what factors/data shall I use to predict the movement of Crude Oil? I am playing around with various methods of forecasting, what data are relevant and therefore I should throw into my forecast model? I am thinking of SP500, what else? Any suggestions? Thanks!
CL IMO is the most "technically" traded contract. Some good ideas would be.. 1. Have a large bankroll 2. Never break your rules one time (automate EVERYTHING if possible) 3. Learn fundamentals like backwardation. I really know nothing about number 3 but that hasn't shown in any results so far. All I know is I see history repeat itself often in CL. GL
Think about what is closest to crude oil (CL) chemically. - Brent crude - West Texas Intermediate - Gasoline - Heating Oil Much closer relations than those with indices or currencies.
Fundamentals historically obviously play a big part.... tightness in the physical cargo / barge markets, refinery shut downs etc etc. Suprisingly whilst having an effect on the term structure these have been having less and less of an effect effect on the flat price, or if they have it's gone in the opposite direction to what you'd expect!!! It's almost like the fundamentals are being completely disregarded. For example, build of about 10m bbls in the last 2 weeks according to DOE stats, yet flat price is strong today, roofing on the early Platts window. Geopolitical tensions - Iran, Nigeria. Agree with last poster - all the benchmarks are highly correlated. If you're trading brent for example, and you see a big spike of volume go through in WTI breaking new highs, then it's a good confirmation that the brent is going up too, etc. But for me, it's all about the USD at the moment. Watch key levels against the EUR. The past few weeks with the Greece situation. Any little bit of news regarding that, e.g. comments from Germany's Merkel, have often moved the market A few locals I know have pretty much stopped trading at the moment, because of so many mixed messages and such indecisiveness. It's a bit tough at the moment!
Sounds interesting... I am starting to look into Neural networks and regression. For those tools, I need to throw in some relevant data. Hopefully we can even find/use some leading factors... Any thoughts?
Forget about leading or lagging. Probably no such things left today where the machines are trading in sub-second timeframes. They are synchronous up to the second (or even below). This can be used for confirmation.