Can AI rebalance my portfolio?

You choose a mix, say 70% stocks and 30% bonds. Markets move, the winners grow into a larger share, and after a strong run for equities the same portfolio might sit at 82% stocks without you buying anything.

Rebalancing sells some of what grew and buys what shrank, returning the mix to what you chose. The point is risk control rather than return: an 82% equity portfolio behaves differently in a downturn from the 70% one you signed up for.

The same drift happens inside an equity allocation. Four technology positions that all did well can quietly become half the portfolio.

Measuring drift requires current prices, current share counts and a target to compare against. Assembled by hand this is a spreadsheet job, which is why it gets postponed.

Software connected to the account skips that entirely. It can state current weights, name the positions that moved most, and produce the specific buys and sells that would restore the target.

One thing it only catches if it looks inside your funds: overlap. Two funds that look like diversification can hold the same twenty companies, so the real concentration is worse than the labels suggest. Not every tool does this; Walnut, for one, measures the holdings you own directly and does not look inside funds.

Analysis and execution are different permissions. A brokerage connection may allow reading positions and nothing else, in which case an assistant can tell you precisely what to do while being unable to do it.

Where trading is permitted, orders still go to your broker and execute under their rules, at their prices, with their settlement. No assistant sits outside that.

Before relying on automation, confirm what your specific connection allows. Assuming an assistant can act, when the connection is read-only, is a common way to discover a rebalance never happened.

In a 401(k), IRA or HSA, selling to rebalance creates no tax event. Inside those accounts the decision is nearly free and the analysis is the whole job.

In a taxable account, selling an appreciated position realises a capital gain that year. A rebalance that looks tidy can carry a bill large enough to make waiting the better answer.

Two cheaper routes usually exist: direct new contributions into whatever is underweight, and take dividends as cash rather than reinvesting them. Both correct drift without selling anything.

Say you set a three-part portfolio a year ago: 60% in a US stock index fund, 30% in a bond fund and 10% in an international fund, $20,000 in total. Stocks had a good year, bonds were flat and the international fund slipped, so the account is now worth $22,400, and the stock fund has grown to $14,560. That is 65% of the account, five points over its target.

The arithmetic is the part software does in a second. At 60%, the stock fund should hold $13,440, so $1,120 has to move. The bond fund should hold $6,720 and holds $6,000, so it is $720 short. The international fund should hold $2,240 and holds $1,840, so it is $400 short. The trades that close the gap are one sell of $1,120 and two buys that add up to the same amount.

The judgement is the part that stays with you. In an IRA, those three trades cost nothing in tax and you would probably just place them. In a taxable account, the $1,120 sale realises part of a year's gain. If you were about to add $1,200 of new money anyway, putting all of it into bonds and the international fund closes most of the gap without selling anything.

There are two common rules for when to rebalance, and either is better than deciding by feel. A calendar rule checks on a fixed date, such as once a year, and trades only if something has moved meaningfully. A threshold rule ignores the calendar and acts when a holding drifts a set distance from its target, often five percentage points, or a quarter of its target weight for small positions.

Calendar rules are easier to keep, and they line up with tax planning if you pick a date late in the year. Threshold rules react to big moves sooner but need something watching the weights in between. Watching is exactly the part an assistant connected to your account does well, because it can report the gap every time you ask rather than once a year.

Whichever rule you choose, write it down before markets move. A rule chosen after a sharp drop tends to be a market call in disguise.

Ask for the current weights first, and check them against your broker statement. If the numbers do not match, nothing downstream is worth reading.

Ask what the trades would cost in realised gains before asking what the trades are. In a taxable account that figure often changes the plan.

Set a rule rather than deciding case by case. An annual date or a fixed drift threshold both remove the part of this that turns into a market call.

Capital gains treatment on sales is covered by the IRS in Topic no. 409, Capital gains and losses and Publication 550. General guidance on asset allocation and rebalancing is published by the SEC at investor.gov. Walnut is informational and is not an investment adviser. This guide is educational and not personalized investment or tax advice.