First, ten figures side by side
Only figures whose source page or paper we opened and checked are included. Anything we could not check is left out (see “Figures we did not use” at the end). Rows are in alphabetical order by country or region. The order carries no meaning, and it does not rank any country or region as better or worse. Each figure belongs to the year it was published and is not a current rate.
Table 1. Figures that count broker data (regulator documents)
Swipe the table sideways to see all columns →
| Source (country or region) | Product | What was counted | Period | How a “loss” was defined | Figure |
|---|---|---|---|---|---|
| ASIC Report 626 and Consultation Paper 322 (Australia, 2019)[4][5] | CFDs (contracts for difference), split into “margin FX” and other CFDs | Clients (the documents do not say whether people or accounts were counted). Data reported by 57 providers | Any 12-month period each provider chose between Jan 1, 2016 and Jun 30, 2017 | “Lost money.” These two documents do not give a detailed definition | 63% for margin FX, 72% for other CFDs (about 80% for binary options) |
| ESMA announcement (EU, March 27, 2018)[1] | CFDs (including currency pairs) | Retail accounts, from analyses by national authorities | Not stated | Not stated | Typically 74–89% of retail accounts lose money (range across countries). Average loss per client: €1,600 to €29,000 |
| Four EU member states (Ireland, Cyprus, Spain, Poland), as tabulated by ASIC[5] | CFDs | The table heading is “% loss.” It does not say accounts or people | Ireland 2013-14 and 2015-16; Cyprus Jan 1 to Aug 31, 2017; Spain 21 months (early 2015 to late 2016); Poland 2016 and 2017 | Not stated | Ireland 75% and 74%; Cyprus 76%; Spain 82%; Poland 79% and 80% |
| Study by Kentaro Iwatsubo in the Financial Futures Association of Japan (FFAJ) bulletin No. 114 (Japan, 2017)[10] | Over-the-counter (OTC) forex margin trading at FFAJ member firms | Accounts. A random sample of 1,000 accounts per month | The year 2015, calculated month by month | Monthly return on realized profit and loss, including swap. The top 5% and bottom 5% were removed. Only accounts with no open position at the start of the month and at least one trade in the month | The average was negative (about -2.9%). The median was close to zero and slightly positive. Pooling all months, more than half of account-months were positive (median 0.008%) |
| FFAJ quarterly data, as shown in Iwatsubo’s slides for a Financial Services Agency panel (Japan, 2018)[11] | OTC forex | Accounts. The share whose funds rose and the share whose funds fell | Each fiscal-year quarter (Japan’s fiscal year starts in April). The slide lists 11 quarters, FY2015 Q1 to FY2017 Q3 | After adjusting for deposits and withdrawals, whether realized profit and loss, change in unrealized profit and loss, and swap added up to a rise or a fall in the account. Accounts with no real change were left out | Accounts that fell: 47.48% to 75.74%. Accounts that rose: 24.26% to 52.52%. The share varied a lot from quarter to quarter |
| FCA consultation paper CP18/38 (UK, 2018)[3] | CFDs | Active retail client accounts. A weighted average of the risk warnings published by 7 firms (about 65% of the UK retail CFD market) | Not stated | Each firm’s own published figure was used. The definition is not given in this document | About 78% of accounts lost money |
| CFTC consumer advisory (US)[6] | OTC forex at registered dealers | “Non-discretionary” accounts disclosed by dealers | Q2 2021 to Q1 2022 | Calculated each quarter by the dealer under the rule (see the table below, “What counted as a loss?”). How the “past year” was put together is not stated | About one-third of customers made a profit and about two-thirds lost money |
Table 2. Figures that count people (studies and a survey)
Swipe the table sideways to see all columns →
| Source (country or region) | Product | What was counted | Period | How a “loss” was defined | Figure |
|---|---|---|---|---|---|
| Chague, De-Losso and Giovannetti, “Day Trading for a Living?” (Brazil, working paper, 2020)[13] | Equity futures | Individuals who began day trading in 2013 to 2015 and kept going for at least 300 days | 300 days or more | “Lost money,” per the abstract. We did not check the full text for details | 97% lost money. Only 0.4% earned more than a bank teller (over US$54 a day) |
| FFAJ survey (Japan, April 2018), as described by Iwatsubo[12] | Forex | 1,000 people in their 20s to 70s who currently trade forex (self-reported). The mix of sex and age was set in advance | The year 2017 | The profit or loss for the year, as the person answered | 60.3% made a profit |
| Barber, Lee, Liu and Odean, “The Cross-Section of Speculator Skill” (Taiwan, working paper, May 2011)[14] | Stocks on the Taiwan exchange | Individuals who day traded (about 360,000 in an average year). The 13% figure is limited to people whose yearly day-trading volume was at least 600,000 Taiwan dollars | 1992 to 2006, calculated year by year | Profit or loss from day trading in that year, after commissions and transaction taxes | In a given year, about 13% had a positive profit after costs, and about 15% had a positive excess return (beyond market moves) after costs; the two are defined differently. Only the group of about 1,000 people (under 1%) ranked top by the previous year’s results kept beating the market after costs in the following year |
It is no surprise that the share of “losers” differs from number to number. One counts accounts, one counts only people who lasted 300 days, one counts survey respondents. These are different groups. Here are four things to check when you read a figure.
How to read these numbers
1. Who was counted: accounts or people, and only those who stayed?
- An account is not a person. The EU and UK figures are shares of accounts. One person can hold several accounts, or open one and never use it. The Australian figure is a share of clients.
- Some figures count only people who stayed. The Brazilian 97% is a share of people who started day trading and lasted at least 300 days. People who quit are not in it. The Japanese survey asked people who currently trade forex, so people who quit are not in it either. Looking only at people who stayed can give a picture that differs from the whole (survivorship bias).
- A snapshot of accounts includes people who are about to quit. In another document, Iwatsubo suggests the low average return of scalpers may also be because many less-skilled investors scalp and leave the market without much profit.[11]
- A self-reported survey is a different kind of number from broker data. The 60.3% in the Japanese survey is what people said about their own profit or loss. It is not comparable to the Table 1 figure from account data in the same country (accounts that fell: 47% to 76% depending on the quarter). The group and the method are different.
- ASIC’s report says that 83% of the clients of Australian-licensed providers are outside Australia (2019 review).[4] A figure from an Australian regulator is not necessarily a figure about people who live in Australia.
2. What counted as a loss?
Even when authorities require firms to publish a figure, the counting rules differ.
Swipe the table sideways to see all columns →
| Source of the rule | Unit and period | What counts as a loss |
|---|---|---|
| EU (annex to ESMA’s 2018 decision)[2] | Each provider’s retail CFD trading accounts. Calculated every 3 months, looking back over the previous 12 months | An account where realized plus unrealized net profit and loss over the period is negative. Costs such as fees and commissions are included. Accounts with no open CFD in the period, profits and losses on products other than CFDs, and deposits and withdrawals are excluded |
| US (CFTC rules 5.5(e) and 5.18(i); NFA Interpretive Notice 9053)[7][8][9] | Each dealer’s “non-discretionary” retail forex accounts. Each quarter | Realized and unrealized gains and losses, minus fees and commissions, plus interest and similar income. Deposits and withdrawals are excluded. A result of zero or below counts as “not profitable” |
| Australia (a proposal in ASIC’s 2019 Consultation Paper 322)[5] | The provider’s retail CFD trading accounts. 12 months | The proposal was to disclose the share of accounts that made a loss over 12 months. We did not check the wording of the current order |
These differences change the figure. A 3-month window and a 12-month window give different shares. In the Japanese quarterly data above, the share of accounts that fell moved between the high 40s and the mid 70s of a percent depending on the quarter. Whether you count unrealized results, fees and swap, and whether you put a result of exactly zero on the loss side or the profit side, also changes the share. The 2018 EU decision asked each provider to publish its own figure. The standard warning (“between 74–89% of retail investor accounts lose money”) was for providers with no figure of their own to publish, and it reflects a range from national analyses.[1][2]
3. An average and a distribution are different things
- In the 2015 Japanese account data, the average monthly return was negative, but the median was slightly positive.[10] That can happen when the distribution has a long tail on the negative side (in that study, the monthly distributions were skewed to the left). “More than half of people are positive” and “the money overall went down” can both be true.
- The share of people who lost does not tell you how much they lost. The ESMA page gives an average loss per client of €1,600 to €29,000.[1] A share and an amount are different numbers.
- In the Taiwan study, about 13% had a positive profit after costs in a given year (limited to people day trading at least 600,000 Taiwan dollars a year), and about 15% had a positive excess return (beyond market moves). The two are defined differently. But only the group of about 1,000 people (under 1%) ranked top by the previous year’s results kept beating the market after costs in the following year.[14] “People who won this year” and “people who keep winning” are different groups.
4. Period, product and country all change the number
- Period: In the Japanese account data, the share of accounts that fell moved from the high 40s to the mid 70s of a percent by quarter (accounts that rose: 24.26% to 52.52%).[11] Even in one country and one product, the number changes with market conditions.
- Product: CFDs, OTC forex, stocks and equity futures are different products. Even within one Australian report the figures differ: 63% for margin FX and 72% for other CFDs.[5] A result from stocks cannot be applied to forex as is.
- Country and rules: Leverage limits, advertising rules and disclosure rules differ by country and region (see leverage limits by country). The tables here cover only the countries and regions we could check. We could not check figures for many countries in Asia, Africa and Latin America.
- Why the number was collected: Figures that regulators gathered to justify a rule describe the situation before the rule. These tables do not show how the numbers changed afterward.
What these numbers cannot tell you
- They do not mean “you will lose.” A figure is a share of a group. It is not a probability that decides any one person’s result.
- They do not mean “you will be fine,” either. Even a figure that shows fewer losers (in the Japanese survey, 60.3% made a profit) may reflect only people who are still trading.
- They do not mean “so quit” or “so keep going.” That decision is yours. This article does not recommend trading, or any broker.
- This article does not compare brokers. A figure a broker publishes depends on that firm’s mix of clients and on the period it chose.
Count your own numbers from your own records
Public figures cannot stand in for your own. You can count your own from your records. If you borrow the authorities’ counting method, you are measuring with the same ruler.
- Pick a period: one month, three months or twelve months.
- Leave out deposits and withdrawals: profit or loss for the period = ending balance − starting balance − deposits + withdrawals.
- Include costs: if you work from the balance, spread, commission and swap are already in it. Decide in advance whether to include unrealized results.
- Decide how to treat zero: does a result of exactly zero count as “not a loss” or as “not a profit”?
Swipe the table sideways to see all columns →
| Period | Starting balance | Deposit | Ending balance | Profit or loss for the period |
|---|---|---|---|---|
| 3 months | 100,000 | 50,000 | 130,000 | -20,000 |
(Made-up numbers as an example, in your account currency.) The balance looks like it rose from 100,000 to 130,000. But 50,000 of that was a deposit, so the result for the period is 130,000 − 100,000 − 50,000 = -20,000. If you do not leave out the deposit, you can read a loss as a profit.
With only a few months counted, a pattern may be chance. For how many trades it takes to trust a win rate, see how many trades before you can trust a win rate. For a study on trading often and results, see overconfidence and trading too much. For looking back on a month, see monthly trading review, and for what to record, what to write in a trading journal. How forex compares with other investments is in forex vs stocks, crypto and binary options. To get average profit and the largest drop from a list of results, the trading stats calculator can help, and the trading journal template gives you a table to start from.
In Tiltnote you can record profit or loss, whether you kept your rules, and how you felt for each trade. The records give you the material to count your own numbers.
Figures we did not use
- ASIC’s documents also quote figures from the UK FCA, Cyprus CySEC, Poland’s KNF and Italy’s CONSOB. The ones given for binary options are left out because this is a forex article.
- US figures that individual dealers publish are left out, because using them would turn this into a comparison of firms. The CFTC “about one-third” figure is the only US figure here.
- We did not open a survey by India’s SEBI, or Barber and Odean (2000) on US stocks (the latter is covered in overconfidence and trading too much).
- For Barber and colleagues, we opened the working paper (May 2011), not the journal version (2014). We did not check the journal version’s figures.
About this page
This article is not investment advice, and it does not recommend any trade. The figures belong to the year, country and product given. They are not current rates and not a prediction of your own results. If money or trading is causing you distress, or you find it hard to stop, please don’t carry it alone: talk to someone close to you or a support service where you live.
Sources
- European Securities and Markets Authority. (March 27, 2018). ESMA agrees to prohibit binary options and restrict CFDs to protect retail investors. https://www.esma.europa.eu/node/84933
- European Securities and Markets Authority. (2018). Notice of ESMA’s Product Intervention Decisions in relation to contracts for differences and binary options (ESMA35-43-1135, decisions of May 22, 2018; Annex II, Risk warnings). https://www.esma.europa.eu/sites/default/files/library/esma35-43-1135_notice_of_pi_decisions_on_cfds_and_binary_options.pdf
- Financial Conduct Authority. (2018). CP18/38 Restricting contract for difference products sold to retail clients and a discussion of other retail derivative products (paragraph 2.9 and footnotes 6 to 8). https://www.fca.org.uk/publication/consultation/cp18-38.pdf
- Australian Securities and Investments Commission. (August 2019). Report 626 Consumer harm from OTC binary options and CFDs. https://download.asic.gov.au/media/5241548/rep626-published-22-august-2019.pdf
- Australian Securities and Investments Commission. (August 2019). Consultation Paper 322 Product intervention: OTC binary options and CFDs (paragraphs 43, 89 to 93 and Table 1). https://download.asic.gov.au/media/5241542/cp322-published-22-august-2019.pdf
- Commodity Futures Trading Commission. Customer Advisory: Eight Things You Should Know Before Trading Forex (footnote 1). https://www.cftc.gov/LearnAndProtect/AdvisoriesAndArticles/CustomerAdvisory_MustKnowForex.html
- National Futures Association. Interpretive Notice 9053 (quarterly reporting). https://www.nfa.futures.org/rulebooksql/rules.aspx?RuleID=9053&Section=9
- 17 CFR § 5.5 (Risk disclosure statement), Legal Information Institute version. https://www.law.cornell.edu/cfr/text/17/5.5
- 17 CFR § 5.18(i) (Required disclosures; calculation of profitable accounts), Legal Information Institute version. https://www.law.cornell.edu/cfr/text/17/5.18
- Iwatsubo, K. (2017). FX証拠金取引の投資戦略とパフォーマンス ―保有期間・リスクと収益率の関係― [Investment strategy and performance in FX margin trading: holding period, risk and returns] (in Japanese, with an English summary). Financial Futures Association of Japan bulletin No. 114, October 2017. https://www.ffaj.or.jp/wp-content/uploads/2019/06/ffaj-kaihou114-i.pdf
- Iwatsubo, K. (March 29, 2018). Slides for the Financial Services Agency expert panel on settlement risk of OTC forex firms, document 4 (in Japanese). https://www.fsa.go.jp/news/29/singi/20180329-1/04.pdf
- Iwatsubo, K. (September 27, 2018). FX取引に関する顧客アンケート結果を踏まえて [Based on the results of a customer survey on forex trading] (in Japanese; describes the FFAJ survey of April 2018). https://www.ffaj.or.jp/wp-content/uploads/2018/09/20180927_hyouron.pdf
- Chague, F., De-Losso, R., & Giovannetti, B. (February 2020). Day trading for a living? FGV EESP Textos para discussão 525 (working paper; abstract checked). https://econpapers.repec.org/RePEc:fgv:eesptd:525
- Barber, B. M., Lee, Y.-T., Liu, Y.-J., & Odean, T. (May 2011). The Cross-Section of Speculator Skill: Evidence from Day Trading (working paper; the journal version is Journal of Financial Markets, 2014, whose text we did not check). https://faculty.haas.berkeley.edu/odean/papers/day%20traders/Day%20Trading%20Skill%20110523.pdf