Robo-Advisor vs Doing It Yourself: The Honest Trade
Last updated August 2026
Short answer
Doing it yourself is cheaper, and the saving is real: you skip the advisory fee, commonly around 0.25% a year, which is about $250 on $100,000, and pay only the funds' own expense ratios. What you take on is four jobs: choosing an allocation, rebalancing, harvesting tax losses in a taxable account, and not selling in a bad month. The first is easy, the second is the one people skip, the third is fiddly, and the fourth is hard for everyone. If you would genuinely not rebalance, the fee is buying something. Walnut is not an investment adviser.
This comparison usually gets framed as thrift against laziness, which is not useful to anyone. The honest version is narrower: a robo-advisor sells four specific jobs, three of which you can do in about two hours a year, and one of which nobody does reliably. Whether the fee is worth paying depends entirely on which of those jobs you would actually perform, and the only person who can answer that is you, honestly.
Side by side
| Robo-advisor | Doing it yourself | |
|---|---|---|
| Who picks the allocation | The platform, from your questionnaire | You |
| Who buys the funds | The platform | You, a few times a year |
| Who rebalances | The platform, automatically | You, if you remember |
| Tax-loss harvesting | Automatic on some platforms, in taxable accounts | By hand, and the wash-sale rule makes it easy to get wrong |
| Advisory fee | Commonly around 0.25% a year | None |
| Fund expense ratios | You pay them | You pay them, and they can be lower |
| Who holds the money | The platform | Your own broker |
| Time it takes you | Almost none after setup | An hour or two a year, honestly done |
What you save, in dollars
The advisory fee only, at the category-standard 0.25%. You pay fund expense ratios in both columns, so this is the actual difference. The ten-year figures assume a steady balance and therefore understate it for anyone still contributing.
| Balance | Saved per year | Saved over 10 years |
|---|---|---|
| $25,000 | $63 | $625 |
| $50,000 | $125 | $1,250 |
| $100,000 | $250 | $2,500 |
| $250,000 | $625 | $6,250 |
| $500,000 | $1,250 | $12,500 |
At a starter balance this is a rounding error and the convenience is obviously worth it. By the time the account matters, the same rate is a real annual number. That asymmetry, not the rate, is the argument. Full arithmetic in robo-advisor fees explained.
The four jobs you take on
1. Choosing the allocation
What it involves. One decision, then rarely revisited. A three-fund portfolio of total US market, total international, and bonds covers most cases.
Where it goes wrong. Overthinking it. The gap between a reasonable allocation and an optimal one is much smaller than the gap between investing and not.
2. Rebalancing
What it involves. Once or twice a year: check the weights, sell what has run, buy what has lagged.
Where it goes wrong. This is the one people skip. It requires selling your best performer, which is exactly what nobody wants to do, and it is the main service the fee buys.
3. Tax-loss harvesting, if the account is taxable
What it involves. Meaningful work: identify losses, sell, buy a similar-but-not-identical fund, and stay clear of the wash-sale rule for 30 days either side.
Where it goes wrong. Getting the wash-sale timing wrong disallows the loss. In an IRA this job does not exist at all.
4. Not doing something silly in a bad month
What it involves. None, in theory.
Where it goes wrong. The hardest one, and no software solves it either. A robo-advisor helps only insofar as friction slows you down.
Job two is the crux. Rebalancing means selling the thing that has done best, which every instinct resists, and the evidence that you will do it is whether you have. If your portfolio has drifted untouched for two years, that is your answer, and it is not a character flaw. It is worth paying 0.25% to have it handled.
Get a recommendation for your situation
Walnut is the AI that knows your portfolio: ask anything in plain English, research any fund, and get an honest second opinion. On the broker you already use, read-only, and you approve every trade. Walnut is not a registered investment adviser.
The honest failure mode of doing it yourself
It is not picking a bad allocation. Nearly any sensible mix of broad funds will do the job over a long horizon, and the difference between a reasonable allocation and an optimal one is far smaller than most people fear when they are choosing.
The failure mode is drift plus inaction. A portfolio set up sensibly in one year and never touched becomes something else entirely by the sixth, usually far more concentrated in whatever ran hardest, and usually right before that thing stops running. Nobody decides to take that risk; they just do not rebalance, and the portfolio takes it on their behalf.
The second failure mode is worse and rarer: selling in a bad month. No product prevents it, including a robo-advisor, though friction and not looking both help. If you know you have done this before, that is real information about which column you belong in, and it is worth more than any fee comparison.
A workable middle: automate the contribution, not the portfolio
One arrangement gets very little attention and solves most of the problem cheaply. Set up an automatic recurring contribution into your own brokerage account, and each time direct it at whichever holding is furthest below its target. That is cash-flow rebalancing, the same technique the platforms use, and it requires no selling at all.
It has two real advantages over doing it by selling. In a taxable account it creates no tax event, because nothing is sold. And it turns rebalancing from a decision you have to feel comfortable making into a consequence of a transfer you already scheduled. Its limit is that it only works while you are still contributing meaningfully relative to the portfolio size, so it fades as the balance grows and eventually you do have to sell something.
The middle option people forget
The choice is usually presented as pay for automation or do everything yourself, and there is a third position: keep your own broker and your own holdings, and use software to do the checking rather than the deciding. The four jobs split cleanly. Knowing your weights have drifted is a data problem. Acting on it is a decision.
An assistant connected to your existing brokerage account can tell you how far each position has moved from target, how the whole portfolio has done against the S&P 500, and where you have become concentrated, without taking custody of anything and without an advisory fee. It cannot rebalance for you, and it will not stop you selling in a bad month. That is the honest boundary of what this route solves. See alternatives to a robo-advisor for the field, and how to rebalance with AI for the mechanics.
Which to choose
Pick the robo-advisor if you are starting from cash, want a diversified portfolio without designing one, and would not rebalance on your own. How to choose one covers what actually varies between platforms.
Do it yourself if you want to choose what you own, your balance is large enough that the fee is a real number, and you will genuinely put an hour into it twice a year. And if the honest answer is somewhere between, are robo-advisors worth it works through the situations one at a time.
FAQ
How long does rebalancing actually take if I do it myself?
For a three-fund portfolio, under an hour once you have done it once: check the current weights, work out what is over and under target, and place two or three trades. The work is trivial. The reason people do not do it is not effort, it is that rebalancing means selling whatever has done best, which every instinct resists.
Can I automate rebalancing in my own brokerage account?
Partially. Some brokers offer target-weight tools or automatic investing that directs new contributions to whatever is underweight, which is most of the benefit without any selling. Full automatic rebalancing that sells the overweight asset is rarer outside managed products, and it is the specific piece the advisory fee buys.
What is the cheapest DIY portfolio?
A single broad total-market fund, or three funds covering US stocks, international stocks and bonds. Expense ratios on the largest of these are a few hundredths of a percent, so the all-in cost is a small fraction of a robo-advisor's. The portfolio is not meaningfully different from what a robo-advisor would build you.
Do I need to rebalance in a retirement account differently?
The mechanics are easier there, because selling inside an IRA or 401(k) triggers no tax at all. In a taxable account rebalancing by selling creates a taxable event, which is why directing new contributions to the underweight asset is the better first tool there.
Is a robo-advisor better than doing it yourself?
Financially, doing it yourself is cheaper, because you skip the advisory fee and pay only the fund expense ratios. Practically, it depends on whether you will actually do the four jobs the platform does: choose an allocation, rebalance, harvest losses in a taxable account, and avoid selling in a bad month. If you would not rebalance, the fee is buying a real service.
How much do I save by not using a robo-advisor?
The advisory fee, which at the category-standard 0.25% is about $250 a year on $100,000 and about $1,250 on $500,000. Over ten years at a steady balance those are roughly $2,500 and $12,500. You still pay the fund expense ratios either way, so the saving is the advisory layer only.
Can I just buy the same funds a robo-advisor buys?
Largely yes, and this is the uncomfortable part of the comparison. Robo-advisor portfolios are built from broad, cheap index ETFs that anyone can buy in a normal brokerage account. What you cannot replicate by buying them once is the ongoing rebalancing and, in a taxable account, the automated loss harvesting.
What is a three-fund portfolio?
A total US stock market fund, a total international stock fund, and a broad bond fund, held in proportions matched to your horizon and risk tolerance. It is the standard DIY answer because it gets you diversified in three purchases, and it is close to what many robo-advisor allocations look like underneath.
How often do I need to rebalance if I do it myself?
Once or twice a year is a common approach, or when an allocation drifts more than a set amount from its target, often five percentage points. What matters more than the exact rule is having one, because the failure mode is not rebalancing badly, it is not rebalancing at all.
Is tax-loss harvesting worth doing by hand?
In a taxable account it can be worth real money, and it is genuinely fiddly: you sell at a loss, buy something similar but not substantially identical to stay invested, and avoid repurchasing the same security within 30 days either side or the loss is disallowed. In a retirement account it is worth nothing, because there are no taxable gains to offset.
What is the biggest risk of doing it yourself?
Not the allocation. It is drift and inaction: a portfolio that quietly becomes far more concentrated than you intended because you never rebalanced, and a decision to sell at the wrong moment. Both are behavioural, and they are what the advisory fee is really priced against.
Can AI help if I do it myself?
It can do the checking, which is the part people skip. An assistant connected to your own brokerage account can tell you how far each position has drifted from target, how the portfolio has moved against the S&P 500, and where you are concentrated, without taking custody of anything. It cannot make you act on the answer. Walnut works this way and is not an investment adviser.
Related articles
Walnut is informational and is not an investment adviser, and nothing here is investment advice or a recommendation of any strategy. Fee figures are widely published category rates used to illustrate the arithmetic; the dollar amounts assume a steady balance and no growth.