Pages

Showing posts with label Nate Silver. Show all posts
Showing posts with label Nate Silver. Show all posts

Tuesday, August 13, 2013

Application of "Signal and The Noise" to Predicting Freight Volumes

I am deep into reading Signal and The Noise by Nate Silver - This is the guy who almost perfectly predicted the outcome of the last election, state by state, while virtually all of the talking heads and big public polling houses go tit all wrong. I have not finished the book yet but so far it is a fascinating read.

So, why discuss this on a transportation, logistics and supply chain blog?  As many of you know, I am a closet forecaster.  I use my data I observe and report on in my Macroeconomic Monday feature to try to determine what will happen in the transportation markets.  I have my ups and downs and so far, however, I would say I have been far more accurate than the official transportation pundits (Magazines which are essentially paid for by the trucking industry, analysts who "cover" the industry but in reality are just trying to push stock prices up.. etc.) who have, for the last few years, reported a dramatic speed up in freight, a dramatic drop off in capacity and a huge inbalance driving rates up.  I am sure they will be right one day but for now, if you had listened to them instead of me three years ago, you would have been paying far higher rates than you should have been.

Nate Silver describes a phenomenon in the book which I think is one of the core reasons why some of my predictions have been just a bit more accurate.  The concept is that of being "Out of Sample".  What this means is people will apply previous history to future results yet they will not realize enough data has changed which causes their examples they are using to not be representative of the current situation.  So, the general belief that when the economy "heats up" there will be a problem with capacity fails to account for:

  1. Growth in intermodal
  2. Smaller packaging and product
  3. Movement of people to cities
  4. Software and collaboration models
  5. 3D printing
  6. The fact that more and more of GDP is not product driven but services and financial driven
And I am sure a lot more.  My point here is that those who just extrapolate previous history to the future are doomed to have a failed prediction - my predictions seem to be a bit better because I am accounting for changes the external environment and accounting for them in my models.  

To be clear, this may and most likely will change however for now I say (as I have for almost two years now) say that capacity / demand is fairly balanced and you should act that way.  In the words of John Maynard Keynes, "When the facts change, I change my mind".  I will keep my eye on the facts and will change my mind but one thing I will continue to work on is making sure I do not succumb to being "out of sample."