Your stock picks were fine. The trading is what cost you

You’ve had a good eighteen months. Two or three of those calls were your own, and you could name them if anybody asked.

Right. Now the part you aren’t braced for. Buy €5,000 of a US-listed share from a euro account, then sell it again. Commission and handling across both legs: about €4. That’s the number you picked your broker on. The currency conversion across those same two legs: €25. Ten of those round trips in a year come to €292.50, on DEGIRO’s published prices ↗ (opens in new tab) checked in September 2026. Sitting on the same shares for the whole year costs €2.50 in the same charges, so the gap is €290. Dated illustrations, not a recommendation.

Every euro of that gap came out of how often you touched the shares.

The same thing shows up where it was first measured properly. Researchers went through 66,465 American broker households covering 1991 to 1996 and sorted them by how often they traded. Before costs, the two groups earned much the same ↗ (opens in new tab). After costs, the busy households ended on a net 11.4% a year and the quiet ones on 18.5%. Seven percentage points, and the shares themselves barely explain any of it. One American brokerage in the 1990s, and no guide to future returns.

Overconfidence bias in investing is a busyness problem. Your picks can be perfectly good and the year still costs you money. There’s a short test further down that measures this on you. You’ll probably do badly on it. So does nearly everyone, which is the useful bit.

What is overconfidence bias, and which kind have you got?

Overconfidence bias in investing is the gap between how good your judgement is and how good you think it is. It comes in three kinds: thinking you did better than you did, thinking you’re better than other investors, and being too sure of your estimate.

All three end up at the order screen.

A 2008 review ↗ (opens in new tab) pulled the three apart. You can carry one and not the others, on the same afternoon.

You can learn to spell overplacement and still trade every Thursday. Then you have a vocabulary and the same bill. The wider set of biases sits in our guide to the ten cognitive biases costing you money. Overconfidence is the one that runs on frequency.

Why does overconfidence bias show up as trading more, not knowing more?

Because the trait has to come out somewhere, and trading is where it goes.

The cleanest evidence is Finnish. It measured overconfidence head on, instead of guessing at it from someone’s sex or their answers to a survey. With wealth, income, age and occupation held steady, overconfident investors traded more often ↗ (opens in new tab). Not better. More often.

Then there’s the loop. It matters most if you’ve just had a good run. Investors recall their returns as higher than they were ↗ (opens in new tab), and the bigger the memory gap, the more confident and the more active they turn out to be. Winners get remembered. Losers get filed. The researchers call that selective forgetting. A good month produces more trading, which produces more cost, which doesn’t feel like anything at the time.

Which of the three does the driving is still open. What’s settled is the level above. Overconfidence raises how often you trade, and every trade costs you something, visible or not.

What does your broker take on a single trade?

Start with the one bill you shopped on. Your broker takes more than the commission, and the biggest line is the one you never compared.

Europe’s markets regulator, ESMA, surveyed 27 EU neo-brokers ↗ (opens in new tab) holding around 10 million client accounts as at 2023. Currency conversion and the gap between the buy and the sell price are both on its list of how those firms make money. Neither is a number anybody picks a broker on.

Take a real, published price list. At DEGIRO ↗ (opens in new tab), buying €5,000 of a US-listed share from a euro account and later selling it works out like this. Published prices, checked September 2026; these are dated illustrations, not endorsements and not advice, and you should check the current tariff before acting.

Each line is charged on each leg: €2.00 and €12.50 to buy, the same again to sell.

The line you compared brokers on is €4. The line you didn’t look at is €25, more than six times larger.

You might file that under corner cases for people who buy American shares. ESMA’s risk monitor for early 2026 ↗ (opens in new tab) puts US stocks at around 40% of EU retail equity transactions. Of the hundred most-discussed shares on social media, one is EU-based. One.

Now give that a year. Say you hold €20,000 in shares and you make ten round trips over the twelve months. One every five or six weeks, €5,000 at a time. A good idea in February, a better one in June, and a tidy-up in October. That’s an ordinary year.

At the prices above, those ten round trips cost €290. The version of you who began the year holding the same shares and did nothing paid the €2.50 exchange connectivity fee, and that was the year’s trading and conversion cost. Anything the holdings themselves charge sits on top, for both of you. That gap is 1.45% of the portfolio, and trading is the whole of it.

Illustrative arithmetic at September 2026 prices rather than a forecast. Investment values can fall as well as rise, and what you end up with depends on how the holdings perform and what you are charged.

A different price list shows you a different number. Trade Republic ↗ (opens in new tab) charges no order commission and a one euro external settlement cost per trade. The firm trades on an exchange and doesn’t publish the spread you cross. One euro is what you can see, not what you pay. IOSCO put it in its November 2025 report on neo-brokers ↗ (opens in new tab): “even with low or no trading fees or commissions, investors who trade frequently shoulder the bid-ask spread costs”. Both firms are authorised, and you can check any provider on ESMA’s registers ↗ (opens in new tab).

Somebody was always paying for all this to feel free. Under EU market rules ↗ (opens in new tab), your broker can no longer be paid by a third party to route your order somewhere particular. Germany had its own exemption from that ban, and used it. It expired on 30 June 2026, and the ban has applied there since 1 July 2026. Germany’s regulator BaFin has told consumers what to expect next ↗ (opens in new tab): higher fees per order, fee models with monthly base charges, and new fund savings plans that are no longer free.

From BaFin’s own consumer page, in regulator German:

“Das PFOF-Verbot kann aber Folgen haben, die Sie im Geldbeutel spüren könnten.”

BaFin, consumer guidance on the payment for order flow ban, July 2026

Roughly: the ban may have consequences you feel in your wallet.

What follows is our reading, not BaFin’s. The spread IOSCO describes was always there. What the ban changes is how much of it now arrives as a fee you can see.

That’s what a trade costs. The frequency is yours.

How do you know if you are an overconfident investor?

From the inside, overconfidence feels exactly like competence. So look at what you did:

• You traded more in the months after a good run than before it.
• You haven’t opened the annual statement, but you’d say roughly how the year went.
• You’ve never held your net return up against a plain index fund.
• You sized a position on how sure you felt rather than on what you could lose.
• You acted on something the app told you, the same day it told you.

Take any of those as a prompt to measure. It settles nothing on its own. The standard test works on ranges and takes a couple of minutes. Write down eight ranges you’re 90% sure of, then count how many hold. You end up with a score.

For each of the eight items below, write down a low number and a high number, so that you’re 90% sure the true answer sits between them. Honesty about the width of your range is the whole test here.

1. The population of the European Union on 1 January 2026.
2. How many of the 27 EU member states used the euro in September 2026.
3. The year the euro launched as a currency.
4. The European Central Bank’s deposit facility rate on 2 September 2026.
5. Euro-area annual inflation in August 2026.
6. How many US dollars one euro bought on 1 September 2026.
7. The Bank of England’s Bank Rate on 1 September 2026.
8. The total area of the European Union, in square kilometres.

How did you do?

Here are the answers, each with its date attached so you can go and check them yourself. Score yourself against the dates given, not against today’s numbers; four of these move.

Scoring. Seven or eight inside your ranges is what a genuinely 90%-confident person should manage. Four, five or six means your ranges are tighter than your accuracy earns. Three or fewer is the documented typical result, and that’s overprecision, measured. Eight out of eight with enormous ranges isn’t a win either; you gave yourself so much room that you told yourself nothing.

In the research, 90% ranges hold the right answer less than half the time ↗ (opens in new tab). A bad score puts you with almost everybody, and it says nothing about your intelligence. All it reads is how wide the range in your head is. It’s why “this is definitely undervalued” feels like a fact when it’s a guess.

If you arrived after a bad quarter rather than a good one, you’re not in trouble. Before costs, the busy and the quiet earned much the same, and that cuts both ways. Whatever went wrong, it probably wasn’t your eye for a company.

Are men more overconfident investors than women?

Yes, in one famous study, and it gets quoted well past what it measured. People remember it as proof that women are better investors. The study counted trading.

In over 35,000 American brokerage households, from February 1991 to January 1997 ↗ (opens in new tab), men traded 45% more than women, and that trading cost men 2.65 percentage points of net return a year against 1.72 for women.

The finding is real. It’s also American, thirty years old, and about brokerage clients rather than about people.

The picture since hasn’t stayed still. A 2024 study of the general US population ↗ (opens in new tab) measured overconfidence about financial knowledge, and women scored higher on that one. Its authors put the gap down to differences in financial knowledge rather than to anything people are born with.

Sex was a stand-in for overconfidence, used back when overconfidence was hard to observe directly. It stopped being hard. You measured yours eight questions ago.

What does a single trade cost a European investor beyond the broker’s charges?

Two more bills, and neither of them is on the price list.

What does the tax office take before you have made anything?

In six European markets, buying shares is taxed on the way in. The tax office doesn’t wait to see if you were right.

Each rate came from the tax authority or the statute itself, read in September 2026. This isn’t all of Europe: a market missing here is one we didn’t check, not one that charges nothing.

Buying €5,000 of shares in an Irish-registered company hands the State €50 in Irish stamp duty on shares ↗ (opens in new tab) before the trade has done anything at all. That’s more than the entire €29 round trip above, commission, handling, currency conversion and all. The charge follows the company. A German buying Irish shares pays it in full, and the US share in the earlier example attracts none of it.

Not every Irish holding pays it. A listed Irish company worth under €1 billion on the previous 1 December can be exempt between 2026 and 2030 ↗ (opens in new tab), though only once it or its market has told Revenue, which publishes the list. Units in Irish-domiciled funds and ETFs sit outside the charge ↗ (opens in new tab).

Belgium is the odd one out in that table. Its tax attaches to the buyer instead of the company, so someone living in Belgium pays the 0.35% on the company shares they buy ↗ (opens in new tab), the US ones from the example above included, through a broker abroad just as much as a Belgian one. The sale is charged too, so that €5,000 round trip costs €35.00 in stock exchange transaction tax on top of the €29 of broker costs. The 0.35% is the catch-all band, so it does not price the classes the statute lists separately, among them shares in a Belgian regulated real-estate company and units in funds and ETFs recognised under Belgian or EEA law.

Selling has its own bill, and the country you live in decides what’s on it. If you want that bill smaller, the subject is tax-efficient investing. Sell at a profit in Ireland ↗ (opens in new tab) and 33% of the gain above a €1,270 annual exemption goes in tax. In Germany ↗ (opens in new tab) the rate is 25%, plus a solidarity surcharge of 5.5% charged on the tax itself ↗ (opens in new tab), plus church tax if it applies to you. It bites above a €1,000 saver’s allowance ↗ (opens in new tab) each year, €2,000 for jointly assessed couples.

The Netherlands works differently. Box 3 is where the Dutch tax system files savings and investments. There is no tax on the gain you realise. Instead, box 3 taxes the holding ↗ (opens in new tab): for 2026, 36% of a deemed 6.00% return on investments, above a tax-free amount of €59,357 per person. That works out at about 2.16% a year of the investments above that amount.

A Dutch reader who trades weekly and one who has not logged in since 2021 get the same bill on the same assets. In the Netherlands, then, churn shows up in fees and spread. The tax bill doesn’t move. It’s still not a reason to trade more; the fees and the spread carry on regardless.

Of the markets in this guide, Slovenia has gone furthest in the other direction. The rate starts at 25% and falls with every five years you hold ↗ (opens in new tab): 20% after five, 15% after ten. Sell after fifteen years and the gain is exempt outright. Sitting still is written into the tax code. Whether a holding is worth keeping for fifteen years is a separate question from what the tax does.

What does the swap itself cost?

About 3.3 percentage points over the following year, in US brokerage accounts between 1987 and 1993.

Suppose the dealing costs were zero. The swap still loses. In a separate set of discount-brokerage records, 10,000 accounts, the shares investors bought went on to do worse than the shares they had sold ↗ (opens in new tab) by that margin. And no, bad timing doesn’t explain it. The same work checked.

Taken as a whole, the busy and the quiet held much the same quality of shares before costs. Trade by trade, the average swap still moved into something that then did worse than what it replaced.

Every one of those swaps felt like a decision at the time. You probably remember the reasoning for two of them.

If your next thought is that you’d have timed it better, that question is in does market timing work.

Do trading apps make overconfidence worse?

Your app makes money when you use it, and using it means trading. Gamification is the polite word for the machinery that keeps you tapping. Regulators have stopped guessing at what that machinery does to people. Now they run experiments on it.

The world’s market regulators, working together as IOSCO, list the practices by name ↗ (opens in new tab): streaks, leaderboards, push notifications. IOSCO also reviewed the research on gamified design, badges and celebratory messages among it. It records that these “can provide immediate gratification for investors, thus potentially encouraging them to trade more frequently when it may not be appropriate for them to do so”.

That’s the world’s securities regulators, in committee prose, telling you the app is built to keep you trading.

The FCA, which regulates UK markets, put that in a controlled experiment ↗ (opens in new tab) with over 9,000 consumers and published the results in 2024. Push notifications raised the number of trades by 11%. A points-and-prize-draw feature raised it by 12%. The share of risky trades went up by 8% and 6% respectively.

Ours arrives at 08:14, in the same stack as the parcel firm and a reminder about the bins.

Design can also work the other way. The AFM in the Netherlands has found that platform design shapes how investors behave ↗ (opens in new tab). France’s AMF ran a laboratory test on gamified design ↗ (opens in new tab), and the effect came out fussier than the headlines suggest. Achievement badges tied to risk-taking moved people. Virtual confetti didn’t shift their risk-taking at all.

The scoring is what gets us. Confetti is only decoration.

What actually reduces overconfidence bias?

More reading won’t help, because the trouble sits in your memory. One fix has an experiment behind it, and it runs backwards. Investors who looked up their real past trades ↗ (opens in new tab), instead of recalling them, came out less overconfident and planned to trade less.

So before anything else, go and read your statement. The real one, with the numbers on it, covering the whole stretch, including the months you’d rather skip.

Then set the rules, while you’re not in the middle of a good run:

1. Set a trade budget. Decide the maximum number of trades you’ll make this quarter, in advance, and write the number somewhere you’ll see it. A note on the fridge counts. The evidence measured churn. A planned rebalance or a deliberate change of position is a different thing.
2. Sleep on anything the app told you about. If a notification is what put a discretionary buy or sell idea in your head, it can’t be acted on the same day. Corporate actions, margin calls and needing the cash are not that.
3. Cap the holding. Decide the most any one big hunch is allowed to cost you, before you have the hunch.
4. Keep a decision journal with a range column. Write the call and the range you’d bet real money on, then score it later. It’s the only one of these that puts a number on your overprecision.

Underneath all four sits the dull option: a standing monthly contribution into something broad. It makes trades that aren’t decisions, so they can’t be overconfident ones. That argument runs in full in passive versus active investing, and it takes more nerve than it sounds.

If you’re carrying expensive debt, or you’ve no cash to fall back on, both of those come ahead of anything on this list.

Go and pull up your real returns, net of everything, over the whole period you’ve been holding. Put them beside a plain global index fund over that same period. Two years proves nothing either way; the longer the stretch, the less of it is luck.

Those good eighteen months were probably real. What you don’t know yet is what they cost you.