Fri. 10Sep2010, 0155GMT
Didn’t I do a post on forecasting not too long ago? Had a link to something … that’s right the CSS Analytics page.
Well, today I wanted to ramble on a little more about forecasting, from a different angle. The reason for this is I don’t want to talk about the market really. It is rollover time, a time I find confuses me in my short-term trading. The theory is that the rollover day occurs 8 days before expiry, and on rollover day the volume shifts to the new contract. Yeah, right. No, it doesn’t. Not necessarily. I can find myself playing the wrong game, on the outside court, while the action is elsewhere. Yes, this is a moan and a rant, I have no basis to complain other than my desire to, so please feel free to ignore this paragraph.
There are a number of reasons put forward why most people are mostly unsuccessful at forecasting. All sorts of biases, the inability, or unwillingness, to recognize new, perhaps conflicting information (let alone to actually seek it out to test your hypothesis), a refusal or inability to recognize the impact unexpected developments might have … the list goes on.
But here I just want to focus on one aspect of the difficulty, the impact of unexpected events, events that can’t be reasonably (or even unreasonably) allowed for, and propose an (albeit imperfect) solution. Simply put, when forecasting I like to cheat. I have a crystal ball (I keep it in the cupboard with my tin hat). LOL – no I don’t have a crystal ball, but I do cheat.
BTW, when I talk about ‘forecasting’, I am talking about actionable, tradeable forecasts … I think the price is going up, therefore I want to be long … along those lines. Don’t get me wrong I’m happy to rabbit on about peak oil, long-term inflation prospects, the future of flying cars, and so on, with the best of ‘em (well, maybe not the flying cars), but just not using these flights of fancy for tradeable ideas.
OK, so how to cheat at forecasting? When I forecast, it is only for a very short time ahead. The longer out the forecast, the higher the probability of some unexpected event occurring to disrupt the forecast. Actually, reading this makes it appear that the only thing standing in the way of my successful long-term forecasts is unexpected events … this is not true, there is my incompetence at forecasting to allow for too…
OK, so I am probably not in the running for a Nobel prize for this amazing insight (LOL) … but there you have it. So when you see me talking about my models and what I expect, the ‘forecast’ is good usually until lunchtime in NY, that’s about the limit of my actionable forecasting aspirations.
Of course, between NY lunchtime and the development of flying cars, there is a massive timeframe that I am pretty much ignoring. That’s not quite true … I do like to ponder on where is it going for the next few days, the next week … but truth be told I am much better in my very short-term time frame than in any other.
When it comes to actually actioning (trading) my forecasts there is a lot more to say, a lot more to consider, but for another day. (Actioning? Making up words like this, I really should be a management consultant).
My models today, through to NY lunchtime … it really is an arm-twisting day, model #2 is giving me nothing at all (and this time I mean it, nothing). Model #1 … it is so hard typing having my arm twisted like this, LOL … nope, there is nothing there. OK, if you want to go further than twisting my arm, maybe you want to hold a gun to my head for a view, I think there is a not insignificant risk of a 15-20 point downmove, it is not high probability, but if there is likely to be any move of magnitude, this would be it, IMVHO. Otherwise I think we may well be 5 points either side of here – we are about 1099.50 on ESZ0 as I type – come NY lunchtime (and probably Friday close) So, in summary, don’t expect much, but the risk is higher for a downmove of 15-20 points than an upmove.