 from 148 bets). There's nothing magical about the 2.0 / evens price. What's magical is when the actual probability is greater than the implied, and that can be at 1.01, 1000, or anywhere in between. But more likely at the lower end of that spectrum. 

G, possibly not his full or real name, left a comment on my Failing Assets post, writing:
Excellent (and often over-looked) comment on detachment.
Something I've found recently that is much easier when placing "place and forget" bets.
Much more difficult when trading a match/game etc.
Was curious if you found the same - obviously you're not trading as much/at all now- when your were up in the higher echelons of the trading world along side Messers Webb, Berry, Iverson et al.
I found fairly early on that the key to detachment, which is basically taking the emotion out of trading, is to stake sensibly. Back in 2011 I wrote:
While I don't consciously have an amount that I am comfortable betting, my body soon lets me know if I've over committed - and my wife can hear my breathing pattern change to heavy, so she knows as well and stays away!
It's hard to stay detached when adrenaline is in full flow, but by staking sensibly I was able to do a fairly decent job for a part-timer.

I'm not sure it's fair to compare me with the pros listed above, although at least one of them allows the fact that it's his job to have a negative influence on his trading. (How much the other two made on the Australian Open tennis that they were talking up last month has yet to be revealed). I was only ever a humble part-timer, in it for the intellectual challenge rather than the money.

If you are trading with money that you can afford to lose, then you are more likely to trade optimally, without emotion, than you are if you need the money to pay bills and feed your young children. No update from Mark on his horse racing venture, which may be good news (he's seen sense) or bad news (he's not making minimum wage) but he did have an embarrassing Tweet posted:
As many readers will have spotted, Mark has fallen victim to Survivorship Bias:
Survivorship bias, or survival bias, is the logical error of concentrating on the people or things that "survived" some process and inadvertently overlooking those that did not because of their lack of visibility. This can lead to false conclusions in several different ways.
Trading successfully requires, among other things, believing in yourself. The issue is that for many, this belief is irrational, and they do not have the skills or resources required and thus ultimately fail. Simply really, really wanting to make thousands doesn't overturn the laws of probability in the long run. And praying doesn't help, trust me. 

Trading came along at the perfect time for me. Not only did I have a career, and a job that allowed me some flexibility regarding hours, but I had no small kids to distract me, and was in a relationship with someone who left me alone to do my thing. For some reason she was happier with me trading most nights than with me going to the pub! 

I was also in on, if not the ground floor, certainly the first floor, of the exchange concept, and there were many opportunities back in those days.

Since then, court-siders, green-siders, pitch-siders, sooie-siders (in Arkansas, good joke if you get it) and other-siders have moved in, Premium Charges have been introduced, Betfair's liquidity has declined in most of my sports, and my time is better spent researching trends and betting and forgetting, although the consequent increase in nights out and drinking will probably shorten my life expectancy by a few years. 

Incidentally, if you want to measure how good you are as a trader, compare your average win size with your average loss size. The higher the win to loss ratio is, the better a trader you are.

No comments: