I think es/cl is strongly corelated nowadays because of current economic situation. But it's not permanent. And if you guys are interesed in finding pairs strongly corelated, why don't you try es/ym or es/nq? They are always strongly corelated. When I back tested pairs, if they don't move rapidly, they converge very well. But they move rapidly, they diverge for very long time. So pair ratio keep going up or down for a long time. How can you guys solve that problem? Especially if you do intraday pair trading?
"When I back tested pairs, if they don't move rapidly, they converge very well. But they move rapidly, they diverge for very long time." I'm not sure what you mean by the term 'moving rapidly'; if you are trading fundamentally and statistically correlated instruments (with >93% positive correlation) and have them weighted correctly for both currency valuations and volatility differences then they should have very good cointegration characteristics. There are hundreds of highly correlated spread combinations that work quite well for futures pairs trading outside the stock index ones you mentioned.
Here is example of yesterday and today's es/cl pair. Yesterday, prices were stable and they converged well but today price movements are largers and they diverged.
bone, actually I have a question about considering volatility. For example, there are a pair saying A and B and the avg ratio is 10. One day A is $100 and B is $1000 and B's volatility is twice of A's. If A is up from 100 to 110 and B is up from 1000 to 1200, then do you consider they are still in equivalent?
Well, I wouldn't recommend or advise a client to trade the S&P 500 versus the Nymex CL contract as a spread position. I would trade February CL versus March CL, or HO vs. CL, or RBOB vs. CL, or Brent Crude vs. CL, but not CL vs. ES. Just because instruments are correlated does not necessarily make them a good spread pair - as you point out, intraday cointegration is a very big deal. What I mean by that is that the instruments should track each other closely on a higher frequency basis (essentially tic for tic). The last thing you really want is for legs to converge or diverge wildly in ways that are difficult to model.
Just to clarify, I am definitely not contemplating a pair / spread trade in ES versus CL. However I am using the correlation (see chart here: http://www.elitetrader.com/vb/attachment.php?s=&postid=3021926) as an indication that the airlines will be a strong sell in the future based on the following conditions: (1) The S&P 500:crude oil ratio remains below 14.50 and (2) oil rises to a "high" level, maybe $100 or more. This is because: (1) at a "high" oil price, increased risk appetite / rising stock prices will also mean higher oil prices, and (2) the effect of higher oil prices (as a large expense item for airlines) will have more of an effect than the increase in stock prices generally. ********** price update: in the last 24 hours has moved above $90 (February 2011 contract). Although the $XAL is below its November high of 50.66, and although airlines have underperformed the S&P 500 since that peak, they aren't showing signs of "oil stress". For example, yesterday the airlines went up despite oil going up as well.
The 30-day correlation between SPY and USO in the last two years has varied from +0.31 up to +0.85. It is at +0.75 currently. The 10-day correlation is all over the place from -0.33 to +0.94. Currently is at -0.33. There is nothing there to base a trade on.
I disagree with your statement that there is nothing to base a trade on. By simply eyeballing a 6-month chart of $SPX versus $WTIC (continuous contract) I can see that these two follow a very similar pattern. I'm not really concerned with 10-day or 30-day correlation; this is something that has been "in play" for months. I also note that USO does not track $WTIC continuous contract closely. For example USO reached a high in April of 42.19 but is now at about 38.53. Whereas $WTIC reached the high 80s in April but went over $90 today. Furthermore, my thesis relates to shorting airlines, rather than trading oil: http://www.bloomberg.com/news/2010-...loom-as-crude-climbs-toward-100-a-barrel.html
Link to jet fuel prices: http://www.bloomberg.com/apps/quote?ticker=JETINYPR:IND Link to Bloomberg airline index http://www.bloomberg.com/apps/quote?ticker=BUSAIRL:IND I prefer this index to the $XAL because it follows the prices of RJET, JBLU, LCC, AMR, DAL, LUV and UAL more closely. Specifically, the 52-week range as of today for the $XAL is 30.91 to 50.66 (source: Yahoo Finance) whereas the 52-week range for the Bloomberg Index is 32-ish to 45-ish (source: me eyeballing the chart from the link above).
The missing piece which takes things to the next level is that there are are other instruments outside the equity and energy spaces that have very high frequency intraday and daily close correlations to the S&P 500. No one correlator is perfect, but if you can model a basket the effects and benefits are more consistent. For proprietary (contractual NDA) reasons, I can't specify the specific instruments, but they do exist and they are indeed used as a trading edge on the S&P 500.