Sunday, December 13, 2015

"Getting Forecasts Right Requires Admitting You’re Wrong"

In 2011's "Why Analysts Should Keep it Simple (BAC)" I found myself wasting my time and thought of one of my favorite examples of clear thinking:
I just read half of a manager's lengthy defense of a position in Bank of America that sounds to be seriously underwater.
After a few thousand words I asked myself if this was the highest use of my time and got a ringing "No" for an answer.
I've never cared for the 37-page investment thesis approach to investing.
Instead, I've hung around people who are very quick on their feet and who make decisions-based-on-imperfect-information all day long and I've sort of acquired the trait.

Here's a repost from 2008, I'm with the General on "failing forward fast":
Ag Stocks and The Berlin Airlift (AG; MOS; MON; POT) 
Last night in "Commodities Comeuppance" I said my best guess was that the ag stocks would be up today. When POT and MOS opened down I was reminded of a vignette from the Berlin Airlift.

We're coming up on the 60th anniversary of the Soviet blockade that June.
During the summer the two million people that the Brits and Americans were trying to feed could get by with two tons of coal per day (over the course of the airlift 80% of the weight hauled was coal) but as the blockade went on, it was apparent that the Sov's. intended to starve the city and it became imperative that an efficient method of delivering coal be found.

During winter the absolute minimum requirement was 3100 tons of coal per day. The little C-47's could haul around three tons per flight. The first week of the airlift, deliveries averaged 90 tons per day. The second week, 1000 tons/day. Not enough.

It was decided to experiment with a low-speed, low-level drop of coal onto an empty field, the idea being that if it worked, B-29 Superfortress' with a 105 mph stall speed and 22-25 ton capacity would solve the problem.

On the appointed day the senior commanders went to the field, the plane came over, low and slow, dropped the coal, packed 100 pounds to a bag, the bags landed, exploded open, the coal was pulverized and a great black cloud of coal dust covered everyone watching.

One of the Generals, I forget if it was LeMay, Tunner or Smith, turned to the others and covered in coal dust said, "Doesn't work" and that was that.

When I saw the ag stocks open this morning I thought
"Doesn't work". ...
There's more to the story of the airlift if you follow the link but the key point is: you have to be intellectually honest enough to fess up when you've made a mistake.
Here's today's headline story from the CFA Institute, Nov. 27, 2015:
Economic forecasting is fraught with uncertainty, and even the most seasoned financial professional is only correct a fraction of the time. Yet despite this uncertainty, forecasters rarely revise their predictions when new evidence presents itself.

Tim Harford, behavioral economist and columnist at the Financial Times, has suggested that this lack of revision stems from an economic God complex, and he discussed how analysts can make more accurate forecasts at this year’s European Investment Conference.

Harford draws upon the work of University of Pennsylvania professor Philip Tetlock, particularly his concept of the “Superforecaster,” to explain why a certain psychological attitude enables better predictions. Tetlock found that people with a particular personality type — those whom he dubs “actively open-minded thinkers” — are able to make more accurate forecasts because of their willingness to change their minds in light of new evidence....MORE, including video.
HT: Mr Viskanta of Abnormal Returns writing at the CFA Institute's blog:
Behavioral economics and finance have become central to the education of financial analysts. In this presentation, Tim Harford, author of The Undercover Economist Strikes Back, talks about how good forecasters keep an open mind and are willing to change it when faced with contradictory evidence. (European Investment Conference)
Note: We have so many posts on forecasting that it's probably easiest to, if  interested, start with a search of the blog for "Tetlock".