Sunday, April 04, 2010

Scott Armstrong losing the skewed bet his side set up at Intrade

(Greetings, Andrew Bolt readers!  Please let Andrew know that I'd love to bet him personally over global warming, and my bet offers are here.  I'd also happily bet him even odds over the Armstrong bet offer discussed below.  You might also be interested in a 2006 post I wrote on how to cook Andrew Bolt, as he now seems to be using the same excuse, arguing the current warming isn't part of a long term trend.)

The last we heard from climate denialist Scott Armstrong was several years ago when RealClimate took apart his convoluted offer to "bet" in some form against Al Gore over whether climate models were accurate, where the offer was only against certain fixed locations and while not allowing the climate models to operate properly.

Unfortunately for Armstrong, someone on his side of the fence took the denialism seriously and tried to apply it in a clear format.  Crandles alerted me to some climate- and weather-related predictions at the Intrade betting market (you have click on "Markets" and then click through the left-side menu), and one of them is somewhat inappropriately titled "Climate Change Prediction: Al Gore v Scott Armstrong. Armstrong to Win"

Details are at the "Contract Specific Rules".  Instead of the convoluted original bet, they simply want to compare 2007 to 2010.  A period that short throws a lot of random noise into the analysis, but that only slightly benefits someone who's making a bad prediction.  They've also selected UAH satellite data out of all the half-dozen possible surface and satellite datasets, the one that has shown the least amount of increase, another way to slightly skew the odds.

Most ridiculous though is how they make the comparative predictions for purposes of settling the bet, which is worth quoting:

  • Armstrong forecasts no change in global mean temperature over the next ten years. He predicts the 2010 annual global mean temperature will be 0.284, the same as 2007.
  • Gore has not made an official forecast. He did, however, imply that the global mean temperature would increase at a rapid rate - presumably at least as great as the IPCC's 1992 projection of 0.03°C-per-year. Therefore the IPCC's 1992 projection will to be taken as Gore's forecast. This forecast predicts the 2010 annual global mean temperature anomaly to be 0.374 (0.284+[3*0.03])

Really, Gore implied that?  A citation would be useful, especially one with a date attached.  Actually, two citations are needed since I don't think the IPCC made short-term predictions in 1992 or otherwise indicated the temperature rise would be stable over the next century.  In 2001, the IPCC predicted .15 to .2C rise in the first few decades, and in 2007 it predicted .2C in the first decades.  Armstrong supporters were either unaware of the subsequent 15 years of climatology, or they were purposefully trying to skew the bet - take your pick.

But the irony stems from 2010 shaping up to be a hot year so far even according to UAH, with both January and February over .6C above the the mean.  Armstrong fans lose if 2010 is above .328C.  Unsurprisingly, the bet last traded at 30, or more than 3:1 against Armstrong.  Funny how there are no updates at the denialist website since January data came online.

What would be far more interesting is an accurate Armstrong v. Gore bet, or maybe more appropriately titled Armstrong v. the IPCC.  Make it a ten year bet instead of the tiny, three-year period, and center it halfway between no warming and the IPCC prediction of .2C per decade.  With Armstrong in bad shape in the current unfair bet, his side should do even worse in a fair bet.  Might be worth suggesting to Intrade, though, as something that could accomplish the prediction that James Annan called for several years ago.

UPDATE:  March has come in even warmer, at .65C.  Armstrong's side is looking very embarrassed now.

UPDATE 2:  Intrade now bidding at 20, or 5:1 against Armstrong's side (UPDATE UPDATED:  per the comments, I screwed up the ratio which is actually 4:1 against at price of 20).  I think a more interesting prediction would be when the denialist website that focuses on this bet, quietly folds up shop.


  1. FWIW, the site has been updated; the chart in the sidebar now goes up to March 2010.

  2. Not at the site, they're still using the old chart.

  3. Anonymous3:51 PM

    >"Intrade now bidding at 20, or 5:1"

    The buy sell range is currently 20 to 25. The lowest I have seen is a sell order at 20. Maybe that was the situation at the time update 2 was posted. Only posting one side of the range might be considered to be another way of skewing the impression of the situation.

    Also, to avoid any misinterpretation an intrade price of 20 implies a 20% probabilitity if you ignore fees. This is digital odds of 5:1 but would more normally be quoted as 4:1 if you are used to traditional odds - ie if you win you get back stake and 4 * stake.
    (Intrades system of just having a price that can be interpreted as percentage probability seems much easier to me.


  4. You're right about the ratio being 4:1, I've updated the post.

    The quoted price was 20 at the time of Update 2, which I assume was the agreed price of the last trade when I checked it.


Note: Only a member of this blog may post a comment.