Fulham has had an up and down season. They’ve had games where they look like world beaters and games where they look like they have no idea what they are doing with the ball. The one constant is their desire to possess the ball. The change from game to game is what they do once they have it.
We’ve looked at numbers before, trying to explain Fulham’s home and road form. This time we’re going to take some ideas from game theory, examining possession and passing, to see how they affect the outcomes of Fulham matches.
What is Game Theory?
Game Theory is "the study of mathematical models of conflict and cooperation between intelligent rational decision-makers." Confused yet? In the simplest terms possible, Game Theory is a way to mathematically model the outcomes of games. It can be used for much more than that, but we’ll limit ourselves to that idea.
There are all sorts of games that can be modeled. Some simple games like tic tac toe, can be solved. Other complex games like poker or soccer can probably never be solved, but that doesn’t make the models worthless. Game Theory has been used in soccer extensively to study penalty kicks. There are ways to use it to decide how often you should take a short corner. The way we’re going to use it is to look at Fulham’s possession and passing statistics.
Is Possession an issue?
First, let's take a look at some of Fulham’s stats for their games so far. We’re going to limit this to games in the Championship.
Fulham is a possession oriented team. They’ve only been out possessed once on the year. The low on the season is 45.1% and the high is 70.6%, with an average of 59%. There has been a lot of talk online that Fulham often possess the ball too much and just kick it around without trying to score. To test this, let’s look at the correlation between possession and expected goals. Is there a pattern we can find?
What you see here is that there isn’t much correlation between how much Fulham have the ball and the goals they are expected to score. There’s even less of a correlation between possession and shots (r^2 = .006). Clearly possession isn’t the major issue.
Let’s now take a look at passing. On average, Fulham are completing 82% of their passes. Their high for the season was 90% and their low was 76%. An interesting note is that the r^2 value between pass completion and the percent of passes classified as long is .74. That’s a lot of correlation. This makes sense, long balls are harder passes. If a higher percentage of your passes are difficult, your total completion should go down. So now, let's take a look at the pass completion rate versus expected goals.
This isn’t a perfect correlation, but it’s much stronger than possession. You can also quickly see Fulham’s best expected goals performances came when they completed fewer of their passes. Their worst performances came in the middle (completing between 80-85%. Correlation of course doesn’t equal causation, but it can give us some things to look into.
Game Theory and passing
Why would a team completing less of its passes lead to success? First, keep in mind that at their worst, Fulham are still completing over 75% of their passes. I’d expect that if Fulham ever completed a really low percentage, their xG would plummet. That leaves us wondering though, what if anything makes 76% the ideal rate?
Remember that the chance of a pass’ success is almost an entirely a function of how difficult it is. A through ball that splits two defenders is a much harder pass than a square ball to an open teammate standing 6 yards away. But which pass has a better chance of leading to a goal?
How does Game Theory fit into this? Well let's take a “simple” soccer example to illustrate how defense and offense are intertwined. We’ll look at penalties. If I kick left, and you dive right, I score and vice versa. If you dive the same way, you stop me. That would mean that I should kick left and right 50% each and you should do the same with your dives. But in reality, it’s easier for me to kick right so I will score more often if I kick right (shots to the left are more likely to miss). Should I kick right 100% of the time? No, because the you can then dive that direction every time. If you do that though, that opens me up to kick left. Even though I’m more likely to miss if I go that direction, if you are diving right 100% of the time, I score every time I shoot left and hit the target. That means the you have to dive left sometimes otherwise I’m always going left. I kick left sometimes to keep you from always going right, you dive left sometimes because I will kick that way sometimes. Eventually you reach an equilibrium. Both of us are maximizing our strategies and there is no more advantage to be gained.
But what does this have to do with passing? The defense has to defend against all the pass types. They have to be looking for long balls. They have to watch out for a through ball or a ball that plays a runner in behind. They have to press to make short passing and playing from the back hard. The pitch is a big place, and you can’t defend everything. If a team is only playing short passes and building from the back, the opponent can bring the defenders higher and press. The fear of a long ball over the top isn’t there. Conversely, if a team is only playing long balls, you can sit deep and win balls with a numerical advantage of defenders. Just like with penalties, you need a mixed strategy. There is an optimum number of passes of each type you should be playing to keep the defenders on their toes. While pass completion rate isn’t the perfect test for this, it is a very good proxy for pass difficulty. We can use it as an approximation instead of diving into all the individual pass types.
Explaining the dip
You may have noticed that dip in the chart before it rises back up at the tail end. What can we make of this? To answer, let’s think about a team like Barcelona. At their tiki taka height, they completed a very high percentage of their passes. No one would ever accuse them of not trying enough hard passes. However, they’re a special case. Most often they had a genuine talent advantage over their opponent. In the Championship, it’s rarely the case that you’ll be that much better than the other squad. There are situations where you are though. When Fulham lined up against Burton for example, they have enough of a talent advantage that they can still generate 1.8 expected goals while completing nearly 90% of the passes. Against most teams, Fulham would never be able to approach that rate. So we’d expect to see the expected goals rebound at very high completion rates due to the talent disparity.
What do we make of this?
Fulham’s struggles don’t appear to be something caused by away versus home form, as we’ve previously discussed. They also aren’t caused by having the ball too much. When Fulham struggle, they appear to simply not be trying enough dangerous passes. The squad need to try some dangerous balls to keep a defense honest. This isn’t a large change. In the most successful games, Fulham are increasing their number of long balls by about 9%. This is only about 20-30 passes a game. This also appears to be something the team is learning. Quotes from the players and coach talk about being a little more direct and the trend seems to be that the team is working towards a completion rate of about 77%.
This is a good sign going forward, and something you should take note of when watching the games. If Fulham are completing a high percentage of their passes and not trying anything dangerous, you should begin to worry. If things are hovering around 77% passes completed, have confidence that the chances will come.