You are ready to trade more next year. Two trades stand out: a $300 gain and a $200 gain. You remember getting them right.
Then the year-end statement supplies the part you left out: a $700 loss and $30 in trading costs.
Suppose your US-dollar account started with $10,000 and those trades and costs were its only changes. Does the full record support trading more?
Count the losses as well as the wins
Overconfidence means being more certain than the evidence supports. Here, the mistake is treating two wins as proof that more trading will pay off.
The FOMO example showed how selected winners mislead. The same problem appears when you select from your own trades.
The two wins add to $500. The loss takes away $700. Its bar is longer than both winning bars combined. You can win more often than you lose and still lose money.
Gross means before trading costs: $500 − $700 = −$200. Each completed buy-and-sell trade costs $10 in this example. That made-up amount includes all trading friction, such as the bid-ask spread, the gap between buying and selling quotes. Three trades cost $30.
After costs, the account ends at $10,000 − $230 = $9,770. Its whole-account return is −$230 ÷ $10,000 = −2.3% for the year, before inflation.
Even if trading had cost nothing, you would still have lost $200. Cheaper trades cannot repair an incomplete scorecard.
Cash and investments you still own belong in the account value, including any unsold losses. An annualized return on one quick winner leaves the rest of the account out.
What the brokerage evidence showed
Barber and Odean's 2000 study examined common-stock holdings of 66,465 households at one US discount broker, using records from 1991–1996. They grouped households by how much of their stock portfolio they traded.
| Historical comparison | Annual return |
|---|---|
| Most-active fifth, after costs | 11.4% |
| Market benchmark | 17.9% |
Annualized USD total returns for February 1991–January 1997, before inflation. The benchmark weights NYSE, AMEX and Nasdaq stocks by market value and carries no household trading costs.
Before costs, the groups' returns were similar. Greater activity failed to produce enough extra return to cover its cost. Trading had to pay for itself. In this sample, it didn't.
Commissions were substantial in that era, so this gap is no estimate of trading costs now. Zero commissions still leave spreads, possible taxes and the risk of a poor decision.
The 11.4% describes a historical group whose members chose how much to trade. It cannot forecast your return or isolate what each extra trade caused.
Activity can feel like control
In a lottery experiment published in 1975, Ellen Langer found that people who chose their tickets asked more, on average, to sell them than people handed tickets. The choice did not improve their odds.
This illustrates an illusion of control: feeling that your involvement gives you power over a chance outcome. In investing, a busier screen can give you that feeling without giving you control over the next price move.
An extra trade needs a purpose. New information about a business gives you something to evaluate; rereading the same quote adds no evidence. Rebalancing to restore your planned mix or selling to meet a cash need can also make sense. Frequency alone cannot tell you why someone traded.
When memory rewrites the forecast
“I knew the rebound was coming” is easy to say after it arrives. Hindsight bias makes an outcome seem more predictable once you know it. In Baruch Fischhoff's 1975 study, people told how events ended judged those outcomes more likely beforehand, largely unaware that the ending had changed their judgment.
A dated reason from before the rebound gives you something firmer to check. If it says “I'm unsure whether this will recover,” the later claim of certainty has a problem. Memory can turn a hope into a forecast. An investing journal preserves the original version, including decisions to wait or make a planned fund purchase.
What the full account actually supports
The account needs a comparison, too. Suppose a suitable benchmark, chosen before the year began, earned a 4% total return over the same year. That turns $10,000 into $10,400 before your own costs and taxes. Your $9,770 account is $630 behind.
One losing year cannot settle whether you have skill. But “two winners prove I should trade more” already asks more of this record than it can deliver. The supported claim is smaller:
“The account lost 2.3% after trading costs and lagged its benchmark.”
Any claim of investing skill needs three things behind it:
- All positions and cash. The record includes the losers and investments you still own.
- All relevant costs, counted once. A spread's effect is already in the prices you actually paid and received; don't subtract it again from that trading result.
- A benchmark chosen in advance. Compare the same period; don't shop for an easier opponent after seeing the result.
This record gives you no basis to increase trading just because you remember two wins. The next lesson, confirmation bias, tackles the urge to seek agreement when an idea needs testing.
In short
- Confidence needs a complete record, not a highlight reel.
- You can win most of your trades and still lose money.
- In the brokerage study, extra activity failed to earn enough to cover its costs.
- A dated reason is stronger evidence of what you believed than a memory polished by the outcome.
- Compare the whole account, after costs, with a fair benchmark over the same period.