There’s a received wisdom that says you need a settled team to get good results. Here’s Will Greenwood talking about the England team in the 2015 Rugby World Cup:
“But the most important thing right now – and I am not inside the England camp so I can only speculate on the mood within it – is stability. I cannot emphasise that enough. Chopping and changing helps no one.”
Google “a settled team is a winning team” and you get pages of similar appeals to find a settled line up, from all sorts of well meaning experts and supporters.
If we accept these philosophers’ statements as true, we’re accepting 2 things:
- It’s true – teams do better when they have a settled line up
- We’ve got cause and effect the right way around, i.e. cause=settled team, effect=better results
Let’s test it shall we?
Do Settled Teams Get Better Results?
We can’t do this test by looking at dozens of teams, and seeing whether the ones that have more settled line ups do better, because that doesn’t help us work out cause and effect. To get to the bottom of whether a settled line up is a winning one, and to understand what causes what, we need to choose one team, and see whether the team performs better when its line up is settled compared to when it makes lots of changes.
For my team, I’ve chosen the Liverpool football team in the 2014-15 Premiership season. I chose Liverpool because they’re better than average and so should be able to string a good run of games together if the manager can get that settled line up. They’re also not so good that they never lose, so they should give us a good range of results to analyse.
I’ve looked at Liverpool’s starting team in every Premiership game and counted how many player changes the manager made from the previous game. The number of player changes ranged from 0 to 4 over the season, with an average of 2.3 player changes from the previous game. Not all player changes are voluntary because of things like suspensions and injuries, and players coming back from suspension and injury, but that’s all part of the hurly burly of achieving our target of a settled line up.
Now let’s look at the average number of points Liverpool got when they had a more settled line up than average (0, 1 or 2 player changes from the previous game), versus when they a less settled line up (3 or 4 player changes).
Yes, you read the table correctly. Liverpool did no better or worse, whether they made a handful of changes, or if they changed a third of the team. This doesn’t mean having a settled team is irrelevant, but it does mean that other things, say how good the opposition is, are likely much more relevant. So just having a settled line up doesn’t, even for a better than average team like Liverpool, reliably give you better results.
What if we Got Our Cause and Effect the Wrong Way Around?
It looks like our cause and effect theory doesn’t hang together: the team stability doesn’t give us better results. But what if the cause and effect worked the other way around? What if better results caused a more settled team? Let’s look at the data again, remembering that the manager made on average 2.3 player changes after each match. Here’s a chart of how many changes the manager made, depending on whether the previous result was a win, draw or loss.
He’s making 50% more changes when he loses. Given that some changes are enforced by injury or suspension, win lose or draw, then I’d bet he’s making most of his voluntary changes after defeats.
So it seems that good results cause team stability, and bad results cause the manager to tinker a bit more with his line up. The classic stick while you’re ahead, roll the dice again if you’re behind.
Testing the Theory in A Completely Different Environment
Every good European knows that Britain and France are fond cousins with opposite personalities, so if the theory works in France too, then maybe we’re onto something. I also want to test things away from the pressure cooker environment of football. So let’s look at the France rugby team in the 2015 Six Nations competition.
Can you see the pattern? Let’s look first at how results affect team stability:
- Average number of changes after a victory: 1.5
- Average number of changes after a defeat: 6.5 – Sacre Bleu!
So good results are followed by team stability, bad ones by a petite Revolution.
Let’s now look at how team stability affects results:
- Results after making hardly any player changes: won 1, lost 1
- Results after lots of player changes: won 1, lost 1
So whether Phillipe Saint-Andre made lots of changes, or hardly any changes, turned out to be immaterial.
So how do we square all this with our instinct that the teams we all recall as winners are all pretty stable? The Arsenal invincibles? The great Barcelona teams? The world beating Australian and West Indies cricket teams from over the years? Here’s what I conclude. Winners make fewer changes, but they don’t win because they make fewer changes, they make fewer changes because they’re winners.
Leaving aside some armchair philosophy about whether causality exists at all, it seems to me that it feels natural to think that people’s actions (changing the team) cause outcomes (bad results) and it’s not that natural to think the other way around, i.e. that outcomes prompt or cause people’s actions. That natural way of thinking tricks us into getting our cause and effect the wrong way around. But the direction of cause and effect is obvious when you look at the evidence and think about what’s going on.
Makes me wonder about other cause-effects we’ve been accustomed to hearing about. Do great or shocking governments cause national economies to thrive or flounder? Does a world class university produce outstanding graduates? Do a few corrupt old men create a monopolistic, lucrative global sports governing body? Or does it all work the other way around?
 For non football fans, you get 3 points for a win, 1 for a draw and 0 for a defeat.
 What did affect whether France won or lost? They beat both teams that regularly finish below them in the table, and they lost to all 3 teams that regularly finish above them.