Is anybody profitable by identifying a leading instrument and trading a correlated lagging instrument? Just taking a position in the one leg of the spread? What type of statistics/modeling would be best to analyze this type of trade?
Of course there is - that's the entire premise behind a prominent sector of algo programs. Nobody's going to do the work for you, so get to work solo.
I'm currently looking at pairs such as CL/HO and CL/RB using CL as the lead. My analysis has been mainly based off the delta between the individual % change for the correlated instruments and potential trades as the delta expands. Just looking for a point in the right direction to take this to the next level.
Actually, CL/HO and CL/RB are particularly poor subjects to analyze from a realistic trade-ability perspective unless you had access to a liquid ( no pun intended ) OTC product. You would never be able to get filled execution-wise, even with an automated lead-lag strategy that only executes the lagging product, faster or with better results than the exchange-support Heating Oil or Gasoline Crack Spreads. All automated platforms have to be registered with the CME, and their order messaging is tagged as "conditional". You're always going to be last in the order queue and you would be getting picked off constantly with something as highly correlated and exchange spread-traded as a CL Crack Spread. From a modeling perspective, HO and RB are almost always going to lag CL - and your analysis will not be reasonable unless you can somehow incorporate the on-the-run CL-HO and CL-RB Crack Spread bid/ask spreads in perfect concert with the individual contracts you are looking at. IMHO, you would be better served looking at a good solid intramarket synthetic spread relationship like the JY versus the JGB. Also, please make sure that you are only analyzing the trading session time frames where both products are trading simultaneously. I hope this helps in terms of getting you pointed in a more fruitful direction, and that I have spelled out some of the pitfalls in terms of practical real-world application for intermarket analysis and correlations. Yes, this is a really worthwhile endeavor. But you can also get your ass handed to you in a big way if you haven't correctly modeled and tested these combinations and if you haven't had extensive real world experience trading them. Try to get Bloomberg T&S data, because everyone else's data just sucks in comparison.
You have to put the position on as a pair and give yourself time for market direction to reverse. For this strategy you may want to look at futures pairs that have at least 60 days before expiration. Just my opinion.
And a damned good opinion it is - with the exception of waiting for the market reversal thing. What I have personally found over the years is that it's better in terms of anxiety, quality of sleep, and drawdowns to sacrifice some trading range in terms of profit target in exchange for a reversal that is confirmed. In other words, let the deep pockets turn the market for you. YMMV.
So I am gathering that the preference is to trade the spread rather than a single leg - and that a longer time frame is needed. My previous attempts included intraday spread trading and found the round trip slippage, especially with market orders, is too great for intraday trading. That's why I've been looking at a single leg strategy. Is this going taking the wrong focus?
For a single leg strategy, I think you need other information to pick direction ... using correlated spread info may give the inconsistent directional signals. I understand the lead/lag relationships that "seem" evident in some futures contracts, but use caution. Just grab a pair of futures like ES/YM or HO/RB or SI/GC and learn the tendencies of that pair.