Intelligence is not a one-dimensional affair – as anyone who has seen a nerdy Ph.D. trying to mingle at a party can easily attest. Psychologist Howard Gardner says people have Multiple Intelligences: he identifies no less than eight kinds, which he finds to be mostly uncorrelated. One can be a genius in some dimension and a complete fool in another.

In an interesting new book, Dylan Evans puts forward an additional kind: Risk Intelligence, which he defines as the ability to estimate probabilities accurately. Risk intelligence is also uncorrelated with other kinds of intelligence, so don’t be wary of taking Evans’s test. It consists of 50 general knowledge statements, half of which are true and half false. Answers require giving the probability that a statement is true. So, if you are certain that a statement is true, the answer should be 100%. If you are certain that a statement is false, it should be 0%. If you have absolutely no idea whether the statement is true of false, the answer should be 50%. If you are not sure, but you believe that a statement is more likely to be true rather than false, you should choose an answer between 60% and 90%, depending on your level of confidence. Likewise, if you believe that a statement is more likely to be false rather than true, you should choose an answer between 40% and 10%.

You have high risk intelligence if your answers are well calibrated. Perfect calibration requires that x% of the statements for which your answer is x% turn out to be true. So all the statements for which your answer is 100%, and none of the statements for which it is 0%, should be true. And the same in between the two extremes: 90% of the statements for which your answer is 90% should be true, and so on.

Unsurprisingly, not many people have high risk intelligence – you will be shown the sample statistics after you take the test. And the most common drawback is overconfidence, where the percentage of true statements for which people answer x% is significantly less than x% for x>50, and significantly more than x% for x<50.

Now imagine you are shown 50 companies, and the statement to which you have to give a probability is: This company is rightly priced. An Efficient Market theorist will always answer 100% – and come up with often fanciful ex post rationalizations of why he was right all the time. Mr. Market also says that the price is right. Investors are strongly inclined to believe Mr. Market because they are prone to the Prior Indifference Fallacy – in this context, the assumption that the stock price has a 50/50 chance of going up or down. Notice that always answering 50% in Evans’s test leads, formally, to perfect calibration – which, of course, is perfectly worthless. High risk intelligence requires being close to perfect calibration along the whole probability spectrum. Since overconfidence is a constant threat to risk intelligence, investors should focus on those companies where the probability that the price is right is particularly low – which, once again, underlines the importance of the Margin of Safety.

### 2 Responses to “Risk Intelligence”

1. Very interesting insights! I look forward to taking the test…

2. I’m glad you liked the book. Your blog post on my book summarizes the main ideas very succinctly. I particularly like the way you have put the following point:

“Notice that always answering 50% in Evans’s test leads, formally, to perfect calibration – which, of course, is perfectly worthless. High risk intelligence requires being close to perfect calibration along the whole probability spectrum.”

This is a beautifully economic way of putting it.

This site uses Akismet to reduce spam. Learn how your comment data is processed.