Message Font: Serif | Sans-Serif
No. of Recommendations: 2
Jello: Bettors chase Obama's victory probability up to 69%...

I looked at those graphs. Ugh: way too much volatility for me. True public opinion is much more stable than indicated in the posted odds.

I think the hybrid weighting system used by Nate Silver and several other outfits offers a better clue to what is happening. Nate aggregates all state-wide polls together, using an algorithm that appears to go something like this:

1. Maintain a running estimate of the bias of each polling organization, by comparing its moving average to the grand average of all organizations. Subtract the estimated bias from each state-wide poll.

2. Maintain a running estimate of the unreliability (standard deviation) of each polling organization, using its deviation from the grand running average of (unbiased) state-wide polls. Weight the results of each polling organization by the inverse of its unreliability.

3. Apply a further exponentially-decreasing weight to each poll, based on the elapsed time since its results were recorded.

4. The result, for each state, is a current estimate of the weighted unbiased opinion with respect to each major candidate.

5. Modify these results using the local economy (mostly unemployment) in each state. The closer the date of the election, the less important the economy and the more important the polling estimate. The higher the unemployment, the more the incumbent's projected vote is adjusted downward, and vice versa for the challenger.

6. Use the inferred probability distribution for each statewide vote in a simulation of the election, to calculate the probability of a win in the electoral college and an expected number of electoral college votes.

The result is, I believe, about the best that can be achieved with contemporary technology. If it were me doing Nate Silver's job, I would prefer to use a Bayesian method based on the ensemble Kalman filter, but that technique is still largely unknown in political science -- even though it is now the basis for every weather prediction you see on television.

I would also be happier with a separate estimate of political polarization. That would make an interesting graph!

Incidentally, Nate is now quoting a probability of an Obama win at 79% as of this morning. It has been rising slowly but surely ever since the second week of October. A probability of 79% is roughly 3:1 odds, which means the election outcome is still rather uncertain. Obama partisans can be cautiously optimistic, but no more than that.


Print the post  


What was Your Dumbest Investment?
Share it with us -- and learn from others' stories of flubs.
When Life Gives You Lemons
We all have had hardships and made poor decisions. The important thing is how we respond and grow. Read the story of a Fool who started from nothing, and looks to gain everything.
Contact Us
Contact Customer Service and other Fool departments here.
Work for Fools?
Winner of the Washingtonian great places to work, and Glassdoor #1 Company to Work For 2015! Have access to all of TMF's online and email products for FREE, and be paid for your contributions to TMF! Click the link and start your Fool career.