.jpg)
.
I wanted to return to looking at more energy prices. I started this back when I was looking at gasoline prices and whether they really increase in the summer? Read this for the answer.
So how much are the various energy prices correlated over time? The below plot shows price data for Crude Oil, Fuel Oil, Gasoline, and Natural Gas back to 1993 (data from EIA):
Of course it is in log scaling. If you don't log scale the prices, then last year prices will look parabolic when in fact most energy prices were just continuing the % appreciation vs. time trend since '98 or so. But if you look at this you will see that Crude, Fuel Oil, and Gasoline all follow quite closely. Natural Gas is the only one that looks somewhat more independent from the rest (it has a mind of its own sometimes). All came crashing down late last year into this year and all but Natural Gas have started to recover this year. Makes me think Natural Gas may be due to catch up to the rest and start moving up soon.
Examining Further the Correlations
Here is a multivariate tool I have pulled out before, the loading plot. I think it conveys quite nicely the correlations between various variables (after a log transformation on the data which is essential for doing correlations correctly on this data):
On this plot, variables that are close to each other are highly correlated. So here it is obvious that Crude, Gasoline, and Fuel Oil are highly correlated (in fact all combinations are greater than 0.97 correlation). That is a very strong correlation. So if you know what crude oil is doing, you pretty much know what the other two are doing. Natural gas is off from the rest some showing it is more independent. Natural gas has some price movements that are unexplained by movements in crude.
This all makes practical sense in that Gasoline and Fuel Oil are direct derivatives of Crude Oil where Natural Gas is related but not a direct derivative. But I did think that with their different demand characteristics that Gasoline and Fuel Oil might act different sometimes but apparently not so much.
What are the Practical Implications?
Basically if one wants to trade, invest, or hedge the general direction of Gasoline or Fuel Oil, they could do that hedge via trading one of the many Crude Oil ETF's and get a nearly perfect match. This actually could be useful to one trying to hedge Fuel Oil costs in the winter because you could hedge using crude oil ETF's rather than directly hedging with Fuel Oil futures or through a Fuel Oil distributor as is usually done. This would open alot more options and possibilities. I will try to have a follow-up post working through the details of how that hedge might work. Similarly if one wanted to hedge gasoline costs, they could do it through a crude oil ETF. With many more to choose from including leveraged ETF's the options would be wider.
A Bit More on the Correlations & Regressions (for the stat-geeks):
Here is the regression fit plot for the log of Fuel Oil vs. the log of Crude Oil:
Remarkably nice fit (won't be as nice if you don't do it in a log scale) with no time lag. This is on monthly price data so there is possibly some shorter time frame dynamics being missed here but I was interested in the longer term movements in price.
Now, here is the regression fit plot for the log of Natural Gas vs. the log of Crude Oil:
Notice how much more error there is in this fit. The R-squared still looks decent but R-squared as a fit metric can sometimes be misleading. Although there is still an overall fit here, there is much to be desired. So there is much more to understanding Natural Gas price than just knowing Crude Oil price.
One important point in all of this, I don't yet see any good time series forecasting model for crude oil itself except for a random walk with a slight upward trend over the long term (unlike the interesting forecasting model I found for gasoline). So I will dig deeper and return to that in the future.
Future posts on this topic...
- In doing this, I found some interesting dynamics in gasoline prices vs. crude prices which are probably related to the seasonality I showed in this past post. But in doing so, probably found a better way to trade the gasoline swings so I will examine that further in a followup post.
- Natural gas prices also showed some interesting short term (within a couple of months) dynamics which may have some forecasting opportunity and trading implications.
- Examine in more detail how one might hedge their personal Fuel Price risk and Gasoline or Natural Gas risk.
- Look at some of the fundamental variables related to energy prices and especially crude oil to see what kind of predictive behavior things like inventories, demand, etc. have on prices of crude and others.
8 comments: