My Monthly Income Portfolio Drifted 14% From Target — the Robot Caught It, My Spreadsheet Didn't

12:45 in the morning, and I'm in bed with a tablet propped on one knee, scrolling through two years of rebalancing records for a dividend portfolio I have personally, manually, lovingly maintained since 2022 — when I found the gap.

Two consecutive months, November and December of 2024, where my spreadsheet said "rebalanced" but the brokerage's trade history said otherwise. No trades. No orders. Nothing. I had recorded intentions in November and December, then life happened — a work crunch, then the flu, then the holidays — and by January my spreadsheet's notes had hardened in my memory into facts. I remembered having rebalanced. The database disagreed.

And in those two skipped months, with one REIT position ripping higher and my bond allocation dragging, the portfolio drifted from a 60/40 income-tilt to something north of 72% equity income exposure. My monthly volatility, which I had expressly engineered this portfolio to smooth out, went up approximately a quarter without my feeling a thing.

the portfolio, briefly

The setup: about $310,000, built to produce monthly income. Three dividend-heavy equity ETFs, a REIT fund, an intermediate bond fund, and a short-duration treasuries position. Target weights, set in 2022 after alot of coffee: 58% equity income, 14% REITs, 28% bonds and treasuries, give or take. Every position pays monthly or quarterly; the cash flow averages about $1,540 a month, and I reinvest everything that lands in a quarter where I don't need it.

My method, since the beginning: first Saturday of the month, spreadsheet open, weights computed by hand, trades placed manually when any position drifted more than 3 points from target. Four years of first Saturdays. I have receipts. I have an awfully satisfying chart. What I did not have, it turns out, was a system that worked when the first Saturday didn't.

what the comparison actually showed

I exported four years of the broker's transaction history and reconciled it against my spreadsheet, line by line, an exercise that took one evening and produced one large feeling. Of 48 scheduled rebalances, I had executed 44. The four missed ones — two in late 2024, two scattered — accounted for nearly all of the drift I'd been blaming on "the market being weird this year."

the bias hiding in my own method

Worse, my manual process had a systematic bias. When I rebalanced by hand, I consistently deferred selling the position that had run up most — the REIT, mostly — because selling it meant realizing gains in my taxable account and sitting with a capital gains tax bill. So I'd rebalance everything around it, trimming bonds to buy equities and calling the result "rebalanced." My spreadsheet tracked my target weights; it never tracked how frequently I'd bent the rules to avoid the tax form. The drift wasn't just the missed months. It was baked into the method.

By the time I ran the full audit, actual versus target was 14 points off on the REIT sleeve. Fourteen. My 3-point trigger had become decorative.

what the "robot" turned out to be

Lemme be honest about what I was avoiding. My broker has had automated rebalancing tools for years — I'd dismissed them as overkill for a portfolio I enjoyed managing. The tools do less magic than I imagined: you set target weights and tolerance bands, and the system flags drift beyond the band, or optionally executes trades to bring weights back. No AI. No wizardry. A tolerance band and an alarm clock.

alerts first, trading second

I enabled alerts first, trading second — a configuration I now recommend to everyone. Alerts only: the system emails me when any position breaches its 3-point band, on any day, not just first Saturdays. After two months of watching those alerts fire, I saw my own bias in data: the REIT alert fired eleven times before I acted, cuz acting meant the tax conversation. The robot didn't have feelings about Form 1099. It just measured distance from target.

Then I turned on execution for one sleeve — the bond and treasuries positions, where trades are tax-boring — and kept the equity income and REIT sleeves manual, but now with a written rule I cannot bend: when an alert fires, I either trade or write one sentence in the spreadsheet explaining why not, dated. "Taxable gain too large this year, scheduled for January" is a decision. Silence is drift.

the tax piece, finally dealt with like an adult

The honest fix for the REIT overweight required selling about $18,000 of it, realizing approximately $5,900 of long-term gains. I did it in December, deliberately, alongside a review of gains I'd already taken elsewhere — and it turned out I'd been leaving money on the table in the other direction, too. Two positions had losses I'd carried for a year without harvesting because my first-Saturday ritual never included a loss report. Selling the overweight and pairing it against the harvested losses cut the net taxable gain to about $2,100. The tax bill for fixing four years of drift: slight than the tax bill for one year of pretending I'd rebalanced.

That's the sentence I'd tattoo on the whole experience, if tattoos came in spreadsheet font: avoidance has a price too, its just invoiced later, with interest.

hand versus machine, the honest ledger

Do I think manual rebalancing is worthless? No. Hand-computing weights every month for four years taught me this portfolio's rhythm — which positions lead, how the dividends cluster, what a normal month of drift looks like. That intuition caught the December 2024 gap in the first place, when a weight I remembered as 14% read 19% on the tablet at 12:45 AM. The machine doesn't create understanding; it just doesnt get the flu.

The system now: alerts run every day, the boring sleeve rebalances itself, the equity sleeves get monthly manual attention with mandatory written excuses, and a quarterly audit reconciles the broker's actual trade history against my records — because my spreadsheet records what I meant to do, and the database records what I did, and the distance between those two documents is exactly where the fourteen points were hiding.

The portfolio yields what it yielded. The risk is ultimately what I designed it to be. And the first Saturday of every month is still mine — I just don't ask it to be the only line of defense anymore.