Predict oil price from weather information, how ?

Discussion in 'Strategy Development' started by jacksmith, Mar 20, 2009.

  1. I assume that oil price partially depend on weather information. Since future weather is predictable to a large degree, I figure that if we can find the “dependence” of oil price against “weather”, we shall have some edge.

    Then, these are two sequences of data. One sequence is the historical weather information, another is historical oil price.

    What will be the methods to “discover” the dependence of oil price vs. weather ?

    (I heard the terms of Factor Analysis or Principle Component Analysis, but not sure if they will be the right tools for the job).

    Thanks.
     
  2. start by learning the concept of correlation, for starters. I see you never picked up the book on prediction I rec'd to you earlier...
     
  3. I know basic statistics very well, correlation, covariance, joint distribution, etc

    I do not know Factor Analysis, Principle Component Analysis, etc.
     
  4. Nat gas is going to be more closely correlated to weather, if you are interested.
     
  5. It would be tough to build a system on this alone. You would have to also factor seasonality, macroeconomics, and OPEC into your decisions. If you are correct in weather forecasts, but some outside factor overwhelms the weather correlation, you could get your ass handed to you.
     
  6. then how to find dependence of Nat gas and weather ?
     
  7. Your original request does not require FA or PCA; if you understood correlation, as you say you do, you would know that.
    Those tools have nothing to do with your request. Once again, go back and try to understand the meaning of correlation and regression, and you will have the answer to the question you posed.