Contradictheory: Goals and expectations – and our GLCs


  • Living
  • Saturday, 30 Jun 2018

So, I love being a football fan. It’s one of those things I can get truly excited about. And nothing gets me more worked up than when I want to discuss a game I just watched with a friend – only to find out they completely disagree with me.

For example, after the June 27 (Malaysian time) Germany-South Korea game, he quoted, “Hard work beats talent when talent doesn’t work hard”, and although I agree with the sentiment, I’m not sure it was necessarily the case here.

While he used catchphrases, I wanted to use facts. Instead of looking if the players looked lazy, I wanted to see what they produced in terms of Expected Goals (ExpG).

ExpG is not a new idea – I first saw it appear about eight to 10 years ago, but it has now crept into the mainstream. It tries to measure how well a team played not by the number of goals they scored but by the quality of the chances they made.

Why is this important? Well, to quote football analytic website 11tegen11.net, expected goals are “the single best predictor for future match outcomes, better than points, goals, shots or shots on target”.

The common example given is the Juventus team in their 2015-16 season. The Italian team struggled to score points in their first 10 games. But they were underperforming, scoring fewer goals than their ExpG score suggested. Juventus ended up running away with the league.

So, what is this magical number? It’s a probability, usually measured as a number between 0 and 1. The larger the number, the more likely the player would have scored from that shot.

For example, a penalty is scored 78.3% of the time. That is, if you attend lots of football games and watch a thousand penalties, you might find they are scored only 783 of those times. So, you assign an ExpG of 0.783.

You can take many things into account, for example, and the ExpG will take different values. It matters if a player is right in front of a goal (ExpG=0.387) or if he’s slamming it from outside the box (ExpG=0.036). Whether it’s a shot by foot or a header.

Whether it comes from a through ball or a cross. And although different people have slightly different models, they all end up with quite similar scores. (All ExpG analysis from here come from 11tegen11’s twitter feed, @11tegen11.)

Assuming the goal attempts are independent (they aren’t always, though), then you can add up the ExpG to find out how many goals a team should have scored during that game and compare it to how many they actually scored. If they are far apart, you can say a team were very lucky (or not).

For example, until just after the 90th minute, the ExpG model has Germany at a little under two goals. South Korea was at about 0.3 goals.

Yet it was scoreless at the time. However, after the penalty was awarded, South Korea’s ExpG jumped by 0.783 to just above 1, and after their second great chance, they ended up the game with a total of 1.76.

So we can conclude that Germany were unlucky not to be leading coming into the 90th minute, and should have been two goals up by then. This two goal cushion would have changed the way they played at the end.

The thing is, despite me offering this sort of very objective analysis, most people still want to argue with my conclusions. Sometimes they even say “nerd” out the side of their mouth, like it’s a bad thing or something.

Of course, bias isn’t (and shouldn’t be) always overcome by data, and there’s plenty of subjectivity in the various ExpG models. For a game of football, it isn’t such a big deal. But I worry a bit more if it’s about national institutions.

It feels like in the last few weeks, the future of various government-linked companies (GLCs) and quasi-government bodies have seemed uncertain.

What concerns me is that they are under the lens simply because they had a particular person on the board, or because they were listed under a certain coloured file – or even just because they were set up by the previous government.

It’s like how the new government say they are maintaining BR1M (1Malaysia People’s Aid), but calling it something else. I really hope they are actually implementing a new programme because Malaysia Baru should be about more than just cosmetic changes.

I completely agree that all these national programmes should be reviewed. But the impression I get is that this is being done to find wrongdoing – for example, the CEO is being paid too much or sang the wrong tune. As a result, if a programme is allowed to carry on, it’s seen as because they have not broken the law or been unethical.

The thing is, this review was a great opportunity to build trust in institutions. Look at each initiative and decide if the value they offer is worth the cost. Keep the ones that are doing well, and let go those that don’t meet the mark.

And tell the world about the success stories – even if it’s something that was set up by the previous government. It shows you are impartial, it shows you reward meritocracy.

So when you eventually put your own guys into positions of authority, the public will feel confident that you’re doing it for the right reasons.

There will be an inquest on Die Mannschaft (as the German team is popularly known) when they return home, and that’s more than reasonable.

But I hope those in charge recognise that all in all, Germany didn’t play that badly and should have probably ended the group with five points instead of three – more than enough to qualify for the next stage.

In his fortnightly column, Contradictheory, mathematician-turned-scriptwriter Dzof Azmi explores the theory that logic is the antithesis of emotion but people need both to make sense of life’s vagaries and contradictions. Write to Dzof at star2@thestar.com.my.

 


Article type: metered
User Type: anonymous web
User Status:
Campaign ID: 7
Cxense type: free
User access status: 3
   

Across The Star Online


Air Pollutant Index

Highest API Readings

    Select State and Location to view the latest API reading

    Source: Department of Environment, Malaysia