By Adrian Colarusso, CFA, CFP®
September 17, 2023
On Monday Night Football, the Jets faced a 4th and inches with 2:48 left in a tie game. The ball was on the Bills 17-yard line – comfortably within range for a go-ahead field goal. What to do?
Football fans are well-versed in the decision-making framework Jets coach Robert Saleh faced.
Saleh quickly cross-tabulated an array of probabilistic variables inside his brain’s statistical model to make the kick-it or go-for-it decision.
How real is the risk that the offense can’t gain a few inches? How does the probability that the Bills answer with a score vary with the time left on the clock?
As us fans were making up our own minds about the right decision, the announcers chimed in decisively: “The analytics department says go for it.”
Indeed, the Jets did go for it, successfully converting the first down and ultimately winning the game.
AI enters the chat
Since 2018, the NFL has touted its “Next Gen Stats” powered by Amazon Web Services. These analytics go far beyond “who leads the league in passing yards” into the realm of “expected rushing yards for a quarterback in scrambling-versus-designed-run situations”. Obsessive fans gobble up this additional information as they draft their fantasy teams.
Meanwhile, coaches are turning to this treasure trove of new data to help them make better decisions on the field. From the NFL:
Whether the coach makes the “right” decision is more about process than outcome. The optimal call is the choice that gives the team the best chance to win with the information available at the moment the decision is made. A series of smart decisions can lead to compounding effects on the scoreboard.
Of course, the same principle applies for wealth accumulation and protection.
Process leads to decisions. Decisions – plus luck or fate (depending on your philosophical persuasion) – lead to outcomes.
A cause for celebration
I am glad that sports are opening our eyes to the power of process when making decisions in the face of uncertainty. The rise of Next Gen stats in the NFL, and books like Moneyball teach us valuable lessons about investing.
No one has a crystal ball. Therefore, predicting the future is an unreasonable goal. Instead, we should cultivate the ability to think probabilistically, while clearly defining our objective functions – i.e. the variables we are trying to maximize.
We don’t need to be totally submissive to the AI models, but we must take in a sufficient volume of data through experience to hone our own mental models for making decisions.
Should we “go for it” with a single stock?
Statistical models play their role in our investment process. Most notably, they help us calibrate our clients’ portfolio risk – quantifying “how much” (in terms of expected volatility) and “what kind” of risk (in terms of expected portfolio reactions in the face of various potential market stresses).
Viewing a diversified portfolio through the lens of a risk model helps us align our intention with our execution. We leave room for the art of judgement to express our views of the world through our portfolios.
We say this a lot, but our “analytics department” vociferously contends: you generally don’t get paid to take single stock risk. This is a call-out to those with concentrated positions in one company, perhaps from stock compensation.
We’ve found that people tend to underestimate the additional risk they are taking and overestimate the probability that their chosen stock will outperform more diversified (and lower risk) alternatives.
When might someone get paid to bear this extra risk? One group we work with are the most influential company executives, whose efforts can swing results for their stock.
If embedded tax liabilities are a problem, we have a few tools that can help. Navigating the intersection of investment risk and taxes is our specialty.
Quick thoughts on AI investing
The current hot thing in the investment landscape is artificial intelligence. A couple years ago, every company was “pivoting to blockchain” to give their stocks a boost. The same dynamic is happening now with a new buzz word.
We believe it is far too soon to predict the myriad implications that AI will have across the global economy. It is exceedingly difficult – even for investment professionals – to pick individual winners and losers before their prices move.
One stock that has surged this year is Nvidia. Investors in the S&P 500 benefited by holding a 1.13% position in the stock to start the year. Now, the stock represents 3% of the index.1 If you stop trying to play “needle in the haystack” and just own the haystack, you’ll probably be better off.
The inflection point we are seeing in AI makes us decidedly optimistic about the future. Yes, AI has its potential downsides, but the power of these models to improve our collective decision-making should have compounding positive effects on the world – and, likely, its equity markets.