contango/backwardation

Discussions about the testing and simulation of mechanical trading systems using historical data and other methods. Trading Blox Customers should post Trading Blox specific questions in the Customer Support forum.
Post Reply
wonkabar
Full Member
Full Member
Posts: 20
Joined: Wed Sep 01, 2004 5:27 pm

contango/backwardation

Post by wonkabar » Sun Mar 13, 2005 6:39 am

Has anyone explored the impact of rolling (roll yields) on long term systems? For example now CL is in contango and everyone is likley long and when it comes time to roll will pay .60 (at these levels). This market is ususally backwarded and during the same situtation a long would make on the roll.

Along the same line of thought, do people trade months other than the front month for this or similar reasons?

thanks

Jake Carriker
Site Admin
Site Admin
Posts: 1493
Joined: Fri Sep 12, 2003 10:32 am
Location: Austin, Texas

Post by Jake Carriker » Sun Mar 13, 2005 11:11 am

Hi Wonkabar,

I study rolling fairly extensively, but I don't understand your comments about making or losing value on the roll depending on the spread relationship of the old and new contract months.

I don't believe that for my trading systems I make or lose money on a roll (ignoring the effect of possibly different trends between the two contract months for the moment) due to one contract being priced higher or lower than another. I simply close out the trade in the old month and open it in the new month. I might say that I gain value if I can execute the roll at a narrower spread than I calculate between the two contracts, and that I lose value if I execute the roll at a wider spread than I calculate. However, I am not taking any gain or loss on the raw point differential between the two contracts.

Maybe you refer to the tendency of a futures contract to converge with the cash price as expiration nears. If the futures price is above the cash price and the trader is long, this might create a negative bias for the trade. Is this what you mean?

I have explored trading non-front month contracts in markets where there are liquid distant months. CL and ED are two examples. I find that sometimes a continuous contract seems to have a smoother trend component if it is constructed with non near month expirations. I also find that there is more daily volatililty in the further out months. ED is a good example. Check out the distant expirys versus the near month.

Best,
Jake

wonkabar
Full Member
Full Member
Posts: 20
Joined: Wed Sep 01, 2004 5:27 pm

Post by wonkabar » Sun Mar 13, 2005 4:26 pm

Jake,

Thanks for the response. What I mean is say right now if you are long April CL and it is .60 contango to May. When it comes time to roll your long you are paying .60 to keep your long. If the market were in backwardation then then you would be rolling your longs at a lower price. Some markets like gold or the stock indicies are pure contango markets where the value of the roll is known but in the pure commodities the roles fluctuate greatly based on supply and demand as well as non-speculators who have the ability to deliver or take delivery.

Perhaps I am looking at this wrong but I think rolling longs in contango markets takes away from real return and rollings longs in backwardated markets adds to return. (and vice versa for shorts)

I attached the article that got me thinking on this.
Attachments
ssrn-id650923.pdf
(287.5 KiB) Downloaded 1736 times

JAM
Senior Member
Senior Member
Posts: 41
Joined: Wed Nov 17, 2004 4:25 pm
Location: Toronto

Post by JAM » Mon Mar 14, 2005 4:55 pm

Wonkabar, in my testing, there has been nothing to persuade me to trade just those futures that are normally backwardated, rather than in contango, under the long term trend following systems I have considered.

Theoretically, in an upward trending market, wouldn't you expect higher costs for delivery farther in the future, after accounting for storage costs?

Second, your question only directs going long - what about going short? Backwardation would be a net benefit then.

Finally, the amount of slippage in the roll, although not insignificant, is less important than capturing the big move in a long term trend. Would you want to forgo a 10 point move due to a 0.60 roll?

wonkabar
Full Member
Full Member
Posts: 20
Joined: Wed Sep 01, 2004 5:27 pm

Post by wonkabar » Mon Mar 14, 2005 8:55 pm

JAM,

The same argument holds for going short, only in reverse.

I am not talking about forgoing trades based on backwardation or contango. What I am looking at is if using other months may be better in certain market environments. Also, what is the true cost of trading after paying or receiveing roll yields.

I certianly wouldn't want to give up a 10 pt move to save .60 but if the 10 pt move took 12 months and I had to pay .60 to roll each month, the payoff would look a lot different.

Jake Carriker
Site Admin
Site Admin
Posts: 1493
Joined: Fri Sep 12, 2003 10:32 am
Location: Austin, Texas

Post by Jake Carriker » Mon Mar 14, 2005 9:34 pm

Hi Wonkabar,

I still don't think I understand how you lose 60 cents (given the spread you qoute) on the roll.

I read the paper you posted and I notice that the "roll return" they cite does not pertain to rolling contracts from one month to the next, but is a reference to the implied hypothetical return from a contract converging to the cash price over time. Like most things published in academic papers, efforts to capture such market phenomena in live trading accounts may be hazardous to one's financial health. It always amazes me how these types of papers can use passive tense sentence structure, jargon, and every other syntactical and grammatical tool known to man in order to render the topic basically incomprehensible.

I *think* the point of that paper is that buying and holding a passive commodity index might not work if one's intention is to improve the Sharpe Ratio of your efficient frontiered, asset allocated, purely hypothetical portfolio. However, trading the index or the individual components might be a good idea if it were not already proven time and time again by people that study lots of normally distributed assumed, simulated, and hypothetical returns that such a thing is impossible to do profitably.

OK, so I'm having a bit of fun with hyperbole and the ivory tower crowd :) I really don't have anything against the authors of that paper, its just that the highbrow tone and loopy kind of testing procedures almost gaurantee that none of that stuff is in danger of actually being put into practice. Does that one fella (the one who's not the professor) actually trade? It doesn't sound like it.

...back to rolling futures contracts.

Let's take a pretend example. You buy CL at $35 and have a good trade going on when it is time to roll the position. You sell the current position out at $40 and buy the new month at $41. It goes to $45 and you close the trade out. You make $5 per contract on the first trade and $4 per contract on the second trade for a total gain of $9 per contract. Notice that you didn't "give back" any profits when you rolled to the contract that was priced $1 higher. You took $9 out of a $9 move if you now consult your backadjusted chart that splices the two contracts together in the same manner you trade them. Also notice that the $10 move that you *feel* like happened (bought at $35 sold at $45, right) never did. The two contracts simply traded at different price levels, and you got the whole move by moving from one to another when expiration neared.

The backadjusted chart is key here. It kills the gap between the two contracts just like you do when you trade. You don't lose the value of the gap, you simply close out one trade at one price and enter a new trade at another price. I don't think that the move gets "used up" by the gap between two contracts.

It may well be true that backwardation is a trait of a bull market, but that is a different question than whether rolls cost you the price difference between two contracts.

Hope that helps.

Jake

JAM
Senior Member
Senior Member
Posts: 41
Joined: Wed Nov 17, 2004 4:25 pm
Location: Toronto

Post by JAM » Tue Mar 15, 2005 9:32 am

Wonkabar, I think Jake illustrates this well - although it appears you may be missing part of the move due to the role (ie, as in Jake's example, making $9 instead of $10 on the trade), in reality you are not - if you want to delve deeper, you could also run a comparison of owning the physical commodity and paying the storage cost versus dealding with the the price differnentials on the futures contracts.

Further, in trading, although you do want to maximize your profit from the trade, looking at your hypothetical buy price less sell price profit, rolls are going to be either a negative or positive impact.

However, am I right in thinking the key part of your question is has anyone optimized which contract month to choose when an entry signal appears, taking into account variations in price differentials between contracts? If so, I haven't done so yet...

kianti
Roundtable Knight
Roundtable Knight
Posts: 335
Joined: Fri May 02, 2003 6:10 am
Location: Florence - Italy

Post by kianti » Tue Mar 15, 2005 12:44 pm

Why do we Avoid the Nearby Oil Contract for Trading?
JWH Monthly Commentary September 2004

A recent client meeting focused on trading the crude oil market.

The client asked what contract we traded, and we stated that we avoided the nearby contract. He
looked at us somewhat puzzled; "Isn't that where you find all of the market's liquidity?"

That is true, but there are often considerations besides liquidity that determine our choice of futures contracts to trade. For us, the signal-to-noise ratio is more important, and we believe that there is a clearer signal on the long-term direction of any commodity in the more distant contract months.
??
Futures markets also tend to have a characteristic called normal backwardation, whereby the futures price is often below the cash spot price. The size of backwardation can be significant. Oil is not a carry market whereby the futures price is just the carrying cost of the current spot price. This backwardation (inverted or negative carry) is associated with two factors.
One factor is the risk premium associated with the compensation for long speculators to take positions from short hedgers. This premium will change with market conditions.
The second factor is the convenience yield associated with holding the cash commodity. In the case of oil, this convenience yield is significant and volatile.
Due to the combination of a highly volatile risk premium and a volatile convenience yield, we do not believe it is appropriate to trade the nearby contract. The nearby contract will actually have more noise or volatility as it responds to price pressure, which is unrelated to longer-term trends. Because we are interested in exploiting the longer-term trends, we do not want to be stopped out of our positions based on false signals caused by changes in the convenience yeild or risk premium. Trading in futures markets is more than just finding a model; it requires an understanding of the market dynamics so that we are able to maximize the performance of any model. In this case, the choice of time to maturity has an appreciable impact on model performance because of the market dynamics and structure.

Jake Carriker
Site Admin
Site Admin
Posts: 1493
Joined: Fri Sep 12, 2003 10:32 am
Location: Austin, Texas

Post by Jake Carriker » Tue Mar 15, 2005 12:47 pm

Kianti,

Yep.

Jake

SPW
Contributor
Contributor
Posts: 2
Joined: Sat May 13, 2006 3:27 am

Post by SPW » Sun May 14, 2006 7:35 am

Hello everyone,

I'm new to this forum and in fact I'm writing a thesis about the optimization of asset allocation using fully collateralized commodity futures. I'm trying to understand as well as possible where returns of futures contracts come from.

As far as rolling is concerned I believe it should have no direct influence on returns. If you sell a contract at 50 and roll over to one at 40 that's not a return. What counts is where you initiate the position and where you finally sell it. Comparing the sell price of one contract with the buy price of another one would be like selling a blue chip with high price and buying a penny-stock instead and assuming you had a great return as you now own much more shares?

What's quite difficult to understand is how market backwardation and normal backwardation are related. Commodities that are typically in (market) backwardation had stronger returns than those in contango. But as I argued in the previous paragraph you should not be paid an insurance premium for market backwardation. Normal backwardation leeds to an insurance premium, not market backwardation. It should seem that commodities that are in market backwardation are often in normal backwardation too? (How else could one explain that commodities in market backwardation had historically higher returns than those in contango? So expected future spot prices could be similar to current spot prices...) What do you think?

I want to pass to a different point now.
Often one says that an investment in a commodity index is an interesting investment to benefit from rising commodity spot prices. But is there no concensus that one day crude oil reserves will be exhausted and that prices will be extremely high. My question is: can I expect to benefit from these rising spot prices in the long run (if I am long futures contracts and roll all of the time) or are such expected increases in prices priced into futures contracts.

Hopefully you have some ideas.

Old European
Roundtable Knight
Roundtable Knight
Posts: 144
Joined: Mon Apr 19, 2004 3:04 pm
Location: Old Europe

Post by Old European » Sun May 14, 2006 11:12 am

SPW,

If I understand you correctly, you are talking about taking fully collateralized long positions in commodity futures.

I'm afraid that I don't agree with you that systematically rolling these positions doesn't contribute (positively or negatively) to the return.

In fact the so-called roll yield is a very significant positive contributor to the overall return (besides the T-bill yield and the return on the front month futures position), when one is systematically long the front month in crude oil for example (a market that is most of the time in backwardation) and rolls over when the front month changes.

The oil market apparently is prepared to pay a premium for cash oil availability and delivery. And the long-term holder of long positions in oil is in this way renumerated for the risk he is taking (volatility of the market).

A lot of academic research has been done on this subject. Goldman also published a series of papers on this subject when they launched their Goldman Sachs Commodity Index about 10 years ago. I'm sure that you can find several of these sudies on the internet.

Cheers,

Old European

AFJ Garner
Roundtable Knight
Roundtable Knight
Posts: 2040
Joined: Fri Apr 25, 2003 3:33 pm
Location: London
Contact:

Post by AFJ Garner » Sun May 14, 2006 11:28 am

My big "aha" moment on contango/backwardation came when studying concatenated contracts and discovering why markets such as Crude produced a back adjusted contract that goes negative over time.

SPW
Contributor
Contributor
Posts: 2
Joined: Sat May 13, 2006 3:27 am

Post by SPW » Sun May 14, 2006 11:44 am

Thanks Old European for your answers.
Old European wrote:If I understand you correctly, you are talking about taking fully collateralized long positions in commodity futures.
(...)
In fact the so-called roll yield is a very significant positive contributor to the overall return (besides the T-bill yield and the return on the front month futures position)
Yes, I am talking about taking fully collateralized long positions in commodity futures.
I believe that explaining returns as the sum of spot return + roll yield + collateral yield (as is often done) is very dirty pedagogy. It looks very easy, but I think reality is much more complicated. I know most of the academic research you talked about, and there is not really a concensus about where returns come from. Many authors seem to confuse market backwardation with normal backwardation, which seems to me as the principal reason why they find contradictory results.
One of the most influential papers has been the one by Gorton and Rouwenhorst (Facts and Fantasies about Commodity Futures). They too believe that the rolling is not a source of return, but that the return comes (at least partly) from an insurance premium (that is positive when markets are in normal backwardation). That's what lead me to saying that market backwardation should not lead to returns and that if we do observe that market backwardated futures have higher returns than contangoed futures, it should be because there is a positive relationship between normal backwardation and market backwardation.

cully
Full Member
Full Member
Posts: 11
Joined: Sat Mar 03, 2007 11:28 pm

Post by cully » Thu Aug 12, 2010 4:24 pm

Long-only commodity ETFS have been underperformers relative to the underlying due to cantango/backwardation. This new long-only ETF in taking an active approach to dealing with this source of underperformance. Also includes link to the Gorton and Rouwenhorst article cited above. Any thoughts?


http://shar.es/004IU

Moto moto
Roundtable Knight
Roundtable Knight
Posts: 427
Joined: Mon Jun 01, 2009 4:12 am
Location: once again in the UK

Post by Moto moto » Fri Aug 13, 2010 5:31 am

as you asked for thoughts and this is topical given other discussions/threads regards ETFs and why people use them.

and as I am an old skeptic....
This quote was what I found interesting....

"SummerHaven’s Director of Marketing and Investor Relations Kurt Nelson said he hopes investors recognize that USCI represents something new in the commodities space, which would give the fund the asset-gathering advantages over competitors that have historically gone to first innovators in the ETF industry.

“Our hope is that people look at us and say: ‘You’re not making a small change, you’re actually doing something first.’ And we’d get credit for that. We’ll find out over the next few weeks and months,â€

rgd
Roundtable Knight
Roundtable Knight
Posts: 396
Joined: Thu Aug 10, 2006 5:21 pm

Post by rgd » Mon Aug 16, 2010 1:58 pm

Commodity ETFs underperform the spot price because the spot price does not reflect the cost of owning commodities either physically, or through the conrtol and rolling of a futures contract. In this regard, they do a good job.

The question to ask is: Why has someone come up with an "actively managed", fully invested, long only commodity index? For the same reason tactical investment models have become popular... because they outperform other approaches (or have recently).

If you were to search for an approach that would let you avoid bear markets in individual commodities, and the heavy, debilitating contango of energy markets, then have a commodity index which avoids those scenarios. Backtest this approach, tell everyone why this approach is superior (with the benefit of hindsight), sell the model, then market the heck out of it.

Of course, in a few years someone else will come out with an even better index that avoids scenarios that you lacked the foresight to anticipate. All of this is a walk across the continuum of passive to active management. After some time, they might discover that active management suits commodities well since both the cost of ownership, and a highly cyclical nature make them poor buy and hold investments for extended periods of time.

cully
Full Member
Full Member
Posts: 11
Joined: Sat Mar 03, 2007 11:28 pm

Post by cully » Wed Aug 25, 2010 12:09 pm

A performance comparison (link below) for the Summerhaven Index with competing commodity indexes. It would appear backtesting indicates potential out-performance.

http://secure.elabs10.com/functions/mes ... c0407b1e5b

Post Reply