2012 medal projections: The mega-meta comparison

SportsMyriad.com has completed an exhaustive review of the 2012 Olympic events and come up with the following medal projections: USA 113, Russia 85, China 84, Britain 78, Germany 52.

How does this compare to other projections? Glad you asked …

COUNTRYACTUALMYRIADWSJSIINFOJOHNSONTUCK
USA104113108998899103
Russia82858379818267
China88849297946794
Britain65786665634562
Germany44524949606039
Australia35434240403842
Japan38413939353125
France34343132373739
Italy28303025333126
South Korea28232423272929

These projections are:

  • MYRIAD: Mine, based on reading through results of the past four years (especially the last two). Originally done last year; updated this month.
  • WSJ: Wall Street Journal, based on probabilities assigned for each country to medal in each event. And those probabilities are based on recent results and interviews with experts.
  • SI: Sports Illustrated (saw it in print, didn’t see it on the site), based on predictions in each event. (Similar to mine.)
  • INFO: Infostrada/USA TODAY, also based on results but uses “an algorithm.” It’s a bit more detailed than something I attempted at USA TODAY heading into the Athens Games.
  • JOHNSON: Colorado College professor Dan Johnson, who has made projections for several Olympiads using socioeconomic and other factors.
  • TUCK: Emily Williams, using population and economic data along with historic results. She has picked up this formula from Tuck School of Business (Dartmouth) professor Andrew Bernard. (Yes, I’m also picturing Ed Helms.)

Several of these projections, including mine, are also being compared at Top End Sports.

I’ll now do the time-tested self-interview, partially because it’s an old PR tactic and partially because my brain is completely scrambled after doing the update of each event in about 10 days.

So you think your projection will be closest to the final tally?

Actually, no. The Wall Street Journal’s method looks best. The expert interviews should fill in a few holes in the results. Some athletes simply don’t show their best until the Games, and the experts can often name them. And then converting that information to probabilities is a good way to account for the fact that some sports are sure bets and some are crapshoots. (Let’s say I have a little more confidence in the men’s 400 individual medley than I do in most of the taekwondo picks.)

Looks like the USA will run away with it, and Britain will do pretty well, too.

Perhaps, but this might be a flaw of my methodology. It’s simply easier to get information about those two countries than it is about, say, Azerbaijan’s.

That said, I don’t see any way the USA drops all the way from 110 medals to 88, which is what Infostrada/USA TODAY is projecting.

In Britain’s case, when one of their athletes was roughly even with someone else, I picked Britain. Home advantage means something. But realistically, out of maybe 20 such occasions in which I picked the British athlete to shine at home, it’ll happen perhaps 10-15 times. The question is whether British fans will propel 10-15 athletes who normally wouldn’t medal to do so this time around. If that happens, my projection of the high 70s will come true. If not, it’ll be closer to everyone else’s picks.

How about China?

Yeah, how about that? Infostrada/USA TODAY says they’ll lead the medal count. SI says 97. Johnson says 67. I think they’re going to drop off a bit from Beijing. Going sport by sport, I saw several events in which China’s 2008 results were surely influenced by home advantage — athletes medaled at home but didn’t accomplish much in the next couple of world championships.

Why are you doing this?

I’ve done similar things in my years at USA TODAY, and I wanted to give it a try on my own.

Will you do this again for the 2014 Games?

Probably. The Winter Games are much easier to track.

And again for 2016?

I’ll need a nap first.

This entry was posted in medal projections. Bookmark the permalink.

17 Responses to 2012 medal projections: The mega-meta comparison

  1. Alex says:

    The SI medal projections are here: http://sportsillustrated.cnn.com/2012/olympics/wires/07/20/2090.ap.oly.medal.projections.1st.add/index.html?sct=obinsite

    Not sure how to hyperlink in this comments box.

    The WSJ and the Johnson, Tuck previews are fundamentally unsatisfying because they are not done by event. What use is an overall medal count? Why do I care that Azerbaijan will win 10 medals. That does nothing for me. The event-by-event preview helps me to prepare to watch the games and will make the experience of watching the games more pleasurable. The overall medal count preview does not interest me as much. Although when it is an accumulation of event predictions like in your site, I think it is worthwhile as a summary of the detailed predictions.

  2. Alex says:

    Apart from the awful formatting of the SI site, I just realized that it is only the second half of the projections. I can’t find the first half. And the link to the second half is buried at the bottom of the site. It’s like they are embarrassed of their own projections or something.

    But I did notice one thing they have done which you might want to consider in the future. They seem to emphasize 2012 form more (or at least as much) as 2011 WC performance. For instance, in the W LW Double Sculls in rowing, they have NZ first even though they didn’t even medal in 2011. This is because the NZ team (a newly formed team) has the fastest time this year (albeit by hundredths of a second ahead of the Greek team) and they beat the Greek team (again by a small margin) in the only race in which the two teams went head to head (WC in Lucerne – the NZ skipped the first WC regatta and the Greek team skipped the third WC regatta).

    In the 2016 iteration of your picks I think you should go with a weighted model that incorporates 2012 form and 2011 WC performance (I think 70-30 might be a good weight).

  3. Nicky Charles says:

    I’m not a fan of the overall medal count. How can 100 bronze medals be better than 99 gold medals? Although it is not perfect I prefer the gold medal based table that most countries use. Better still I’d like to see a points based table (Gold=3pts, Silver=2pts, Bronze=1pt) which would give a more accurate picture of each NOC’s performance.

    Luciano Barra the Italian NOC member who has been producing these medal projections for years has China 43-29-28[100], USA 35-17-26[78], Russia 30-23-26[79], GB 19-27-16[62]. Unfortunately, I don’t the info to hand re other countries.

  4. Marius says:

    Hello Alex,
    These are not the SI projections (only on print until now).

  5. Alex says:

    Thanks for the info, Marius. You can see how I got confused since the projections were on the SI site. I wonder why they put these AP staffer projections on the site and not the ones they have in their print version. I will have to visit my local pharmacy to take a look at the print issue…

  6. Andrew says:

    Great work Beau. Do you think you can do the same meta-comparison chart, but with gold medals only?

  7. Alex says:

    I tracked down the SI picks in the print version of the magazine. And I must say I am not impressed. I will just give some examples from events I know a little about. In the W LW Double Sculls, they don’t have New Zealand or China medaling even though these two teams along with the Greeks are the most in form (based on Lucerne performance). In M Basketball they have Lithuania getting the bronze medal, even though most experts predict Brazil, Russia or France are bronze medal favorites. In the wrestling, they have Arzoumanidis getting the bronze and he didn’t even qualify for the games. So I think they are relying on 2011 WC performance too much and not enough on 2012 form. I think the AP staffer predictions might be the best because they put more emphasis on 2012 performance. Beau – you should do a comparison of SI, AP, USA Today and your predictions after the games.

  8. Marius says:

    Alex, no problem, will try to get a print issue as well.
    Beau, thanks for the hard work you put into this.

  9. Alex says:

    Brian Cazeneuve’s medal rredictions for all events have been posted to the SI website: http://sportsillustrated.cnn.com/2012/olympics/2012/writers/brian_cazeneuve/07/26/2012-olympics-medal-predictions/index.html

    The formatting is nice.

  10. charles says:

    I like this. Anyway of adding the AP or other ones?. Nate Silver came close for the winter olympics averaging about 9 or 10.

  11. Pingback: Sunday Link-Off: Let The Games Begin | The Lowdown Blog

  12. yumyum says:

    WSJ does look the most scientific. However, if you check the 2010 Winter Olympics predictions they made with the same method, you would find that they were way off (Canada getting 14 gold instead of the 4 predicted, and US getting 37 medals instead of 25 predicted). I think USA will lead in both gold and metal counts, but everything else is gonna be difficult to say for sure.

  13. Pingback: Olympic Daily Glance, Day 10 — I do like Mondays | SportsMyriad

  14. Pingback: The China-USA medal count battle and Olympic projections: Don’t panic! | SportsMyriad

  15. Pingback: 2012 medal projection redux: What we learned | SportsMyriad

  16. Pingback: Predicting the winners | Business Analytics Blog

Leave a Reply