Thursday, February 01, 2007

Forecasting your own actions...

Forecasting is extremely hard. Even when the forecast is of your own actions. Consider how accurate you'd be in predicting what you'll be having for dinner next Sunday. Its an activity which you probably have full control over, and yet there are too many variables and too many decisions between now and then for most people to make an accurate prediction.

An article in the Financial Times highlights a similar problem, albeit one with much more consequence. I found the link courtesy of Macroblog. The article includes a chart of where actual central bankers in New Zealand predicted they would be setting rates in 3 months time. The bottom line is that the banker's predictions seem to be of little value. And the reason why is fairly obvious, there are just too many variables in economics for any prediction to be particularly accurate.

In my time as an investment manager, there are a few core tenets in which I have come to believe very strongly. The italicized statement above is one of them. No matter what the prediction, no matter who the forecaster is, and even no matter what special information s/he might have, any specific economic forecast has a low probability of being accurate.

What's the consequence of this? How does realizing and accepting this make you a better investor? It isn't just about hedging and/or limiting the portfolio bets you make. That might reflect proper investment thinking, but it might also reflect indecision or cowardice.

The right reaction to the fact that forecasting is so difficult is to think of your own forecasts in terms of possibilities rather than predictions. This is something I wrote about in more detail earlier in the month, in conjunction with my 2007 Forecast post. You may have a particular forecast that you think is most likely, but you should have alternative forecasts which you view as reasonable possibilities. You should then build a portfolio that will perform well in various possible scenarios, if possible. Or else build a portfolio which will perform very well in your most likely scenario, but perform at least fairly well in other scenarios.

Note that this is not the same as simply limiting your portfolio bets. For example, at the beginning of 2007, I thought rates would shift slightly higher and the curve would steepen as the Fed cut only 1-2 times. I also thought there was a possibility that the economy slipped into recession and the Fed cut several times. While my primary forecast was for higher rates, my secondary forecast was for drastically lower rates. Therefore a duration bet is a tough one, because my payoff was small if the first scenario came about, but my pain was large if my second scenario emerged. So I looked to make a larger bet on the curve steepening. I'm willing to make this a large bet, because in both of the most likely scenarios I see, the curve steepens. So not only have I positioned my portfolio in the middle of the curve, I've put on a non-inversion position, and I've loaded up on MBS, which tend to perform well in a steepener.

My MBS stance is another example of how you just limiting your bets is the wrong way to go about it. I'm heavily overweight MBS. The MBS position performs well when rates are fairly stable, which means that if my primary prediction is correct, MBS will be a big performer. I own mostly discount MBS, so if the Fed cuts aggressively, I should see significant prepays on a discounted item. Even more aggressive is my play in discount hybrid-ARMS. There borrowers should use any period of Fed cutting as a chance to reset their fixed (or even IO) period and stave off the ARM portion of their loan. So the discount hybrid-ARM play is a home run if we have a bull steepener. And if none of this happens, I just ride the extra carry off into the sunset.

These are just some examples of how you can make a series of aggressive plays but still wind up performing well in various scenarios. Contrast this with the more common approach of just limiting portfolio bets. The result there is mediocre performance no matter what. I just don't think that's what clients pay investment managers to produce.


Anonymous said...


core tenants?, of course, mean core tenets.

Keep up the good work!

The Grammar Police

Accrued Interest said...

No no... I mean the people living in my brain who give me investment advice. I should obviously read over my posts more often.