Skip to main content
Skip to content
Crypto Trading

Trading Profit Statistics 2027: The Missing People Behind the Success Rate

Who is missing from a trader success rate? Compare original studies and use DN’s cohort calculator to separate observed profit, costs and missing outcomes.

A selected gold ticket above other tickets illustrating trading success statistics
Reading options: audio, text size & sections

Decentralised News · 2027 planning series · Sources reviewed 22 November 2026

What Matters

A trading success rate is meaningful only with its population, period, costs and definition of success. A positive trade, profitable account, repeatable skill and living income are different outcomes. DN’s evidence comparison and cohort calculator show how missing participants and changing denominators can transform a headline without changing a single recorded result.

One reader sees a study reporting that fewer than 1% of day traders demonstrated reliably positive abnormal returns after fees. Another sees a broker’s percentage of accounts that lost money. A third sees screenshots of profitable users in a trading group. Each has encountered a different measurement. Putting the numbers on one chart would not make them comparable.

The temptation is understandable. Someone considering a trading course, signal service or AI bot wants a usable answer: what are my chances? A statistic from a defined historical sample can inform that decision. It cannot automatically become a personal probability, a current global rate or evidence that a particular product changes the odds.

The more useful starting point is to reconstruct the sentence hidden behind the percentage. Who was counted? Who disappeared? What counted as profit? When did measurement stop? Those questions turn a slogan into a claim that can be checked.

DN Evidence Block

Edition and review
2027 research edition; sources checked 22 November 2026.
Evidence reviewed
Original Taiwan day-trader research, the authors’ Brazilian study abstract, SEBI’s 2024 announcement and 2026 report, a BIS crypto-loss model, FCA disclosure rules and a CFTC trading-system advisory.
Original DN work
A study-comparison framework, missing-outcome bounds and a deterministic cohort calculator.
Observation boundary
No DN trader accounts were collected. The calculator’s example is fictional.
Decisive limit
These sources do not establish one universal 2027 profitability rate for crypto traders.
Review status
Prepared by Decentralised News Research; no independent external statistical reviewer assigned.

Five meanings of “successful trader”

Begin by naming the outcome. If a marketing statement changes the outcome halfway through the argument, its impressive percentage can survive while its practical meaning disappears.

OutcomeWhat it establishesWhat it leaves unanswered
A profitable tradeOne completed trade met the stated P&L definition.Position size, other losses, open exposure and account result.
A profitable periodThe defined account or person finished a period above the chosen threshold.Risk, capital required, consistency and omitted costs.
Benchmark outperformancePerformance exceeded a specified comparator under a stated method.Whether the result is distinguishable from chance or repeatable.
Persistent skillA declared test finds predictive performance across separate periods.Future persistence, changing conditions and the ability to identify that trader beforehand.
Living incomeNet income meets a stated local expense or wage threshold.Income stability, tax, benefits, capital preservation and time commitment.

A trader can satisfy the second row while failing the fifth. Someone with substantial starting capital can produce enough cash to live on while underperforming a simpler benchmark. A high trade win rate can coexist with an unprofitable account when the losing trades are larger. These are arithmetic possibilities, not findings about a named platform.

What the well-known research actually says

Barber, Lee, Liu and Odean studied Taiwan day traders over 1992–2006 in The Cross-Section of Speculator Skill. Their conclusion that fewer than 1% could predictably and reliably earn positive abnormal returns net of fees concerns persistent performance under the paper’s method. It is not a count of everyone who ever had a profitable day.

The researchers ranked traders using an earlier year and examined subsequent performance. That ordering matters: identifying winners afterward is easier than identifying skill before the next period. The study’s geography, market, historical period and abnormal-return definition remain part of the finding. They cannot be removed when the statistic is reused to discuss crypto.

Chague, De-Losso and Giovannetti’s Day Trading for a Living?, in the authors’ June 2020 abstract, reports a 97% loss rate among individuals who began day trading Brazilian equity futures in 2013–2015 and persisted for more than 300 days. The persistence condition is essential. It identifies a selected group of participants, not all prospective traders worldwide. DN uses the verified abstract’s cohort and headline result here; it does not reproduce inaccessible underlying account records or claim an independent replication.

SEBI’s September 2024 announcement reported that 93% of individual equity futures-and-options traders incurred losses over FY22–FY24. Its August 2026 report separately reports 87.7% loss-makers in FY26 versus 90.9% in FY25. The later client analysis uses 15 brokers and realised results adjusted for transaction costs; it is distinct from the report’s market-wide gross analysis. Annual and combined-period outcomes must retain their own denominators.

EvidenceUnit and horizonSafe interpretation
Taiwan studyDay traders, 1992–2006; prior and subsequent performance.A historical test of persistent net abnormal performance.
Brazil study abstractNew equity-futures day traders from 2013–2015 who continued beyond the stated threshold.A conditional result for persistent participants.
SEBI reportsIndividual equity-derivatives participants in specified financial years.Dated outcomes for the defined product and participant samples.
Provider risk disclosureEligible accounts at that provider over the required calculation window.Provider-account experience, not a worldwide personal forecast.

The shared lesson is that costs and losses require serious attention. The studies cannot be pooled into a single percentage merely because their conclusions feel similar. That would combine distinct populations, products, horizons and outcome tests.

Why a broker’s account percentage is another kind of evidence

The FCA’s COBS 22.5 disclosure rules illustrate a specific measurement design. Relevant firms recalculate their loss percentage every three months over the preceding 12 months. The calculation includes realised and unrealised net results and charges for the covered products, while excluding accounts without a relevant open position, unrelated products, deposits and withdrawals.

That definition is useful precisely because it is bounded. An account is not necessarily a unique person across providers. An account-level loss does not disclose the person’s complete portfolio, and an account without a measured loss is not necessarily a sustainable livelihood. Preserve the provider, calculation date and product scope when citing a warning. The example describes the cited UK rules, not a universal global disclosure requirement.

Comparing two providers’ percentages also requires caution. Their customers may differ in experience, products, leverage, trading frequency and market exposure. A lower loss percentage does not by itself demonstrate that the provider caused better outcomes. It is a descriptive observation that needs a compatible comparison design.

A modelled crypto loss is not an audited trade ledger

The BIS paper Crypto Shocks and Retail Losses uses crypto-app adoption data and explicit purchase assumptions to investigate retail outcomes through 2022. Its loss analysis considers purchases at app-download timing and a simulation of repeated monthly bitcoin purchases. Those assumptions let the authors estimate outcomes without observing every user’s complete trading account.

This is a different evidence class from audited realised P&L. An app download need not equal an immediate purchase; a modelled holder need not match someone’s actual deposits, withdrawals, sales or other assets. The study is useful when described as the model it is. Calling it proof of every user’s actual loss would make a stronger claim than the method supports.

The same distinction applies to an AI-generated dashboard. A simulation, a reconstructed estimate and an authenticated account statement can all contain numbers labelled “profit.” Ask how the number was obtained before deciding what it means.

The people who disappear from the denominator

Consider a fictional training cohort with 1,000 entrants. At the end of the declared measurement period, 400 provide complete records and 80 have positive net outcomes. A promoter could accurately calculate that 20% of observed participants were profitable. It would be inaccurate to describe that as a 20% success rate among all entrants without addressing the missing 600.

Only 8% of the starting cohort have confirmed positive outcomes. That is a confirmed fraction, not proof that the remaining 92% lost. Some missing participants could have stopped trading, withheld records, moved accounts or made money. Without their outcomes, the reason remains unknown.

Under a deliberately simple binary classification, the starting-cohort positive-outcome proportion could range from 8% to 68%: the lower endpoint assumes none of the missing participants were positive; the upper endpoint assumes all were. Those endpoints are logical bounds, not a confidence interval and not equally likely scenarios.

This is why asking for the original cohort matters. Publishing only continuing users, responders, challenge passers or accounts that remain open can answer a legitimate conditional question. The label must say which condition was applied. It cannot silently answer the question a new entrant is asking.

DN Trading Outcome Denominator Checker

Enter counts from one defined cohort, period and outcome standard. Use people, accounts or programme entries consistently. “Positive” means above zero after the costs included in the declared result. The example below is fictional and does not describe a broker, course or research study.

DN Trading Outcome Denominator Checker

Compare observed success with the original cohort and show logical bounds when outcomes are missing. Use one fixed cohort, one period and one net-outcome definition.

This instrument runs locally in your browser. It does not connect to accounts, fetch evidence or submit instructions. Use aggregate values and non-sensitive references. Downloaded records contain your inputs.

Enter your inputs or load the labelled example, then select Calculate.
Machine-readable illustrative record

An illustrative record of the example assumptions, not an observed transaction or provider test. A null calculation timestamp identifies this static example.

{
  "schema_version": "1.0.0",
  "article_id": "N005",
  "evidence_basis": "illustrative",
  "source_review_date": "2026-11-22",
  "computed_at": null,
  "inputs": {
    "starting": "1000",
    "observed": "400",
    "positive": "80",
    "income": "20"
  },
  "status": "Valid cohort analysis: missing outcomes retained as unknown",
  "outputs": {
    "tool_id": "N005",
    "version": "1.0.0",
    "starting": 1000,
    "observed": 400,
    "positive": 80,
    "income": 20,
    "missing": 600,
    "observed_positive_pct": 20,
    "confirmed_cohort_pct": 8,
    "missing_pct": 60,
    "logical_lower_pct": 8,
    "logical_upper_pct": 68,
    "income_observed_pct": 5,
    "income_cohort_pct": 2
  }
}

For the 1,000-person example, 400 observed records and 80 positive results produce an observed positive rate of 20%, a confirmed starting-cohort fraction of 8%, and 60% missing coverage. If 20 of the observed participants also exceed a predefined income threshold, that is 5% of observed participants and 2% of entrants. The threshold itself must be stated; the tool does not invent a living wage.

Change only observed records to 800 while keeping positive results at 80. The observed rate falls to 10%, while the confirmed starting-cohort fraction remains 8%. Nothing happened to those 80 outcomes. The measurement gained information about the denominator. This is the mechanism the checker makes visible.

Costs can change who crosses zero

Population analysis depends on sound individual accounting. A fictional participant with $1,200 of trading gains before separately charged costs, $900 of transaction expenses and $400 of external tools finishes at −$100 under that defined economic boundary. Omitting the external tools would classify the same participant as positive. Do not subtract charges twice when the reported P&L already includes them.

The CFTC’s trading-system advisory highlights costs such as systems, data subscriptions, commissions and fees, alongside the limitations of hypothetical performance. DN’s account reconciliation guide provides the detailed individual worksheet. This article addresses what happens when those classified outcomes become a population percentage.

Time is another boundary. A positive net result can be inadequate compensation for hundreds of working hours. Capital also has an alternative use. Keep these comparisons separate from exchange-account P&L: a labour-income test and a benchmark-return test can explain economic usefulness without pretending that unpaid time was a cash charge deducted by the venue.

Persistence needs a second period and a rule fixed beforehand

To examine persistence, define the first period, select a rule using only evidence available at its end, and measure the selected group in a later period. Do not move the dates to preserve a winner. Include participants who stop, explain how their outcomes are treated and report whether the same identifier follows them across accounts.

Repeated positive results still do not automatically establish skill. The sample may be small, exposures may share a common market direction, or many strategies may have been tried before the successful one was selected. Statistical testing requires a design that addresses those dependencies. The DN calculator deliberately supplies no probability of skill.

For AI products, add an attribution question: what changed because of the AI? A profitable user who also trades manually, receives rebates or holds appreciating collateral is not a controlled demonstration of the model’s effect. Compare the relevant strategy and costs under a declared protocol before crediting the product.

The evidence packet to request before paying

Ask for one complete measurement sentence: “Among [defined eligible population] observed over [dates], [count] met [outcome definition], after [included costs], with [missing or excluded cases].” Then request the source and the method used to authenticate records. A privacy-respecting aggregate report can still explain its coverage and classification rules.

Look for a separate account of selection. Were results solicited voluntarily? Were unsuccessful users removed? Do multiple attempts belong to one person? Was an account reset treated as a fresh success opportunity? Are refunded customers still in the starting cohort? These questions matter to course and bot evaluations even when no deception is alleged.

Finally, ask what the evidence does not establish. A provider willing to show a bounded outcome may be more informative than one presenting a universal claim. The absence of adequate data is a reason to withhold the claimed conclusion; it is not proof that every user lost or that the provider committed fraud.

DN Alpha Thesis

The useful trading statistic preserves the question it was designed to answer. Market, population, period, costs and attrition are part of the result, not footnotes to remove for a stronger headline. A spectacular percentage with those boundaries missing is less useful than a modest finding that another reader can reconstruct.

For a 2027 decision, demand evidence relevant to the product and capital at issue. Use historical research to understand risks and measurement problems. Use authenticated, complete current records to evaluate a specific claim. Neither category becomes a promise about your next trade.

Methodology, limitations and corrections

DN compared the definitions in the cited original research and official documents as reviewed on 22 November 2026. This is an evidence-reading framework, not a meta-analysis or independent replication. Different studies use different records and methods. We have not reclassified their participants, harmonised their datasets or calculated a combined success rate.

The cohort checker assumes the observed records and missing records partition the same starting cohort. Its bounds concern an unknown binary outcome, not future returns. They become invalid if records are duplicated, periods differ, the positive count includes people outside the observed group, or eligibility is changed after measurement. The 2027 edition label does not imply that future account outcomes have been observed.

No platform receives a ranking or recommendation in this article. Source links are neutral research references. For broader due diligence, use DN’s AI trading-agent evidence audit. Send corrections with the relevant definition and primary record to the DN editorial desk.

Frequently asked questions

What percentage of crypto traders are profitable in 2027?

The sources reviewed do not establish one current universal percentage. They examine different markets, time periods, populations and outcome definitions. A useful claim must name those boundaries and explain costs, missing participants and record quality.

Does a high trade win rate mean the account is profitable?

No. Winning trades can be smaller than losing trades, and fees, funding, subscriptions and open losses can change the account outcome. Reconcile the complete account before interpreting a population success percentage built from it.

Why is an account-loss percentage different from a trader-loss percentage?

One person can have multiple accounts or use several providers. An account statistic measures the defined account boundary. Converting it into a person-level claim requires a reliable deduplication method and a decision about which other accounts or holdings belong in scope.

Are people missing from a study automatically unsuccessful?

No. Their outcomes are unknown unless records establish them. Report the missing count and the confirmed fraction. The DN calculator shows logical best- and worst-case bounds; it does not assume why participants are missing.

Does a positive annual result prove someone can trade for a living?

No. Living income needs a stated cash requirement, relevant taxes and expenses, adequate capital and a sustainable income pattern. A small positive result or one unusually good year does not establish those conditions.

Can the denominator checker estimate my chance of success?

No. It calculates observed proportions and missing-outcome bounds from counts you supply. It does not authenticate records, model your strategy or infer an individual probability from a historical cohort.

Can I average the published loss rates from different studies?

A simple average would usually be misleading when markets, units, periods or outcome definitions differ. A defensible evidence synthesis needs compatible definitions and a justified statistical design. This article deliberately does not pool the reviewed figures.

Primary sources and review record

  1. Barber, Lee, Liu and Odean: The Cross-Section of Speculator Skill. Taiwan 1992–2006 study; prior-year ranking and next-year performance; less than 1% reliable positive net abnormal returns. Reviewed 2026-11-22.
  2. Chague, De-Losso and Giovannetti: Day Trading for a Living? June 2020 author abstract. 97% of the Brazil 2013–2015 entrants persisting beyond 300 days lost money; used only the verified author abstract, no unsupported cost breakdown. Reviewed 2026-11-22.
  3. SEBI: September 2024 equity F&O study announcement. 93% loss incidence for the specified FY22–FY24 combined period. Reviewed 2026-11-22.
  4. SEBI: Profitability of Individual Traders in Equity Derivatives, FY25–FY26. FY26 87.7% vs FY25 90.9% loss-makers; 15-broker client dataset, realised P&L adjusted for costs; separate market-wide gross data. Reviewed 2026-11-22.
  5. FCA Handbook: COBS 22.5. Current provider-account risk-warning calculation: trailing 12 months, quarterly refresh, realised/unrealised results and charges, exclusions. Reviewed 2026-11-22.
  6. BIS Bulletin 69: Crypto Shocks and Retail Losses. App-download-based purchase assumptions and monthly bitcoin simulation; modelled loss classification distinct from account audit. Reviewed 2026-11-22.
  7. CFTC: Commodity Trading Systems Sold on the Internet. Hypothetical-performance limitations and costs such as tools, data and commissions. Reviewed 2026-11-22.

Author and original DN instrument: Heath Muchena. Edition: 2027 planning. Source review: 22 November 2026. Version: 1.0.0. First publication adds the sourced article, declared-input tool and record schema. Refresh when the cited specification, provider terms, case status or source methodology changes.

Corrections: press@decentralised.news. No independent audit or transaction test is claimed. No paid placement or affiliate recommendation is included in this article.

Get the most talked about stories directly in your inbox

Join the Decentralised News briefing for independent crypto, DeFi and AI analysis. No spam, unsubscribe anytime.