Writing teams off after a poor first
half of the season is as false as failing to see when a team has
Over the second half of a season, good teams are likely to do
worse and bad teams are likely to do better. It is not what many people would
expect, so it is worth knowing if you bet either on individual matches or in
In the first half of the 2003-04 season Carlisle
accumulated only 11 points. Not surprisingly it left them bottom of League Two.
But over the course of the second half they managed to pick up 34 points.
In the first half of the same season
Oxford racked up 48 points. Not surprisingly this sub-total enabled them to
challenge for the leadership of League Two. Over the second half, however,
Oxford gained only a further 23 points. So Carlisle, who had been rock bottom,
compiled almost 50% more points over the second half of the season than Oxford,
who had been flying high.
Admittedly this is an extreme example, but it
illustrates a generally observable truth: teams who have done well in one part
of a season tend to do less well in the next, and vice versa. It is what
statisticians call regression to the mean.
Some teams near the top of a
table will be in the correct position. Others will have overachieved, and they
will then fall away. Overall, these teams will do less well in the second half
of a season than they did in the first. Similarly, some teams near the bottom
will be in the correct position whereas others will have underachieved and
later start to climb back. Overall, these teams will do better in the second
half of a season than they did in the first.
The best estimate of the
number of points a Football League team will collect in the second half of a
season is 54% of the number they collected in the first half plus 46% of the
average for all teams, which is 31.
In other words, a team on 41 points
after the first half of a season could be expected to gain only 36 during the
second, because (41 x 0.54) + (31 x 0.46) = 36. Meanwhile a team who gained 21
points during the first half of a season could be expected to gain 26 during
the second, because (21 x 0.54) + (31 x 0.46) = 26.
Not many people
would anticipate this narrowing of the gap. Do not be surprised if some of the
leaders in each division start to slow down and some of those who were being
left behind start to catch up.