In an earlier post regarding 0-0 draw refunds with Bet365, I raised the question of what proportion of football fixtures end 0-0. I have now found a more reliable figure, based on English Premier League fixtures between 1993 and 2010. At the same time, I have determined the historical frequencies for a range of other scorelines. Hopefully you will find this data useful when making correct score bets.
The data set
I have used raw results data from Football-data.co.uk. I’ve used the English Premier League results as my data set because I presume this is the most popular for Australian punters. I have included everything from the first week of the 1993-94 season through to last weekend’s fixtures in the 2009-10 season.
Methodology
I first pooled every fixture into one large data set and calculated the scoreline statistics over the past 16 1/2 seasons. I then calculated the corresponding statistics for each season individually to see if there are any trends over time. For example, are 0-0 draws becoming more or less frequent over time? Or are they relatively constant?
Results
The first observation I made was that the scoreline statistics fluctuate over time, but with few discernible trends. For example, 7% of fixtures ended 0-1 in 1993-04, and 6% in 2009-10. There was little variation over the seasons in-between. This result justifies the use of older data to create a larger data set.
If I had to find and point out one trend, it appears that the odds of the home team winning have increased over time, but this trend is modest.
The tables below provide my results along with some comments. Table 1 provides win/draw/loss statistics, while Tables 2A and 2B provide scoreline statistics. For each table the top row provides the overall statistics covering all seasons, while the row below it provides the corresponding “fair” betting odds. These odds have been calculated as the reciprocal of the frequency.
Table 1 – Win/draw/loss statistics
| Home Win | Away Win | Draw | |
| —————- | —————- | —————- | —————- |
| All Data | 46% | 27% | 27% |
| —————- | —————- | —————- | —————- |
| Odds | 2.15 | 3.74 | 3.73 |
| —————- | —————- | —————- | —————- |
| 2009-10 | 52% | 22% | 26% |
| 2008-09 | 45% | 29% | 25% |
| 2007-08 | 46% | 27% | 26% |
| 2006-07 | 48% | 26% | 26% |
| 2005-06 | 51% | 29% | 20% |
| 2004-05 | 46% | 26% | 29% |
| 2003-04 | 44% | 28% | 28% |
| 2002-03 | 49% | 27% | 24% |
| 2001-02 | 43% | 30% | 27% |
| 2000-01 | 48% | 25% | 27% |
| 1999-00 | 49% | 27% | 24% |
| 1998-99 | 44% | 25% | 30% |
| 1997-98 | 48% | 27% | 25% |
| 1996-97 | 43% | 26% | 31% |
| 1995-96 | 49% | 25% | 26% |
| 1994-95 | 44% | 27% | 29% |
| 1993-94 | 42% | 28% | 31% |
The probability of a home win is currently equal to the combined probabilities of a draw and a home loss. There appears to be a modest upward trend in the frequency of a home win, suggesting that home advantage is playing a larger role over time.
Table 2A – Scoreline statistics
| 0-0 | 1-0 | 0-1 | 1-1 | 2-0 | 0-2 | 2-1 | 1-2 | 2-2 | |
| —————- | ————- | ————- | ————- | ————- | ————- | ————- | ————- | ————- | ————- |
| All Data | 8.8% | 11.1% | 7.7% | 12.1% | 8.5% | 4.3% | 9.0% | 5.6% | 4.8% |
| —————- | ————- | ————- | ————- | ————- | ————- | ————- | ————- | ————- | ————- |
| Odds | 11.38 | 8.97 | 12.94 | 8.30 | 11.83 | 23.12 | 11.12 | 17.80 | 21.02 |
| —————- | ————- | ————- | ————- | ————- | ————- | ————- | ————- | ————- | ————- |
| 2009-10 | 7% | 9% | 6% | 12% | 11% | 5% | 11% | 3% | 6% |
| 2008-09 | 11% | 11% | 8% | 9% | 9% | 4% | 10% | 7% | 5% |
| 2007-08 | 7% | 11% | 8% | 14% | 9% | 6% | 7% | 5% | 4% |
| 2006-07 | 9% | 11% | 9% | 12% | 8% | 4% | 11% | 5% | 4% |
| 2005-06 | 8% | 14% | 7% | 8% | 11% | 6% | 8% | 7% | 3% |
| 2004-05 | 8% | 11% | 9% | 14% | 9% | 3% | 8% | 4% | 6% |
| 2003-04 | 11% | 9% | 7% | 11% | 7% | 5% | 9% | 7% | 5% |
| 2002-03 | 6% | 13% | 8% | 11% | 9% | 4% | 11% | 5% | 7% |
| 2001-02 | 9% | 9% | 8% | 13% | 8% | 5% | 9% | 5% | 4% |
| 2000-01 | 7% | 11% | 7% | 13% | 11% | 3% | 10% | 6% | 5% |
| 1999-00 | 6% | 10% | 8% | 12% | 9% | 4% | 9% | 6% | 5% |
| 1998-99 | 13% | 10% | 8% | 11% | 8% | 3% | 10% | 6% | 5% |
| 1997-98 | 9% | 13% | 8% | 11% | 6% | 4% | 9% | 4% | 3% |
| 1996-97 | 11% | 10% | 7% | 13% | 8% | 6% | 8% | 6% | 7% |
| 1995-96 | 7% | 16% | 8% | 14% | 7% | 3% | 8% | 4% | 4% |
| 1994-95 | 11% | 9% | 7% | 12% | 8% | 4% | 8% | 6% | 5% |
| 1993-94 | 9% | 11% | 7% | 15% | 7% | 5% | 8% | 7% | 5% |
Table 2B – Scoreline statistics continued
| 3-0 | 3-1 | 3-2 | 0-3 | 1-3 | 2-3 | 3-3 | |
| —————- | ————- | ————- | ————- | ————- | ————- | ————- | ————- |
| All Data | 4.0% | 4.2% | 2.1% | 1.8% | 2.6% | 1.6% | 1.1% |
| —————- | ————- | ————- | ————- | ————- | ————- | ————- | ————- |
| Odds | 24.70 | 23.54 | 47.42 | 56.49 | 37.99 | 63.69 | 94.14 |
| —————- | ————- | ————- | ————- | ————- | ————- | ————- | ————- |
| 2009-10 | 6% | 3% | 2% | 1% | 2% | 2% | 2% |
| 2008-09 | 4% | 4% | 2% | 2% | 3% | 2% | 0% |
| 2007-08 | 3% | 6% | 1% | 2% | 2% | 1% | 1% |
| 2006-07 | 5% | 6% | 2% | 2% | 2% | 1% | 1% |
| 2005-06 | 4% | 4% | 2% | 2% | 3% | 2% | 1% |
| 2004-05 | 4% | 4% | 2% | 1% | 3% | 1% | 1% |
| 2003-04 | 3% | 4% | 3% | 2% | 2% | 1% | 1% |
| 2002-03 | 4% | 4% | 2% | 2% | 2% | 3% | 0% |
| 2001-02 | 3% | 4% | 3% | 2% | 4% | 2% | 1% |
| 2000-01 | 4% | 4% | 2% | 3% | 2% | 1% | 2% |
| 1999-00 | 3% | 4% | 3% | 2% | 2% | 3% | 1% |
| 1998-99 | 4% | 4% | 2% | 2% | 2% | 0% | 1% |
| 1997-98 | 4% | 4% | 2% | 2% | 3% | 2% | 2% |
| 1996-97 | 3% | 5% | 1% | 2% | 3% | 1% | 1% |
| 1995-96 | 5% | 4% | 2% | 2% | 3% | 3% | 1% |
| 1994-95 | 5% | 5% | 2% | 1% | 3% | 2% | 1% |
| 1993-94 | 3% | 3% | 2% | 2% | 3% | 1% | 2% |
I have graphed the above figures and can’t discern any clear trends in the frequencies of individual scorelines. Having said that, based on the upward trend concerning home victories one can conclude that the combined probabilities of the home win scorelines will have increased.
One other point to mention is that the combined probabilities of all of the above scorelines (0-0 to 3-3) is 89.3%.
How to interpret these figures
Obviously, every game is different, and these figures and corresponding odds are for all previous English Premier League fixtures. One wouldn’t expect 2.15 odds on Manchester United to beat Burnley at home, for example. Having said that, I feel these figures are useful for providing benchmarks based on which to adjust your perceived odds. For example, if you anticipate a low scoring affair for a particular fixture, you can take the 8.97 odds for a 1-0 home victory and adjust them downwards. If you foresee a 0-0 draw as being less likely than average, you could adjust the 11.38 odds for a 0-0 draw upward accordingly.
If you have any suggestions or requests for other football data analysis let me know.
Hi
Is it possible to get the spreadsheet on an e-mail ?
Good work !
/ Jacob
Hi Jacob, I’m afarid I don’t have the spreadsheet on me but you can paste the results above into Excel by highlighting the data and selecting Paste -> Paste Special… –> Unicode Text (or Text). I would paste in the data one table at a time.
–
This article is 2 1/2 years old. If you want to recreate the study using more recent data you can obtain the necessary data sets here:
http://www.football-data.co.uk/englandm.php
–
Let me know if you have further questions.