The key to Magnus Carlsen’s success as a chess grandmaster

For the fifth installment of my series of posts analyzing a data set of over 650,000 chess tournament games ranging back to the 15th century, I wanted to focus in on Magnus Carlsen and try to understand what makes him such an exceptional chess player.

In November 2013, Magnus Carlsen soundly defeated the reigning World Chess Champion, Viswanathan Anand, and added “World Chess Champion” to his wardrobe of prestigious titles that he’s earned in the past decade of playing chess. Many seem to think that it’s Carlsen’s ability to wear his opponents down and win games that’s been the key to his success lately, but a recent analysis of his games since 2001 seems to suggest otherwise.

Anand vs. Carlsen in the 2013 World Chess Championship

Anand vs. Carlsen in the 2013 World Chess Championship

Magnus Carlsen’s meteoric rise to the top started like many of the other chess prodigies. He appeared on the scene in the early 2000s as an ambitious young grandmaster, quickly rising to the ranks of the elite (2750+ Elo rating) within half a decade.

Viswanathan Anand, Vladimir Kramnik, and Veselin Topalov all followed similar trajectories in the 1990s, but eventually stagnated in the sub-2800 territory under the shadow of Garry Kasparov. In 2009, Carlsen seemed to be suffering the fate of the many prodigies before him. Then in 2010, Carlsen made an incredible move: After winning a flurry of chess tournaments, he surpassed Kasparov’s Elo rating and never stopped growing from there. At this rate, Carlsen seems poised to be the first ever chess player to break a 2900 Elo rating.


What has been the key to Carlsen’s success? Below, I charted Carlsen’s win, loss, and draw rates since 2001. Surprisingly, Carlsen doesn’t seem to be winning more games today than he did in 2001. Instead, it appears the key to his success is taking games that he used to consistently lose — especially games as Black — and instead forcing them into a draw.


Interestingly, this trend mimics the evolution of chess game outcomes since 1850: Every year, more games are ending in draws rather than conclusive wins. It seems that the ideal modern grandmaster is better at forcing a draw to prevent a loss than checkmating their opponent.

Dr. Randy Olson is a Senior Data Scientist at the University of Pennsylvania, where he develops state-of-the-art machine learning algorithms with a focus on biomedical applications.

Posted in analysis, data visualization Tagged with: , , ,
  • Pingback: [English Learning Special] – 09.06.2014. | Džecin blog()

  • Would love to see a chart that does a comparison between either Carlsen vs Overall (avg grandmasters) or perhaps a chart that shows how those outcomes have changed over time (didn’t yet click the link to see if that’s already out there)

    This was a nice, quick analysis – and very visual! Thanks for sharing.

    • Good idea! I subset all of the data down to games on or after 1990, and only with both players at >= 2700 Elo rating. Here are the average win/loss/draw percentages:

      Magnus Carlsen (over all years)
      wins as white: 47.375%
      losses as white: 16.25%
      draws as white: 36.375%

      wins as black: 35.4%
      losses as black: 21.11%
      draws as black: 43.49%

      overall wins: 41.42%
      overall losses: 18.67%
      overall draws: 39.91%

      All other players
      wins as white: 29.22%
      losses as white: 19.02%
      draws as white: 51.76%

      wins as black: 18.62%
      losses as black: 29.97%
      draws as black: 51.41%

      overall wins: 23.93%
      overall losses: 24.48%
      overall draws: 51.59%

      So it’s fair to say that Magnus plays better than most grandmasters out there (and always has). The only thing that’s changed since he began is that he draws more games that he previously used to lose.

  • Sergiy Stepanovich

    great data summary. It would be interesting to see/add ELO-opposition.

    Magnus Carlsen faces today tougher opponents than in 2001, 2005… From what year Magnus Carlsen faces constantly strong opponents?

    A quick check at

    For October 2001 three tournaments were calculated. His opponents average rating was: 2288, 2123 and 2219.
    For October 2002: 2349, 2300, 2285, 2274
    For October 2003: 2304, 2344, 2267, 2348, 2366
    October 2004: 2449, 2397, 2482 and 2645 (just two games).
    October 2005: 2427, 2581, 2660
    October 2006: 2655, 2322, 2446, 2441, 2674, 2590
    October 2007: 2731, 2675, 2474
    October 2008: 2707, 2669, 2767
    November 2009: 2762
    November 2010: 2621, 2776
    November 2011: 2772
    November 2012: 2769
    October 2013: 2775
    June 2014: 2760.

    • Here’s the chart you’re looking for. Carlsen’s opponents have indeed gotten better (according to their Elo) over time.

  • Pingback: An Investing Lesson | Mortality Sucks()