Saturday, May 14, 2011

Analyzing the Wager to Death

The great thing about a wager, is even if you are not involved, you can waste time, cheery pick, try crazy stuff to rationalize the outcome, just do about anything and it is perfectly acceptable irrational behavior. So I am analyzing the wager to death!

The stack cluster thing I am using is not predictive, it is just a ballpark look a things. The timing of the wager fairly short term though. It is will the temperature recorded by GISS in 2015 be above or below the average of 2008. The data is noisy, this La Nina will fade, a new El Nino will start and the period of the bet may be in the middle of a new La Nina. It pretty much is a coin toss.

Just to try to get a little better picture, I changed the length of the individual stacked regressions to 24 months. Since I am more concerned with the future, I made a chart of the most recent regressions with overall linear regressions from 2001 and 2008 to the end of the data in March of 2011. The most recent 2 year regressions indicate a down turn. That is of course subject to change. So I added one more regression starting at March 2009 to March 2011.

The main clustering was positive, so temperatures should rise, but with 2015 just around the corner, I biased the stack for the closer term. At least that is what I think. Because the RSS and UAH data for April are already out, I know that there is an up turn as the La Nina fades. The orange 2009 to 2011 line should be a lower limit for the next few years. Hansen's GISS data tends to be the warmest of the surface averages and GISS on occasion makes new adjustments to their data. By 2015, I would not be surprised if GISS adjusts a little to get more in line with the other guys. The bet was against the 2008 average which is already published, so I am thinking 0.5 with remain the target, even if it is adjusted down a touch.

After all this playing around, it still looks like a toss up with the winner being the one guessing the start of the next La Nina. I may waste some more time and download the ENSO to see how that may come into play.

No comments: