A guide to making annual financial market forecasts.

I have posted this before but repost this slightly updated version as it is exceedingly pertinent considering what is happening in the markets after many institutions and departments have already published their top trades and financial calls for the next year. I would like to know how many have already been stopped out of their next yeartrades before next year has even begun. So here is a somewhat cynical guide to the whole process with some top tips for practitioners.

WHY

First, there is absolutely no point in embarking upon this process because only an idiot would put on a trade on Jan 1st and not look at it again until 31st Dec, with no positional adjustment during it. If everyone did this then the financial internet, press and even markets or exchanges would not have to exist between Jan 2nd and Dec 1st. Perhaps that's a good idea.

This process is much more like a yacht show where forecasting agencies can show off their shiny products of intricate design and process, wrap them with a glitz of gravitas to sell you a dream that you can sail off into next year aboard. If it doesn’t hit any rocks and sink then they hope you will be back to buy a bigger one next year.

HOW

The usual first rule of forecasting is never to put a price and time frame in the same forecast. For example ‘Oil will be over $100’ gives you an infinite time run within which to be right and 'Oil will be lower than here in the next year’ gives you a very high probability of being right unless you have unfortunately hit the absolute low at the time of prediction.

With this in mind, make sure your predictions are not for where prices will be at the end of next year, just somewhere during next year. This leaves you with a year of optionality to either get out or take profit, GS style.

Only publish your exit trades way after the event. For example, you called oil down, oil went down until Jan20th then up for the rest of the year - You cut the winning trade on Jan 20th. Of course you did.

WHEN

If you have to publish before the end of the year then don’t do so until Dec 31st. Doing so in November wastes a month of information to consider. Preferably publish your trades for next year in March, or even later. The first few months of the year nearly always go the wrong way (as 2014, '15, '16 and '18 have shown) and you will have three extra months of information to trade on. You also have the advantage of everyone else’s trades looking pretty stupid by then and investors looking for alternative sources of advice.

If you can, make the predictions for longer than just next year. Make them 5 year rather than one year forecasts and you get an extra four years of optionality with regard to the above methods but it also has two other advantages. First, you give off a greater air of authority having a five-year view rather than a one year view. Second and more importantly, after five years everyone will have forgotten your predictions except you, so if they are wrong it doesn’t matter. If you are right, you can wave them around in success and write a best-selling book and set up a large hedge fund doomed to lose all its assets.

WHAT

Of course what people actually care about is probabilities of outcomes and though I have often suggested that forecasters publish their own probability curves of price outcomes they still seem to stick with razor sharp defined price targets that, by definition, they are bound to miss. So best to put your 1 standard deviation range around a call. No one will mind and may prefer it. Either way, it won’t matter but does give you a wider chance of success.

Write exactly the opposite of the GS forecasts. If framed under the conditions of the first point you will still have opportunity to get out at a profit at some point and if you can show a profit against GS your views will be spread wide and far and you will appear on Financial TV, more because most people enjoy seeing GS look like a mug than your views being spectacularly right. If last year is anything to go by then you will have outperformed them by March.

Finally, even though the regular market faces of doom-mongering are beginning to be treated like quaint village idiots, remember to have an absurd couple of low-delta calamity calls in there. They can be camouflaged in sensible calls and will only get noticed if they are right.

WHO

Pick eight prices you are going to call, they can be as varied as you like, calling long or short for each. Then set up 256 new blog, Twitter, Seeking Alpha or some such identities. Using all the available permutations (2^8= 256) one of them is going to nail it exactly. This is who you now are. Wave this one around, write that book and set up that hedge fund. There are plenty of people out there still being dragged out for comment because they happened to have the lucky dice come up their way in 2000 or 2008. It could be you next year.

WAFFLE

All forecasts are now couched in as many pages of disclaimer as there are forecast. Make sure that each of your calls has 16 pages of arguments and appendices behind it. Come to the great reckoning there is bound to be something in this small print that can be held up as a caveat. E.g. 'though we call for US 10 year yields to go to 6% in 2017, should the Fed not raise to 5%, GDP not hit 10% and unemployment not fall to zero, we may see this trade underperform". Better still create your own benchmark and reference against it. "We see the suggested portfolio outperforming the Polemic risk-adjusted emerging developed market average weighted corporate bond fx hedged equity benchmark by 30 basis points on a chronological biased heuristic". Complete nonsense, of course, but a get out of jail free card when the time comes.

However, don't forget the 'unforeseen events' claus, such as Brexit or Trump which, though unforeseen by you, were foreseen by the majority of the population, proven by the fact they voted for it.

REPEAT

If all of the above is too complicated then just repeat the trade ideas that you put out last year, or if you are feeling very adventurous just extrapolate the last month's moves. It seems to be what most people do