Wednesday 14 May 2014

Chewing The Numbers EPL and Bundesliga

As I mentioned in my previous post, the 2013-14 seasons for the English Premier League and Bundesliga are now in the books.

Looking from a draw perspective, the two seasons have much in common. In Germany, the goals per game average went over three goals per game for the first time this millennium, which reduces the probability of draws of course, and the resulting 64 draws (20.9%) was just one higher than the lowest total of the previous ten years.

In England, the goals per game average was actually the lowest in five seasons - by a single goal. The averages have been remarkably consistent over this period:

With two teams each scoring 100+ goals, one might expect the draw percentage to have been in line with expectations, but the draw count was 78 (20.5%). It rises slightly with Liverpool and Manchester City's draws excluded to 21.9% but not by much at all.

As was the case with the Bundesliga, the 2013-14 draw count was just one higher than the previous (last ten seasons) low. Percentages around 20% are both well below the 5 and 10 year averages which in England were 26.6% and 26.3% respectively, and 24.8% and 25.2% in Germany.

More significantly, the 0:0 and 1:1 draws were down in both leagues. In the five seasons between 2009 and 2013, there were an average of 74.4 low-scoring draws in England. In 2013-14 this total was 55, a decline of 26%.

In the Bundesliga, the five season low-scoring draw average was 55.6. In 2013-14, a similar 23% decline resulted in a total of just 43. Th 2:2 draw was also at a new low in Germany (12) although the nine 3:3+ draws were at a high. Unfortunately these rare flukes are not the profile of match we are looking for - in other words, the wrong type of draw.

The Premier League is especially interesting, because the one goal games (1:0 / 0:1) were up on the five year average, totalling 75 instead of the expected 60. A 25% increase in these scores, with the 0:0 and 1:1 down, appears to be rather bad luck. Would it have been too much to ask for a handful of these to end up as draws? Apparently!

For the heck of it, I looked at the 2:0 and 0:2 score too, and these were also up - by 20%. And just to rub salt in the wounds, the 2:2 (semi-fluke) draws came in 21% less than the average, and the 3:3+ draws (massive fluke) was about the same.

At least the Bundeslayga had the decency to have a decline in one goal games as well as 0:0 and 1:1 draws.

As I have discussed in the past month or so, it has become all too apparent that making a profit is not a good way to measure the merit of a system. Of course, it's human nature not to look too closely when the profits are coming but the silver lining in the cloud of the XX Draws making a loss is that I came up with the idea of comparing the selections to a benchmark. In the case of draws I chose all draws priced below 4.0, and they produced some curious results.

Using Pinnacle Sports draw prices, in both England and Germany, backing all draws priced at under 4.0 would, not surprisingly have resulted in a loss; 34.88 points in England, while in Germany it would have been 21.34.
Coincidentally, and not the last coincidence of this post so read on, the percentage of winners in both leagues is almost identical, at 24.6%. The XX Draws performed rather well in England, ending the season with a profit of 3.54 points, and beating the benchmark by 38.42 points. The Bundesliga is not quite so impressive (and while I understand that making 3.54 points is hardly impressive, it is relatively impressive), with the benchmark beaten by just 4.24 points, despite having a slightly lower strike rate. Overall in these two leagues, the XX Draws beat the benchmark by 42.66 points.

Just for fun, I looked to see if the Favourite-Longshot bias continues in the draw price, and it does. Backing the draws from the shortest up to 3.51 in England - 3.52 in Germany - would have resulted in small profits in both leagues, with the winning percentage in both leagues again identical - this time at 29.7%.
I'll have more data once the seasons end in France, Italy and Spain, but from the first two leagues I would say that the XX Draws were a little unlucky to be competing in a season that appears to be an outlier in draws.

It was a record season in the EPL for Away wins though, (32.4%) and anyone with a system backing Aways should be aware that the benchmark for these would have resulted in a 19.3 point profit (35.68 points if restricted to selections priced at under 9.0).

No comments: