As many others, I have been watching Natural Gas (UNG) lately.  I have done a number of posts on my blog analyzing natural gas price and other data.  Most recently I looked at the price ratio of natural gas and crude and developed an ARIMA model looking at the time behavior of that ratio and based on that model I think this is a good time for a pair trade of shorting crude and going long natural gas.

Here is a quick summary of the background, if you want more details you can see my blog post.  To start, here is a chart of the price ratio of natural gas to crude.  This is monthly data going back to 1992.

(I think you can click on the picture for a better view, I downsized it so it won’t get cut off while the image downsizer gets fixed.)  It is on a log scale which is important for all price data but also for a price ratio.  But you will see this price ratio has trended through the years up until about 2002 and then since it has trended down.  But the shorter term swings are more what I am interested in and so I fit an ARIMA model to that dynamic shorter term behavior.  I won’t show the details of the ARIMA model but the next step is to look at the residual error from the ARIMA model.  This basically shows you when the ARIMA model is not fitting the short term behavior well.  That is an indication of the price ratio getting too high or low on a short term basis (kinda like an oversold or overbought reading but it is based on a statistical model and statistics).  Here is what that residual looks like with a Moving Average control chart:

When the residual plots over the upper control limit it is a sell signal and when it plots below the lower control limit it is a buy signal.  So I tried a basic trading system where you buy when the ARIMA residual MA plotted above goes below the lower control limit and then you sell when the MA goes back above the center line.  And vice versa on the top side.  Here is what it looks like trading natural gas since 1992:

(again hopefully you can click the image to see a larger and clearer view).  But as you can see, this really produces some amazing results.  It is just a backtested result so it may be overly optimistic and no slippage of course included.  But since this is a statistics based system with little to no over-optimization, I am still pretty impressed by it.

Taking us to now, this system issued a buy on natural gas in March.  That buy is still open.  This analysis is a monthly price analysis so in reality when I would see a buy or sell on this system, I would then watch the daily charts more closely for a purchase.  Technically this analysis is on the Ratio of natural gas and crude so if it issues a buy you would buy NG and short crude.

Taking a quick look at the UNG and USO charts, I like the idea of being long UNG and short USO:

Trade Execution/Strategy: I will probably do this via DTO, the 2x Crude inverse ETF which would allow me to only use 1/2 the capital and not outright short.  And then buy UNG.  So I would for example buy $10k in UNG and buy $5k in DTO. 

Disclosure: I have not done this yet but probably will.