Only for one group, in one country. The 97% comes from a study of Brazilian futures traders who kept day trading for more than 300 days.
Day trading is buying and selling the same asset within one trading session.
Chagué, De-Losso and Giovannetti (2020) found 97% of those 1,551 traders lost money after fees. They were 7.9% of the 19,646 people who started day trading between 2013 and 2015.
The 97% describes the people who stayed, not everyone who tried.
Other studies measure broader groups and land between 71% and 93%. None finds a majority making money.
Where does the 97% figure come from?
The 97% comes from "Day Trading for a Living?", an SSRN working paper by Fernando Chagué, Rodrigo De-Losso and Bruno Giovannetti. They used records from Brazil's securities regulator, the CVM, on everyone who started day trading mini-Ibovespa index futures from 2013 to 2015.
Of the 1,551 who day traded on more than 300 days, 97% lost money net of fees.
The 300-day cut fits the paper's question, whether a person can day trade for a living. Someone doing that trades on most days.
What the 97% does and does not mean
Sticking with it did not rescue most people. Only 17 of the 1,551, or 1.1%, earned more than Brazil's minimum wage of US$16 a day.
Only 8 beat a bank teller.
It does not mean 97% of all day traders everywhere lose money. The sample is one futures contract in one country over 2013 to 2017.
It also does not describe the 92% who quit before 300 days. Quoting it as "97% of day traders" moves the number onto a group the paper did not measure.
Every major study at a glance
| Study | Market and period | Who was measured | Finding |
|---|---|---|---|
| Chagué, De-Losso & Giovannetti (2020) | Brazil, mini-Ibovespa futures, 2013-2017 | 1,551 who day traded 300+ days | 97% lost money net of fees |
| Barber, Lee, Liu & Odean (2004) | Taiwan Stock Exchange, 1995-1999 | 130,000+ day traders per six months | More than 80% lost money in a typical six months |
| Barber, Lee, Liu & Odean (2014) | Taiwan, 1992-2006 | All day traders | Fewer than 1% earned predictable abnormal returns after fees |
| Barber, Lee, Liu, Odean & Zhang (2020) | Taiwan, 1992-2006 | 3.7 billion transactions | Lost 23.9 basis points a day after fees |
| SEBI (2024) | India, equity F&O, FY22-FY24 | Over 1 crore individual traders | 93% lost money, about ₹2 lakh each |
| SEBI (2024), intraday study | India, equity cash segment, FY23 | 69 lakh intraday traders | 71% lost money |
| ESMA (2018) | EU, CFD brokers | Retail CFD accounts | 74-89% lose money |
| SEC (2005) | United States | None | No percentage given |
The Taiwan studies are the closest to the classic picture of a day trader: individuals buying and selling the same stock on the same day, on an exchange that recorded every trade.
Which number should you quote?
Match the number to the claim.
For "most day traders lose money," Barber, Lee, Liu and Odean (2004) is the cleanest source: more than 80% of Taiwan's day traders lost money in a typical six-month period. SEBI's intraday study gives 71% for India's cash market in FY23.
For "almost nobody makes a living at it," use Chagué et al. with its condition attached: 97% of those who lasted 300 days lost money, and 1.1% beat the minimum wage.
For "skill is rare," Barber, Lee, Liu and Odean (2014) found fewer than 1% of Taiwan's day traders could predictably earn positive abnormal returns after fees.
ESMA's 74-89% counts CFD accounts, not day traders. The SEC publishes no loss rate at all, so a percentage credited to the SEC is misattributed.
Why do so many day traders lose?
Costs come first. SEBI found Indian individual F&O traders spent ₹26,000 each on transaction costs in FY24 alone.
The other side of the trade is often a machine. In the same data, 96% of proprietary traders' profits came from algorithmic trading.
Losing does not make most people stop. Nearly three-quarters of Taiwan's day trading volume came from traders with a history of losses, Barber et al. (2020) found.
Of the 1,551 Brazilians who day traded for more than 300 days, 8 earned more than a bank teller's starting salary of US$54 a day. The other 1,543 did not.
