The Sound Craftsman

Active investing in the age of AI

I would not be one of those who will foolishly drive a nail into mere lath and plastering; such a deed would keep me awake nights. Give me a hammer, and let me feel for the furring. Do not depend on the putty. Drive a nail home and clinch it so faithfully that you can wake up in the night and think of your work with satisfaction - a work at which you would not be ashamed to invoke the Muse. So will help you God, and so only. Every nail driven should be as another rivet in the machine of the universe, you carrying on the work.

-Thoreau, Walden

If we render unto AI what should be ours, then our profession is lost.

Another investor recently suggested I try Claude Pro for modeling.

I have to admit: it’s pretty extraordinary.

With a simple prompt of “Build a DCF model for Company XYZ. Include five-years of historical financial data” into the Claude Excel add-in, it did exactly that. It sourced data from the 10-Ks and other public sources, estimated cost of capital, included a scenario analysis, and built a five year forecast.

It took about five minutes to accomplish this task.

By comparison, building my own model and entering five years of financial data by hand takes a few hours. Deliberating over the forecast adds perhaps a few more hours.

From an efficiency standpoint, AI is the winner. (And it will only get better and faster.)

However, something important was missing - I didn’t understand the company as well as I normally do after building a model.

Photo by Craft Kitties on Unsplash

Entering historical data by hand might seem like a waste of time, but going through the process requires you to pay attention to every stitch of the finished product.

Why did the company decide to change the segment reporting in 2023?

That was an odd jump in receivables. What explains that?

Even if the same numbers show up in the AI-generated model, it’s much easier to gloss over them and never ask the important questions.

When you do it by hand, you can’t ignore them.

To be sure, AI can and should play a role in the craft of investing, but it’s important for us to understand the line between craft and efficiency.

Here’s an example of how AI can be used to enhance the craft. Let’s say you notice that receivables have been ticking higher after you entered them by hand. Now you can engage with AI to consider angles you hadn’t thought of before - Channel stuffing? A change in payment terms? A deteriorating customer? - and develop an even better understanding of the company’s underlying reality.

AI can’t answer questions it’s never asked. Engaging in the craft is the only way to generate the questions that really need answering.

Ultimately, investing is a decision-making business. The craft of investing lies in our ability to synthesize and discern information so that we can make better decisions.

To be a sound investing craftsman, then, it’s important to hold tight to what the craft demands - and it is demanding - and not yield to the easier path. We do the demanding manual work, however, because when the market tells us we are wrong, we are better positioned to hold our nerve.

Stay patient, stay focused.

Todd

Todd Wenning is the President & CIO of KNA Capital Management, LLC, an Ohio-registered investment advisor that manages a concentrated equity strategy and provides other investment-related services.

At the time of publication, the author, his immediate family, and/or KNA Capital Management, LLC or its clients do not have positions in any company mentioned.

Please see important disclaimers.