Excerpts from Dean Williams speech: Trying Too Hard
Effort to output in investing is weakly linked compared to other endeavors
“We probably are trying too hard at what we do. More than that, no matter how hard we try, we may not be as important to the results as we’d like to think we are.”
“One of the most consuming uses of our time, in fact, has been accumulating information to help us make forecasts of all those things we think we have to predict. Where’s the evidence that it works? I’ve been looking for it. Really…The
consolation prize is pretty consoling, actually. It’s that you can be a successful investor without being a perpetual forcaster.”
“Confidence in a forecast rises with the amount of information that goes into it. But the accuracy of the forecast stays the same. And when it comes to forecasting—as opposed to doing something—a lot of expertise is no better than a little expertise. And may even be worse.”
Related: Horse bettor example
Adam Robinson on the diminishing returns of data. Confirmation bias increases your confidence without increasing your accuracy
- Study of horse handicappers found that the accuracy of their predictions did not improve from the original 5 variables they selected from a large menu of data. As they were given more variables there confidence went up (confirmation bias effect) although their accuracy did not!
- In addition, the handicappers with only 5 variables were well-calibrated. They were close to 2x better than chance at predicting winner 20% vs 10% and they estimated their confidence as such. When they were given more variables their accuracy remained 20% but confidence grew to 30%!
Gather the right type of info — namely how to measure value (Me: this is how I trade)
“We have security analysts. We get research reports from brokers. We get forecasts about the economy, interest rates, the stock market. We process that information and act on the basis of it. For all of that to make any sense, we all have to believe we can generate information which is unknown to the market as a whole. There is an approach which is simpler and probably stands a better chance of working. Spend your time measuring value instead of generating information. Don’t forecast. Buy what is cheap today. Let other people deal with the odds against predicting the future.“
Sources of edge
“There are two ways we can try to gain an edge over the market. The one that most of us choose is to try to generate superior information. To know more than anyone else. The other choice is to be better at measuring value than others and not to care very much about what other investors think they know. To hold cheaper securities by today’s standards and to let the future speak for itself.”
Growth describes a phase not a category of company
“It is generally recognized that growth stocks produce a superior risk-adjusted rate of return. However, this is only true for stocks that are expected to grow in the future, and correlations between past growth and future growth are low”. There is no such thing as a growth stock. Only passing phases of growth in almost every company’s life. Phases whose beginning and end usually appear in disguise.”
Regression to the mean
“The tendency toward average profitability is a fundamental, if not the fundamental principle of competitive markets. It is an inevitable force, pushing those profits and their valuation back to the average. It can be a powerful investment tool. It can, almost by itself, select cheap portfolios and avoid expensive ones. Its plain English equivalent is that something usually happens to keep both good news and bad news from going on forever.”
4 thoughts on “Trying Too Hard”
“Don’t forecast. Buy what is cheap today…measure value better than others…hold cheaper securities by today’s standards and let the future speak for itself.” Can you give 2 trading/investing examples where “buying cheap” didn’t involve a future forecast and the future spoke for itself? Regarding not playing for superior info, does not “value” insinuate you know something everyone else does not?
My view falls much more along the lines of “nowcasting” instead of forecasting. The effort is put on measurement. Buy what’s low, sell what’s high. No idea which leg will drive the profits and whether there will even be profits. But size this correctly and do it repeatedly and you have a business that didn’t really rely on forecasting. But there is a lot of value to be added in measuring. What’s expensive and what’s cheap leaves a lot of latitude for discussion and this is where people who touch the proverbial elephant from different angles will disagree.
Also 99% of trades I’ve ever done had no notion of a forecast. They were based on measures of value. I trade options so mapping this logic to other strategies might be non-obvious but if you poke around nowcasting I think you might find some.
Also scroll down to “gather the right type of info” here.
The lens I start with is what is this thing worth. If I buy it cheaply and sell other things expensively it works in the end if you risk manage in a way that respects path. But there’s no forecast that pretends to know where values will go. And many times things that get cheap get cheaper until they are actually expensive (since everything is always changing that happens on a relative basis). So I go from buying X when it was cheap. Watching it go down then selling X because my model now thinks it’s expensive. No forecasts.
I hope this is helpful.