How many points for CL qualification?

One of the perennial discussions we hear near the end of the Premier League season is “What is the point threshold that guarantees a team is safe from relegation”. Right now, as we already know two of the three teams to compete in the Championship next year (well, ok, Middlesbrough still have a mathematical chance), it is the battle for a CL spot that is way more interesting. Chelsea is assured a top four spot as practically are Spurs (97.5% chance even if they lose all remaining matches), leaving two places for the challengers Liverpool, Man City, Man United, and Arsenal.

Before we go on to answer the question “How many points secure a CL spot”, we need to make clear that, theoretically, the answer is different after every match day is played. In a more balanced season, where teams closer to the top lose often to teams lower, we expect a lower threshold. In a season where five or six teams win more often, the point threshold for CL qualification is expected to be higher. That is, if at the beginning of a season we can make the statement “70 points send you to next season’s CL with 70% probability” it doesn’t mean that the same is true after half of the matches are already played.

First of all we take a look at the number of points of the top four teams since 95/96 (first PL season with 20 participants).

We see that there is an upwards trend in the points of the top four teams over time. This suggests that the gap between stronger and weaker teams is growing. In order to have a more rigorous analysis we should take this trend into account.

Now to the main question: What is the probability of qualifying for the CL for a given amount of points earned?

We will use two different methods to estimate the probabilities in question. The first is based on the final tables of the past 21 PL seasons. We don’t want to go deep into technical details, but just briefly, for a given amount of points X, we look at how often a team with X points made it into the top four. If there is more than one team with X points and not all of those teams either succeeded or failed, we count the season as partially successful (for an X-point team). If there are no teams with X points, we look at where an X-point team would have finished in the final table. Here’s what we get.

So, it looks like, based on the past, with 66 points a team has about 48% chance to qualify for next season’s CL. If we take into account the trend seen in the first plot, we tend to believe that the chances are at most that much. 79 points almost certainly guarantee a CL spot. But even with as little as 61 points there seems to be a realistic chance (17%).

The other method to tackle this, the parametric method, is based on simulating PL seasons using a model which associates different strengths to different teams.  The way to do this is to take the team strengths and, allowing for some fluctuation of them along the season, simulate full PL seasons of matches. Then we can treat the simulated seasons like we treated the past seasons in the previous method and estimate the conditional probabilities in question. This second method has some advantages and some disadvantages: One of the disadvantages is that it is parametric, i.e. it assumes that the PL match results are generated according the underlying model. This is always an assumption in statistics and it can deviate from the truth. A big advantage is that it allows us to simulate only a part of the season, i.e. we can take the PL table as it stands right now and simulate only the remaining matches. Additionally it allows us to avoid the trend problem of the other approach, if the team strength estimates are current.  In the chart below you can see what were the CL qualification chances (given points earned at the end of season) before the season started, and what these chances are now (again, given points earned at the end of the season).

First of all we see that the CL qualification chances at the start of the season (blue line) look a bit different in this plot compared to the previous one. One of the main reasons that drives the differences is that the previous plot is a blend of the sets of team strengths at the start of the previous 20 seasons, whereas the last plot is based solely on the strengths of the 20 teams of the current PL season. Interesting is also the difference between the blue and red lines in the plot above. As the current season is nearly over, there is a much smaller range of final points that can take a team to next season’s CL, compared to the situation at the start of the season.  We also learn that this season the bar is higher than we would have expected.  At the start of the season 70 points would have got you more than a 50% chance of CL qualification.  Now, 70 points only has a 5% chance of being enough.

So, how many points do you need to make it to the CL?  It depends when you ask.  At this moment 73 points gets you a 51% chance, and 76 points a more than 95% chance.

Africa Cup of Nations: Update

We’re now half way through the Africa Cup of Nations and it has gone relatively to plan so far. The big 4, Ivory Coast, Nigeria, Ghana and South Africa are all through to the quarter -finals with only Nigeria unable to top their group. The North African teams have been something of a disappointment with Morocco, Tunisia and Algeria all knocked out.

Continue reading

Europa: The other competition

The Champions League knockout stage might have the glamour of Manchester United v Real Madrid, but what of its little brother (whatever AVB might say), the Europa League? We decided to look at our predictions for the Europa League knockout draw.

There are four English teams in the Europa league, Premier League position in brackets: Chelsea (4), Liverpool (8), Newcastle(15) and Tottenham (3).  This is double the amount that have made it to this stage in the previous two years. But how much further can they go?

Continue reading

Are Bayern Munich at Home in the UCL Final?

Seems a silly question, right? The final will be at Bayern’s home stadium. But it won’t be full of Bayern’s fans, as it is when they usually play. So are they really going to enjoy the full effects of home advantage?

Home European Cup Finals

There are previous instances of European club finals being played at one of the teams’ home grounds: Madrid won the European Cup “at home” in 56/57, Inter won in 64/65 and Roma lost in 83/84. In the UEFA Cup, Feyenoord won in 01/02, Sporting Lisbon lost in 04/05 and finally in the European Cup Winners Cup Barcelona won in 81/82. Thanks Wikipedia!

So four wins and two losses, that’s exactly what you’d expect with full home advantage in effect and two teams of equal ability. But most of these matches are far in the past: maybe the home teams were strong favourites and maybe the stadia all those years ago really were full of home fans. Is there anything else that can help us?

Home Domestic European Cup Finals

I was surprised to find (thanks again, Wikipedia) that French, Norwegian, Danish and Swedish have been played at club grounds, and there have been many instances of cup finals “at home”.

Continue reading

Some Graphic Illustration of This Season’s Premier League

This season’s Premier League has not let us down. With most teams having just four league games left, there is still all to play for at top of the table for the Manchester teams. At the other end of the table four teams are battling it out to stay in the top division and of course, there is also race for the fourth Champions League spot. I was interested in seeing how each team has featured in those battles. Drawing charts is usually the best way… Continue reading

How To Increase Arsenal’s Title Chance

After the weekend’s Premier League matches, I am pleased to find Arsenal sitting 3rd in the table. In fact they have won seven Premier League matches in a row and are now three points above Tottenham. A friend pointed out (probably sarcastically) that Arsenal might be late title challenger. I stared at the table for as long as I could but the gap of 15 points between them and the top didn’t reduce. I then looked at our predictions published on Castrol.com, they actually have a 0.01% chance to win the title as I found after digging very deep into the decimal places. But I wasn’t going to give up.

Continue reading

Premier League Goal Glut – What Goal Glut?!

What an exciting start to the season. At nearly three goals per game, there have been plenty of great matches. Outlandish score-lines such as 8-2, 1-6 and 3-5 have led commentators to scratch their heads.

The Guardian asks whether it’s money spent on strikers, and Yahoo cites improved attacks and failures at the back. But is the goal glut real?

Short Term Variation

The power of statistics is that it accounts for short term variation. When information is limited, we should be wary about drawing conclusions from our observations.

So when we get an “unusual” result, such as 2.98 goals per game so far this season, we can use statistics to tell us if that really is a surprise, or if it’s to be expected – after all, we’ve only seen 99 games.

How to Predict Goals Scored

We use our team strength model to predict Premier League matches. Total goals is something we can predict. At the start of the season, for example, we predicted Liverpool to score 1.69 and concede 1.02 goals at home to average Premier League opposition.

Continue reading

On the Uselessness of Draws

How many times does a manager or a player claim “a draw was a good result?” Whether this is justifying the result after the fact (i.e. if you’ve been 3-0 down a draw is certainly a good result), or whether the team went out to play for a draw, I decided to see how often a draw really was a good result given the pre-match predictions.

Who’s Happy With a Draw?

A draw is better than a loss and the one league point it brings is some comfort to fans, but more often it is two points lost rather than one point gained. Consider a match between evenly matched sides. In that case the home team has about a 50% chance of victory, and the away team about a 25% chance. Continue reading

Analysing Major League Soccer

We’ve been working with Castrol and MLS since March to deliver the MLS Castrol Index. Beyond player ratings, we’ll be predicting the MLS Cup Playoffs once the regular season is finished. So we’ve been taking a look at MLS teams and the structure of the league…

Team Strengths

We applied our team strength model, using weighted historical full time scores to rate teams, to MLS. Who’s looking good this season? Seattle is the standout team in terms of goal scoring threat – we rate them as scoring 1.6 goals on neutral territory against the average MLS opponent. After Seattle, the next best teams are remarkably evenly matched, with Red Bull New York in second scoring only 0.2 more goals against the average opponent than Columbus Crew, ranked 12th. MLS Teams Scoring and Conceding rates Continue reading

Let Spurs Entertain You

There’s clearly more of an emphasis in sport on winning than there is on entertaining, but we can turn our football modelling techniques towards a different target: goals per game.  What you really want is for your team to be the best, but if you think about things from a neutral’s perspective the total number of goals per game might be more interesting.  So who, as a neutral, should you go to see?

Continue reading