Recently, I looked at a large sample of away teams losing 1-0 at half-time and discovered that they find it very difficult to make a comeback and win the fixture. In this article, I explore the degree to which we might expect that the home team, similarly fighting a 0-1 deficit, fights back to win the game.
I fired up my Excel spreadsheets, dating as far back as the 2008/9 season, and calculated how many times the current crop of Premiership teams have come back from 0-1 down to gain all three points. From this I was able to designate each team with a win ratio applicable to these situations.
Home advantage is worth taking into account when using a predictive model to work out the odds of two teams winning. But at HT, if the team that you follow are losing at home by the scoreline of 0-1, then for a number of teams, motivational speeches by the manager, it would appear, are simply not going to work.
Although the sample is small, we see in the table below that there are some clear and discernible trends. Just over 15% of the teams in the sample won when losing 0-1 at HT at home, so it appears that the chance to “fight back” is suppressed in this scenario (although not as strongly as it is in the 1-0 half-time data).
Where the sample size of potential come-back opportunities is greater than one, Arsenal, Spurs, Fulham, and Chelsea lead the way in terms of win ratio.
Table : Teams losing 0-1 at home at HT – Results since 2008
(You can sort the table by clicking on the column headers)
In a research paper by Dixon and Robinson (1998), it is asserted that “teams tend to score more or fewer goals depending on the current score”. They concluded that “if the away team is leading then rates [of scoring] for both home and away teams tend to increase”.
So is the game more open when one team breaks the deadlock?
Data from last season in the Premier League shows that there was an average of 2.81 goals per game. In 2010/2011 it was 2.80, and during the 2009/2010 season the average goals per game stood at 2.77. We can see, then, that the figures are reasonably consistent.
We now compare this to the average number of goals scored in matches where the away team is winning 0-1 at half time in order to find out if there is any difference in the full-time-home-team-come-back goal average compared to the season average.
The average number of goals over all games where the home team is 0-1 down at half-time in the sample is 2.63, with 46% of games ending over 2.5 goals. Just 15% of game did not have a further goal in the second half.
I recognize that my sample is small. But I suggest that in 0-1 half-time situati0ns (and we need to further investigate the effect, if any, of the time of the first goal) despite the expectation of a further goal (at least amongst the home fans), it appears that there is no discernible increase in goal rate.
I have looked at the last three full seasons’ data (2009-2011) as well as the current season, and have seen a divergence from the Dixon and Robinson research, for Liverpool, Everton, and Sunderland, in particular.
- Liverpool: 2.11 Average- 77% of 0-1 Half Time games at home under 2.5 goals Full Time
- Everton: 2.00 Average- 63% of 0-1 Half Time games at home under 2.5 goals Full Time
- Sunderland: 2.1 Average- 66% of 0-1 Half Time games at home under 2.5 goals Full Time
Dixon and Robinson’s sample size was 4000 games, so my sample is very small by comparison.
But at this stage unless the trend breaks and reverts to the average, I would suggest that backing under 2.5 goals when Liverpool, Everton, or Sunderland are losing 1-0 at home at HT. The price will be around 1.9.
If you were betting on Liverpool vs Newcastle, for example, then you would have probably:
- Backed Liverpool before the game and lost
- Backed over 2.5 goals at some point during the game and lost
As has been pointed out, not all games will fit into a predictive model, and as long as the predictive model results in long-term success then that is all that matters.
There are many people who discard a strategy after a few losses. But, unless the strategy fails over a long time period, then this does not necessarily mean that the strategy itself is a failure.
People should continue backing Fulham and Spurs to win when they are losing 0-1 at Half Time, and Liverpool, Everton, and Sunderland under 2.5 goals at 0-1 Half Time, as the punters will be getting value (unless the trend – and I will call it only a ‘trend’, as the sample size is small – ends). I note that Liverpool did draw 4-4 with Arsenal in 2008 when they were losing 0-1 at HT but, as has been documented by previous bloggers, there will always be outliers, exceptions, and counter-examples.
I question whether sample size matters. If we look at Portsmouth as an example there is no point looking at their Premier League data to work out the expectation of them winning a game this season.
Rather, the key may be to understand the difference between trends and academic research on football. If a trend starts then why not follow it even if it cannot be explained by quantitative modelling.
With all the data available I think that the time has come to analyze academic papers and determine whether they have any substance in their conclusions.
I am certainly going to look at the time of the second goal (first goal by the other team) in my next post to see if the time of the first goal has any effect on the next goal.
The data has now been open to everyone, so it is a great time to engage in statistical football analysis. If two Premier League teams play the same style one season to the next then there is a possibility that you will see consistent data. Fixtures which pit Stoke against Arsenal, for example, offer a good opportunity to do just this.
- This season at Stoke, Arsenal had 17 Shots , attempted 159 final third passes, had 8 shots in the box, and 9 out of the box, and attempted 45 long balls.
- Last season at Stoke, Arsenal had 17 Shots , attempted 154 final third passes, had 9 shots in the box, and 8 out of the box, and attempted 46 long balls.
Is the above random or can it be explained?
- Good Read
Categories: Arsenal (NN), Aston Villa, Betting Tips, Chelsea, EPL Index Featured Article, EPL Index Statistical Comparisons, Everton, Fulham, Liverpool, Manchester City, Manchester Utd, Newcastle Utd, Norwich City, QPR, Reading, Southampton, Stoke City, Sunderland, Swansea City, Tottenham Hotspur, West Bromwich Albion, West Ham United, Wigan
This article has had 2,333 Views