Predicting Carlsen-Nepomniachtchi

Real life doesn't have to make sense

When predicting the outcome of a first-to-four NBA playoff series, often you say the number of games you think it will take, like “Bucks in six (games).” It’s not only a prediction about how long the series will last, but a statement of how you think the teams match up.

4-0 -> Total domination.

4-1 -> Not close, but a little competitive.

4-2 -> Competitive but decisive advantage.

4-3 -> Very closely matched, but someone has to win.

If the final series score is what you predicted, of course you’ll pat yourself on the back for judging the teams accurately. If the final score is different, you’ll probably wonder if you misjudged the relative strength of the teams. But there’s another possibility that’s often missed: you were right about how the teams stack up but still wrong about the outcome.

Clearly, in any given game, the weaker team can win. The same is true of a series, although in general the longer the series, the harder it is to pull off an upset. Say you call the Bucks-Suns series “Bucks in six.” For the sake of simplicity, let’s assume the Bucks are a 55% favorite to win each individual game. Here is the distribution of series outcomes based on that individual game win percentage:

Your take on the relative strength of the teams was right on the money - the Bucks’ advantage in a single game corresponds perfectly with a 4-2 series outcome - but you will still see “Bucks win 4-2” occur in real life less than 20% of the time. The reason is that, while Bucks 4-2 is the most likely outcome, it is still only one of eight possibilities and the others are quite possible as well. Even Suns 4-0 occurs about 4% of the time.

Imagine if someone predicted Bucks in six, the Suns actually won in a sweep, and after the series they said, “I was basically right about the teams, but the Suns got lucky this time.” They would be roasted alive on Twitter, but actually this take is not unreasonable.

Similarly with the United States presidential election in 2016, many people took Trump’s victory to mean that political polls are basically worthless. There are a lot of problems with polling in general and how pollsters estimated the 2016 election specifically, but the underdog winning was not inherently one of them. In their last prediction before the election, FiveThirtyEight gave Trump a 28.6% chance of winning. That means they thought he would win more than one in four times. It’s not shocking, then, that he actually did manage to win: 28% is not 0%.

All of which brings us to the upcoming chess World Championship, where Magnus Carlsen will take on the challenger Ian Nepomniachtchi. It’s pretty clear that Carlsen is the stronger player - he’s currently 55 points higher rated and his tournament finishes have consistently been higher than Nepomniachtchi’s. Nonetheless, it’s close, and Carlsen was not able to win the classical portion of his previous two World Championship matches against Sergey Karjakin and Fabiano Caruana. (In both cases the classical portion ended in a tie and Carlsen won the rapid tiebreak.)

Additionally, Nepomniachtchi has performed well head-to-head in classical games, winning four games to Carlsen’s one, a fact some people make a great deal out of. I don’t put too much stock in the classical head-to-head score (Carlsen is way ahead in the rapid head-to-head), but it does reinforce the overall conclusion that this is likely to be a close match.

The match will be contested over 14 games, an increase over the 12-game format of the previous World Championship. In general a longer match favors the stronger player because it makes it harder for the match to be decided on a fluke. However, as with all classical contests between elite players, most of the games are likely to be drawn. This means in a sense the “real” match is much shorter than it appears. It’s likely to be decided in a few games, which could come down to a few moves, a few moments. In those moments anything could happen.

For instance, in his 2014 World Championship match against Vishwanathan Anand, Carlsen blundered with Kd2?? which Anand could have exploited with Nxe5! gaining a big advantage. Anand missed the move and Carlsen went on to win the game and the match. Had Anand spotted that one move, everything could have been different.

We like to explain things with stories, but there’s not always a grand narrative behind a mistake. Sometimes you just miss something.

In order to win the match, Nepomniachtchi does not have to be better at chess than Carlsen. He doesn’t have to clean up his sometimes dubious time management. He doesn’t have to get over the fact that as kids Magnus surged past him in the rankings after he had dominated their age group for years. He doesn’t have to quit Hearthstone. He just has to win.