Henry Hub or Henry Higgins? How Not to Trade Natural Gas
Thu, Dec 3, 2009 | Jared Woodard
In early September, I tried to take the other side of the case against UNG, the popular natural gas ETF, regarding its perceived failure to track spot natural gas prices (“The Fuss Over Natural Gas“). At the time, I admitted that my defense – roughly, “it isn’t as bad as all that” – was anecdotal and didn’t address any sof the fundamental worries that caused the dislocation in the first place. I also warned that UNG shareholders weren’t being compensated for the weird risks they were taking – i.e., risks that had nothing to do with the fundamentals of the physical commodity.
Three months later, it looks like everyone was right to warn retail investors away from the fund. Since early September, as natural gas futures have recovered much of their 2009 losses, the ETF has just kept chopping and dropping (h/t Maoxian for revisiting the issue and tradingpoints for a nicely annotated chart).
Near the end of Pygmalion, Higgins remarks: “Would the world ever have been made if its maker had been afraid of making trouble? Making life means making trouble.” Investors could be forgiven for wondering whether the issuers of UNG have essentially the same ethos in mind. Instead of getting the price at Henry Hub, the central delivery point for North American natural gas, UNG shareholders have received something more akin to the efforts of a meddling professor attempting to run a social experiment. The analogy isn’t entirely apt – Eliza Doolittle performs admirably, after all, while UNG has done anything but.
Notice the effects this Fall’s dislocation has on the correlation coefficient between the futures and ETF, versus their relationship since UNG’s inception (April 2007):
| Inception | 2009 | |
|---|---|---|
| Correlation | 0.919 | 0.403 |
| R^2 | 0.845 | 0.162 |
Given the steady decline in the ratio of UNG to NG (below),
a viable strategy would have been to short UNG at inception and buy the futures in equal dollar amounts, and hold the positions since then. Log-scaled results for that strategy are below, versus buying and holding the futures alone.
I’m not advising anyone to pursue such a strategy at this stage; for one thing, UNG shares are hard to borrow (although the options look active enough to warrant a synthetic short) and, more importantly, the two vehicles could easily converge from here. But this equity curve suggests that UNG’s current set of problems – the halt on new shares, CFTC eyebrow-raising over position limits, swap market size, etc. – are not the only troubles plaguing the fund. Even if all the regulatory troubles disappear, the smartest and least painful thing for any trader to do who wants to take positions in natural gas would be to simply trade the futures.
I think this lesson generalizes pretty well, in fact: no one should trade a new non-equity ETF unless and until it has proven its ability to avoid massive tracking error.
Tags: etf, futures, natural gas, ng, ung
December 3rd, 2009 at 12:36 pm
Did you take into account the contango of the futures markets? The jumps you see in the NG graph are from changes in the front month contract (e.g. when the september contract expired and october became the front month, a jump of almost 1$ was seen on the charts, but this is simply becuse october was already priced higher). If you shorted UNG and bought futures, you would probably at best break even, because you would have been losing money every time you had to roll your futures contracts. Similarly, if you had just bought long term futures, you probably would have at best broken even. Look at a chart of just the January 2010 futures. You will see the steady decline.
December 3rd, 2009 at 12:46 pm
thanks, the guy i listen to has been loseing for awhile in ung and when news comes out graffles with the incongruity between news and price.
December 3rd, 2009 at 1:15 pm
Matt, good point. I reran the above studies with NG data assuming that the contract is rolled the day prior to expiration, and adjusted on a ratio basis. This removes much of the UNG/NG ratio decline in the second chart above, although the disparity is still present (a decline from 2.11 to 1.8). Similarly, the toy strategy still comes out profitable, which it seems to me shouldn’t be the case if the ETF is replicating properly.
December 3rd, 2009 at 2:42 pm
jared… your strategy will have different results since UNG rolls 14 days prior to expiration over a period of a few days and your strategy rolls in one shot the day before expiration. I think you will find the tracking error to be close to the management fees once you compare apples to apples.
December 4th, 2009 at 11:55 am
Jared, this doesnt work for the above stated reasons. Rolling on the last trading day is very risky btw, so you should use an equivalent, which is rolling in a suitable period, such as GSCI NG: go to bloomberg and do a HS on SPGCNGP vs. UNG and you will see that there is no advantage