On the Power of Heuristics and Reducing Complexity
Before American Pharoah won the Triple Crown in 2015, no one expected much from the horse. Jeff Seder felt differently. Seder had worked as an analyst at Citigroup before quitting and following his passion to predict the outcome of horse races. But, Seder was different than other equine researchers because he didn’t care about the thing that other horse breeders obsessed over: pedigree.
The traditional view among horse breeders was that a horse’s mother, father, and general lineage were the primary determinant of its race success. However, after looking through historical records, Seder realized pedigree wasn’t a great predictor. This meant that something else was. Seder needed data.
And it was data that he collected. For years Seder measured everything on horses. Nostril size. Excrement weight. Fast-twitch muscle fiber density. And for years he came up empty-handed. Then, Seder got the idea to measure the size of a horse’s internal organs using a portable ultrasound. Bingo. He hit pay dirt. Seth Stephens-Davidowitz tells of Seder’s discovery in Everybody Lies:
He found that the size of the heart, and particularly the size of the left ventricle, was a massive predictor of a horse’s success, the single most important variable.
That was it. Heart size was a better predictor of horse racing ability than anything else. And this is what Seder knew when he convinced his buyer to purchase American Pharoah and disregard the other 151 horses at auction. The rest is history.
Seder’s story highlights how deep insight can be gleaned from a single piece of information. Hans Rosling echoes this sentiment in Factfulness when he discusses the importance of a single measure in understanding a country’s development—child mortality (emphasis mine):
Do you know I’m obsessed with the number for the child mortality rate? … Because children are very fragile. There are so many things that can kill them. When only 14 children die out of 1,000 in Malaysia, this means that the other 986 survive. Their parents and their society manage to protect them from all the dangers that could have killed them: germs, starvation, violence and so on. So this number 14 tells us that most families in Malaysia have enough food, their sewage systems don’t leak into their drinking water, they have good access to primary health care, and mothers can read and write. It doesn’t just tell us about the health of children. It measures the quality of the whole society.
Rosling’s use of childhood mortality and Seder’s use of heart size perfectly exemplify the power of heuristics. A heuristic is defined as “any approach to problem solving or self-discovery that employs a practical method, not guaranteed to be optimal, perfect, logical, or rational, but instead sufficient for reaching an immediate goal.” While Rosling and Seder don’t perfectly describe the systems they are studying with a single measure, they are able to gain important insights with little information. Sometimes it helps to have lots of variables/measures, but sometimes you only need to know one big thing.
The idea of finding powerful heuristics to understand systems can apply to every part of your life including investing. For example, if you wanted to know the biggest determinant of your future investment returns, I would suggest you look at your asset allocation. As David Swensen, Yale’s Chief Investment Officer, once explained, “more than 90% of the variability in returns for institutional portfolios had to do with the asset allocation decision.” That’s it. Which assets and what proportions you choose to invest in will explain most of your future returns, all else equal.
But, what if we wanted to know something far bigger? Instead of asking what determines investment returns, we could ask, “What determines retirement success?” The American Society of Pension Professionals and Actuaries (ASPPA) has an answer. They published an article in 2011 where they found that 74% of retirement success had to do with one thing: savings rate. The other 26% was explained by asset allocation and related decisions.
So while asset allocation is important in one context (investing), it is less important elsewhere (personal finance). The famous Mr. Money Mustache took this idea to its logical conclusion when he built a community around the idea that savings rate is the most important decision in personal finance (and I agree with him). He knew one big thing and started a movement.
My point in writing this is to get you to answer the following question: What one piece of information would help you make better decisions in different areas of your life? To give you an example, I once proposed this thought experiment to one of my single, male friends:
Imagine you had a list of 10 single women your age and you can only go on a date with 1 of them. However, you know nothing else about any of these women. Not what they look like. Not their personality. Nothing. If you could only have one piece of information on all of them (no photos), what would you ask for before making your decision?
My friend thought about it for a moment then said, “How often they go to the gym.” I asked him why and he said that gym frequency was indicative of other positive attributes such as: hygiene, self-care, motivation, etc. My friend had just used the same heuristic shortcut as Rosling and Seder.
So, take this idea further.
If you are trying to improve your business, what one big thing distinguishes your great clients from your bad clients?
If you are trying to get healthier, what one big thing separates good health from ill health?
If you are trying to be a better parent, what one big thing differentiates an amazing childhood from a subpar one?
Though I don’t have the answers, when I frame questions this way I find myself saving time because I only focus on the most important aspects of a particular system. It’s not perfect, but I hope it can help you as much as it has helped me.
No One Big Thing?
I know what you might be thinking, “But, Nick. What if there isn’t one big thing? What if I have lots of variables/measures that aren’t correlated with the outcome I want?” I agree. The world is complex, which means that there won’t always be one big thing that is easily measurable. In data science, we attack this problem by weighting different variables/measures (regression) or combining and eliminating different variables/measures (dimensionality reduction/PCA). These different statistical methods have the possibility of creating “one big thing” out of a combination of many small things.
For example, while I insinuated that Jeff Seder only needed to know heart size to predict racehorse performance, this isn’t entirely true. In fact, Seder’s firm has found 48 variables/measures that are predictive of racehorse success. Heart size just happens to be the most important.
So, if you find yourself without one big thing, you may need to hire a data scientist or learn data science yourself. Trust me, it’s fun. Lastly, if you are interested in learning more about Jeff Seder and his journey into horse racing, check out this video and read Everybody Lies for the full story. Seriously, this was probably my favorite book I read last year. As always, thank you for reading!
This is post 109. Any code I have related to this post can be found here with the same numbering: https://github.com/nmaggiulli/of-dollars-and-data