PsyFi Search

Saturday 17 September 2011

HONTI #4: Disconfirm, Disconfirm, Disconfirm

Rule #4: Look for evidence that you're wrong, not that you're right.

Be Prepared to Change Your Mind
"When the facts change, I change my mind. What do you do, sir?"; John Maynard Keynes (allegedly: see Quote Investigator)
Once you've done your careful research, or blindfolded a bemused ape and got it to stick a pin in a list of stocks, and bought into some corporation or another, you immediately become exposed to one of the nastiest behavioral biases on the block: confirmation bias. You will, whether you realise it or not, start to favour information that supports your decision and to discount that which doesn't.

Ever get a nice warm feeling when someone tips1 a stock you own?  That's confirmation bias in action and if you let it blindside you it'll take you down, along with your portfolio.

As a general statement, we should only buy into firms about whose prospects we're convinced. Strong corporations with good managements and defensive moats and which are cheap on fundamental grounds are ideal. Sadly such opportunities are rarer than hen's teeth, so usually we've got to make a few compromises along the way.  And, by and large, the less experienced an investor we are the more mistakes we'll make. Or so you would think ...

If we start from a position where we know we'll make errors then we should make every effort to counteract the powerful persuasive lure of confirmation bias.  Unfortunately we're not very good at doing this because, in general, we don't spend our lives looking for evidence to disprove our convictions; quite the opposite, most of the time. The classic psychology experiment that demonstrates our disconfirming weaknesses is the Wason 2-4-6 Test: you have to identify the rule governing the sequence of triplets by generating new ones and asking if they fit the rule.  The research participants nearly all failed because the came up with a theory and then sought to confirm it.  What would you guess?2 

In order to figure out the right answer you to find need a triplet or two that breaks the rule, but most people never found one, largely because they never looked for one.  And this is what happens if all you ever do is look for confirmation that your ideas are right.  In fact, it's worse than this: because after people were told that a rule they'd suggested was wrong they still carried on looking for evidence to confirm it!

Once an idea has taken hold in our heads it's very, very hard to shake it.  And, perhaps surprisingly, being an expert doesn't seem to reduce the problem, in some cases experts are more likely to suffer from confirmation bias3.  Novices can generally be persuaded to cross-check their hair-brained theories because they understand that they're inexperienced but experts often ignore disconfirming evidence, presumably because they've developed semi-automatic ways of making decisions.

There are  studies that show this type of expert failure playing out in stockmarkets.  Long-term evidence shows that more experience leads to worse forecasts once you get past a fairly low level of expertise.  There are multiple things going on here, including some sort of herding effect in expert commentators4, but the overall findings suggest that the forecasters simply don't look for evidence to disconfirm their theories.  Why would they?  After all, they're the experts.

Investors also get caught by confirmation bias in strategies as well as in individual stocks.  So, for instance, there's a general theory that a low market price-earnings ratio is likely to lead to large gains, and vice versa.  It's quite easy to find confirmation of this.  However, if you look at the evidence as a whole what you discover is that years with low price-earnings ratios are followed almost equally by two year periods of high and low returns5: the market price-earnings ratio (and dividend yield) has no predictive period whatsoever over two year timescales. 

Another area of extreme danger for investors are internet based bulletin boards and their like.6   The evidence suggests that people use these to gather evidence to confirm their prior views, rather than the opposite.  This behavior appears to result in lower returns.7  It's quite likely that a secondary bias known as polarization is at work as well8: polarization results in a group of like minded people gravitating towards the views of the most extreme rather than, as you might expect, the average. 

Apart from recognising that relying on experts is an error and that bulletin boards are full of people with similar opinions what should the beleagured investor do to counteract this problem?  Well, there are a few techniques which may help, although if you insist on taking a positive view of an investment you'll end up misusing these.

Firstly, you should try to imagine that your investment has gone bust, and then try and come up with reasons why this could happen.  This should help to debias you somewhat, as it forces you to address the worst case of an absolute loss.  Secondly you should absolutely avoid making decisions under time pressure: stressful situations accentuate confirmation bias.9  This is good general policy in investing anyway; when you're making a decision you'll have to live with for years you don't want to rush into it, no matter how much hype surrounds a stock.10

Thirdly, try to stick to the facts.  If you absolutely have to frequent internet bulletin boards (I do, they're a constant source of amusement, if nothing else) then weigh every statement for the evidence behind it.  Phrases such as "I think" or "I believe" or "I expect" offer about as much explanatory power as a stoned astrologer attempting astrophysics through shell scrying, however respected the contributor. And this applies to all sources of investment advice, including blogs on behavioral finance ... differentiating between opinion and evidence is important, even though evidence itself is not always impartial.

Finally, working with someone on an investment idea in a Good Cop/Bad Cop pairing definitely helps.  Contrary opinions are important and no matter how much we try it's almost impossible to carry two sets of opinions in our heads.  If that's not possible then running a paper portfolio alongside your real one is recommended: if the two differ, and they will, you need to use the cognitive dissonance this provokes to see why there's a difference.11

In the end, unfortunately, you're not going to be able to completely eradicate confirmation bias.  Despite your best efforts you'll make mistakes; as we've discussed before the critical thing is to learn from them, not ignore them.  Eventually, hopefully, you'll spot the more obvious warning signs: just don't think becoming more expert means that the problem goes away.  It doesn't. Ever.




Notes to the article:
  1. Have a look at Don't Lose Money in the Stupid Corner.  The best advice about tipsters seems to be to find the worst ones, and then do the opposite of what they recommend.  In the words of Stuart Sutherland: "Remember nobody is always right, though some people are always wrong"; (Irrationality).
  2. In fact the rule was "any three numbers in ascending order" and we covered this in Confirmation Bias: The Investor's Curse.  There's some evidence that the result of the Wason Test is a facet of the experimental condition: mostly we don't end up in situations where the answer we're seeking is quite so broad.  However, one of those less-common situations is the stockmarket, where we deal in uncertainty on a daily basis.
  3. This issue was covered in James Randi and the Seersucker Illusion.  Experts don't think they can be wrong, so what would be the point in looking for evidence that they are?
  4. Also covered in James Randi and the Seersucker Illusion.  Younger analysts are particularly prone to the problem, because they get punished for being wrong and out of step with the crowd and don't get rewarded for being right in the same circumstances.
  5. See Cognitive Biases in Market Forecasts by Ken Fisher and Meir Statman.  Good read all round.   
  6. We looked at this in Bulletin Boards Are Bad For Your Wealth, touching on committment bias as well, and there's also some research in Overconfidence and Over Optimism.  Look especially at Lynn Stout's research.
  7. See Confirmation Bias, Overconfidence, and Investment Performance: Evidence from Stock Message Boards by JaeHong Park and colleagues.
  8. Lots of evidence for polarization, in lots and lots of different settings.  Some of this suggests that the mere exposure effect is at work (see Fooled by Fluency), others that it's down to novel and persuasive argumentation.  Whatever causes it, it definitely happens; see Daniel Isenberg's summary in Group Polarization: A Critical Review and Meta-Analysis.
  9. Evidence for this can be found in Countering Positive Confirmation Biases in Command Teams which oddly enough comes from the Head Battlelab of the Singapore Armed Forces Centre for Military Experimentation.  Which sounds fun, whatever it is.
  10. The temptation to turn stock investing into a time critical activity is very powerful, we're continually presented with "once in a lifetime" offers; as we saw in Robert Cialdini and the Weapons of Influence time-dependent scarcity is a powerful, if unconcious, persuasion technique.  However, this triggers our emotional decision making machinery, which is designed for rapid responses, rather than our cognitive and analytic capability.  This is a mistake.  See Stocks Aren't Snakes for a description of the issues.
  11. Terrific piece on this and fighting confirmation bias in the WSJ by the constantly good Jason Zweig: "How To Ignore the Yes-Man in Your Head".

1 comment:

  1. made up a story to explain confirmation bias (and other concepts), to my kid.
    http://www.bpsundar.in/home/croak

    ReplyDelete