I Spent a Career Managing Money, Then I Let an AI Test My Life's Work

Retired, with nothing left to defend, I finally ran the test on my own strategy that I could never afford to run while my income depended on the answer.

I'll start with the ending, because it's the straight way to tell it: The most useful thing I carried out of a long career managing money was a quiet distrust of my own track record.

Let me tell you how I came by it — and why I'd put the same question to you. It begins with how sure I once was.

I started in 1986 as a trader in a bank’s trust department and spent the 1990s and early 2000s as an equity-strategy partner—fifteen years, earning about a point a year over the S&P 500. I say it that plainly, because rounding it up to “we beat the market” is the small self-flattery this essay is about. Later I ran my own firm and after that served as one of Mitch Zacks' early regional vice presidents, selling his earnings-revision approach for a living. I believed in it. Then, in 2009, I left the business.

One major reason was that I could no longer square it with myself that I was adding value in a market that had come unmoored. I spent the next fifteen years in a different industry—still the same job at heart: keeping the customers who trusted us.

Last year, I reactivated my CFA charter. With nothing to sell and no book of business to protect, I finally did what no working manager can afford while his income depends on the answer. I took the strategy I still invest my own money by—Peter Lynch's improving story, Hilary Kramer's cheap and overlooked, the earnings-revision discipline I'd sold for Mitch Zacks—handed it to an AI, and told it to tear the thing apart.

It did. I tested the strategy as a skeptical allocator would, correcting for survivorship bias, using point-in-time data, measuring returns before fees, and comparing the results with random selections from the same investment universe. The strategy showed no advantage over chance.

A plain index fund beat it. I was in ordinary company: the SPIVA scorecard finds a large majority of active large-cap managers trailing the S&P 500 nearly every year, and close to ninety percent over fifteen years. Of the few who reach the top half, almost none stay—S&P's persistence data puts their odds worse than a coin flip. What I'd taken for skill was mostly strong years in a rising market and an ordinary basket of names.

Humility cuts both ways. What I tested was my own crude, mechanical fusion of these ideas, not the careful processes the originals run—the ranking method I sold for Mitch Zacks is far more sophisticated than my shorthand, and I still believe in it. What was humbled was my implementation and my certainty, not Lynch, Kramer, or Zacks. I learned only that I was partly fooling myself, and that the finding out was worth more than the belief.

Have you ever genuinely tried to disprove your own edge?

Ask any of us whether we run a clean, transparent shop and we'll say yes—and mean it. We disclose the fees, we don't lie, we send the quarterly letter. All true. But that's honesty with the client. The rarer, harder discipline is honesty with yourself.

I know why I never did it, because for twenty years I lived on the other side of it. Upton Sinclair said it plainly: It is difficult to get a man to understand something when his salary depends on his not understanding it. Mine depended on not asking. Now it doesn't.

Ask yourself these five questions for a more complete investment self-honesty:

• Have you benchmarked against the right index, net of your fees, across multiple market regimes?
• How much of your track record is one or two positions—or one market?
• Have you ever run your picks against random selection in your own universe?
• Does your record include the accounts and strategies that failed—or only the survivors?
• Can you name the exact conditions under which your edge would stop working?

I ask these as one who, for most of my career, couldn't have answered them either. If you can, you're ahead of where I was. If you can't, it doesn't mean your edge is false — only that you've been believing it rather than testing it.

Our industry is uneasy about AI coming for money management, and analysts feel it most sharply. It's the wrong worry. Used the way I used it, AI didn't do the analyst's job — it did the drudgery that was never where the value lived: the survivorship math, the null tests, the patient checking of ten thousand names without getting bored. What it hands back is room for the judgment that actually matters. For a solo manager it's a research team he could never afford; for an analyst, relief from the grunt work, not a threat to the craft.

AI won't replace the manager or the analyst. The advantage goes to whoever uses it to check their own work—and is unafraid to say what it found.

And here is the irony my audit handed me. By proving the mechanical part was beta, it proved the human part—the read on the story, the catalyst, the management, the things a model can't smell—is where the edge lives, if it lives anywhere. The machine didn't diminish the analyst; it pointed straight at them.

Here's the objection I raised myself for years: admit you might be lucky and you lose the account to the manager down the street promising alpha. I sold that dream myself, and I won't pretend I didn't. But it misreads the sale. You don't market the confession — you market the proof. A proven edge, or a genuinely proven discipline, is worth far more than a merely believed one to any client worth keeping.

Underneath it sits a principle I learned in the market and relearned over fifteen years, keeping customers who could have walked. In a portfolio the math is unforgiving: Avoiding catastrophic losses matters more than picking winners, because compounding punishes a wipeout far more than it rewards a windfall. Retention is the same law in different clothes—you grow not by winning the flashy account but by not losing the people who already trust you. Sober expectations survive a drawdown; oversold ones don't.

Defense is the edge—for the portfolio and for the practice.

You might say I've been out of the game since 2009—what do I know about today? Fair. But I don't think these principles expired; if anything, they're truer now, and my distance from the fee game is what lets me raise them without an agenda. I'm not after your clients. I have nothing to sell but the question I finally worked up the nerve to ask myself. So point the mirror at your own strategy first. The manager of the next decade won't be the one with the best story. It'll be the one who was willing to try to prove himself wrong, and honest enough to tell his clients what he found.