Comintel, he is being incredibly sarcastic there - it is called "the widowmaker" for a very good reason. Please google the phrase "Bryan Hunter Amaranth CFTC final report". If you had charted the cracks using tic data, or even one minute data, you wouldn't have entertained that notion either in your quest for mean reversion. Mean reversion, in a simple two-legged energy futures pairs ? Godspeed. Also, how do you plan to address CME's 18:1 message-to-fill ratio policy ? Also, I know guys at CME Electronic Member Firms that spend a ton of money per month on ECN infrastructure (including their own lines and co-located servers) who still get picked off all the time - and they are located within spitting distance of CME's server farm.
I realized he was being being ironic. I am also aware of Amaranth although I did not recognize the reference here. Very good! Of course they had lots of opportunity to get out and failed to do so. Well I am only looking into this. It does not have to just be too legs. I take it you trade from your other posts or blog that you trade one crack spread against another. I assume that that is a mean reversion trade?
I don't think that either of those statements are good assumptions. What you are not really acknowledging in your mind is that the mechanical details required to automate the trade - which I alluded to in my previous post here, you have not addressed. Those details are a much larger and more critical task than your strategy quite honestly. In fact, you are looking at this bassackwards - those execution semantics and exchange rules steer what strategy you use.
I do not intend to do high-frequency trading. Obviously that is not feasible for me. It would be closer to trades that might last a few days with mechanically-assisted entry and exit and position monitoring.
There are systems that already exist that can handle the automated execution of the trades, so that is not the issue. I suggest looking at stocks with a high average trade range, for day trading. regardless of the sector, if a stock moves 5% between the high and low each day, that leaves a lot of room to make your 1% or 2% profit.
OK sorry about that - you do mention the Nymex Heating Oil Crack Spread versus the ICE Gas/Oil Crack Spread but in another context, not mean reversion. Some trades you have referred to in connection with mean reversion are multiple instrument spread positions like butterflys, condors, and baskets, when properly constructed and modeled, as you have noted. I obviously need to modeling of some of those.
if you want mean reversion i would suggest what you got to here- butterfly's and condors... in my experience i've found the more i hedge the more mean reverting the positions are (in the short term- i never hold positions overnight i'm strictly intraday)
comitel, look into z1/m2/z2 crude fly, f/g/h and m/n/q natural gas flies. And on the dangers of H/J ng: http://www.mresearch.com/pdfs/317.pdf