Putting probabilistic thinking into practice – with Jake Taylor
Value investor, author and podcast host Jake Taylor discusses how decision-makers can seek to escape their own ‘little narrative tunnel’ and so better see the truth of any given situation.
Thomas Bayes was an 18th English statistician and philosopher, whose name is now associated with an approach to reasoning or decision-making where probabilities are continually calculated and updated as new information becomes available. We mention this now because we realise we have touched on ‘Bayesian thinking’ twice in recent weeks, here on The Value Perspective – and we are about to do so again.
In The impulse strikes back, which emerged from our podcast conversation with fund-of-funds manager Simon Evan-Cook, Bayesian thinking suffered slightly from its association with C-3PO – in contrast to Han Solo’s more seat-of-the-pants approach to decision-making. In Shed light on your analytical blind spots, our most recent podcast guest Jake Taylor mentioned it in the context of a missed investment opportunity.
To balance things up a little, let’s focus now on Taylor’s take on the applicability of Bayesian thinking when it comes to investing – and how challenging it can actually be to put into practice. “What is really hard about the investment process is we are seeking to find that ‘outside view’,” begins the value investor, author and podcast host, who is making his second appearance on The Value Perspective podcast.
“By that, I mean a way of getting out of our own little narrative tunnel and really trying to see the truth of whatever the situation is – whatever the truth is for that business or that marketplace or whatever it may be. And what is hard is that we all have blind spots and, by definition, they are hard to fix – because, otherwise, they wouldn’t be blind spots.”
There are multiple ways of seeking the outside view, Taylor points out: “You can talk to other people who maybe understand parts of a subject better than you do. Or you can read widely so you can find different tools and so see things through a different prism and start to uncover blind spots. And then you have base rates – basically, of all the times this type of situation has occurred, what has been the outcome and how often?
“The trick is to try to refine the base rate you are using to get closer and closer to what the actual truth looks like. That is what Bayesian probabilities are all about: starting with a top-level base rate and then working your way down – with the data – to truly understand, OK, as I get closer and closer to what the actual phenomenon is, what can I start to expect?”
That, argues Taylor, is really how our brains work too. “The human brain starts out with very basic sort of Bayesian ideas of, oh, if I drop this ball, then 100% of the time it is going to fall to the ground – and you start building up from there,” he explains. “And eventually, by the time you have a fully-formed brain, hopefully you are understanding a lot more nuance in the world and those probabilities start being reflected.
“And, with the brain being a Bayesian updating machine, this is why surprise is so important – because surprise is the feedback telling you, hey, one of your base rates is wrong. What you thought was going to happen did not happen – like, red alert, we need to update our models. That is why failure is so important – and a big part of the Journalytic software I am currently developing is to uncover those surprises.
“The aim is to rub your nose in your mistakes a bit by going back and looking at what your thinking was at the time you made the decision so you gain a better understanding of why you were surprised and you update your models better. The whole idea of closing the feedback loop is to learn faster – you recognise your mistakes faster and then you take measures to fix them.”
One last thing – aside, of course, from buying Journalytic when it hits the market – does Taylor have any tips on how to become better at adopting probabilities into our mental and investment processes? “A lot of times, probabilities come from intuition,” he replies. “We have all these mental models, where a lot of stuff happens at the subconscious level that your cognitive, executive function does not even really have access to.
“So just writing things down – and then going back to see how it turned out – will start to close that feedback loop. Intuition does not work unless feedback is provided to start building that intuition up – if you are just making intuitive guesses in a vacuum, I don’t think you ever get better at it. So you just have to start writing things down and keeping track a little bit better and then you can start to trust your intuition more.”