Three portfolios against true-flation.

The robo-adviser says plus 30. CPI says plus 20. I measure against true-flation.

Everything in life is a trade. You trade your time for money, and the money is supposed to preserve that time until you need it. So every time someone shows me a return I ask the same thing: measured against what? Is it a return in kronor, or is there something more we have to beat? CPI says the krona lost 20 percent in five years. CPI only measures consumer prices. The money supply grows faster than that: since 1971 there is 37 times more money in the system, and that affects all prices, not just the ones in the basket.

That is why I measure against true-flation: how the gold price has moved against the krona over the last 55 years, plus the new gold being mined, 11.6 percent a year. That is the average, not a good gold period. Over five years that makes plus 70 percent just to break even. That is quite a lot of money when you think about it. The result first: the robo-adviser reached 30, the average Avanza customer 13, the rule-based portfolio 61. Measured against the bar that is minus 24, minus 33 and minus 2. Only the portfolio with 45 percent bitcoin made it over.

The first let a robot handle it. The second handled it himself. The third followed rules.

The robo-adviser is Lysa's broad portfolio with 50 percent equities and 50 percent bonds, the mix that corresponds to a typical balanced management mandate. The numbers are Lysa's own, after fund fees, and I have deducted Lysa's own fee of 0.27 percent per year. The Avanza customer is not a person but an average, Avanza's own line "return, average account" in every quarterly report. The rule-based portfolio is our model Balanserad, run in our backtest on daily prices: three assets, one band, one rebalance when the band breaks, and tax paid every year on what was sold. It is a backtest, not a customer portfolio.

It took 70 percent to stand still. The robo-adviser barely got halfway.

The bar is not a gold price. It is gold's average pace over 55 years plus the new gold mined every year, in other words the pace money needs to grow at to keep its share of what is scarce. Over five years: plus 70 percent. The robo-adviser with half equities reached 30, the average customer 13. The robo-adviser with only equities reached 66. The rule-based portfolio Balanserad reached 61 after fees and annual tax, Försiktig 47, Tillväxt 72.

The uncomfortable part for the third column is here: the robo-adviser with 100 percent equities and Balanserad ended up the same distance from the bar, two percentage points below, with the same largest drawdown. Only the portfolio with 45 percent bitcoin cleared the bar after tax. What separates Balanserad from Tillväxt is the tax: 30 percent on gains that to a large extent are the krona's fall. The only protection is to own directly and sell as little as possible, which is what the rules do: five sell orders in five years.

The robo-adviser measures risk as swings in kronor. It is the krona that swings.

Volatility is the standard deviation of the portfolio's value in kronor. It measures how much the number on the screen wobbles, not whether you can buy the same thing with the money when you need it. The measure is built for a world with stable money. Since 1971 the US money supply has grown 37 times, and in five years the krona lost a fifth. In such a world the portfolio that swings the least often carries the greatest risk, the risk that purchasing power disappears, without anything showing up in the risk measure.

Counting risk as volatility is a way of looking at things that assumes money keeps its value. That is the biggest problem in our financial system right now: it does not, and the pace is increasing. Then the old theories about how to measure risk start to crack. Here are the same five years measured in the three ways that matter to a saver: how deep it fell, how long it stayed under water, and where it landed against the bar. Volatility is included because it can be compared with the robo-adviser's language.

You did not beat the robot. The average Avanza customer got half of the robo-adviser.

I have said in three analyses that you should manage your own money. Here is the price of that. The average Avanza customer got 13 percent in just under five years, against the robo-adviser's 30 with half in bonds, and lost to the Stockholm stock exchange three years out of five.

It is not because he is stupid. It is because he does what everyone does: buys what has gone up, sells what has gone down, and keeps ten percent in cash waiting for a moment that never comes. And the average flatters. In 2025 the average was 7 percent and the median 2. Half of all Avanza customers got two percent or worse that year.

Bitcoin fell 60 percent. The rule-based portfolio held up anyway, and not thanks to bitcoin.

Anyone who calls a portfolio with 30 percent bitcoin a bitcoin portfolio has the right to ask how it handled 2022. The answer: worse than the robo-adviser and better than Svensson. It was the gold and the cash that held, and the robo-adviser's bond part fell as much as its equity part that year. Then what always happens after such a year happened: what fell the most rose the most. In 2023 bitcoin returned 160 percent in kronor. The rules sold gold and bought bitcoin on the way down, and sold bitcoin and bought gold on the way up. Four rebalances in five years. None of them required an opinion.

However many equities you choose at the robo-adviser, you do not reach the bar. But you get close with all of them.

Lysa lets you choose anything from zero to a hundred percent equities. I ran the whole ladder over the same five years. With only bonds you stood still in kronor, plus 0.1 percent after fees, that is minus 41 against the bar and minus 17 against CPI. With only equities you reached 66 percent, two below the bar, with a 21 percent largest drawdown. Every step up in equities gave roughly 7 percentage points more over five years and one percentage point more in volatility.

It is a straight line, and it says two things. The robo-adviser does exactly what it promises: the market's return for 0.37 percent. And the market's return was not enough, not even with everything in equities, to keep pace with the money supply over these five years. So equities are not the answer either, not as an index. In the analysis about true-flation I showed why: it is a small number of large companies that pull, and the large majority of companies do not reach the bar. A broad exposure that does not happen to be heavy in the ten largest tech companies gives no excess return measured against true-flation.

The cautious robot: lowest swings, longest under water, largest loss against the bar.

Anyone who wants calm chooses 30 percent equities, or only bonds. This is how it went: 30/70 gave 18 percent after fees over five years, against 20 in inflation. Zero in real terms, and 709 days under water. Only bonds gave 0.1 percent, 933 days under water, and minus 41 against the bar. The portfolio built not to swing was the one that spent the longest below what the customer put in.

The risk you take when you buy lower swings is that you lose purchasing power. The money you have left in kronor and öre does not pay for as much after these five years. In the analysis The history of fiat currencies I showed that whoever saved in government paper after tax has lost 43 percent over a hundred years. Seventy percent bonds is not a low-risk portfolio. It is a certain, slow loss with a calm chart, and a calm chart is exactly what the volatility measure rewards.

Management is a commodity now. What remains is what you measure in, and what you own.

Management in Sweden today sits in three layers. The major banks' traditional mandates charge around one percent or more. The robot-assisted managements, like Avanza Sigma from 31 August 2026, charge just under the percent, according to Avanza half of traditional management, and pulled in SEK 710 million in a month. The robo-advisers, which have found their own niche, sit well below half a percent. Our own model is a fourth kind: no management fee at all, a fee per order. All of the first three deliver roughly the same thing, index minus fee. You think you are buying judgement. You are buying index.

What speaks against this comparison, said before anyone else says it.

This is a snapshot of five years, and it depends on when you start and stop measuring. I chose January 2022 because it gives five full years. Had I chosen the bitcoin top two months earlier, Balanserad would have given 56 percent after fees and annual tax. From the bottom in November 2022: 96. From January 2024: 37. The worst window was still enough, but it is a window, not a law. The median across all five-year windows starting 2018 or later is 20.6 percent a year, the worst 10.4.

The rule-based portfolio is a backtest, not a customer portfolio; the robo-adviser's and the Avanza customer's numbers are real. Gold did the work, not bitcoin: gold plus 134 percent over the period, bitcoin plus 39. Over the next five years it may be the opposite, or neither. The rule-based portfolio's series end in July 2026, the robo-adviser's in October; bitcoin and gold have risen since July, so the row for 2026 changes when the series is extended. And I am not a neutral reader of the third column. That is why every number comes with a source, so that you can calculate it yourself.

Three sentences to take with you

Everything in this analysis can be checked: the sources are below, and the backtest's data and rules are open. If you see something that does not add up, or have an angle I missed, say so. I do not want this to be read as a truth I cling to, but as a comparison that can withstand being challenged.