97% Lose Money. The Interface Is Why.

Charlie Munger called it a gambling parlor. Warren Buffett called it a casino. ESMA called it an embedded conflict of interest. They are all describing the same thing.

The architectural pattern shared by retail stock apps, crypto exchanges, and CFD/forex brokers.

Table of Contents

  • Foreword: Why This Document Exists
  • Part I — The Pattern
    • 1.1 The Architectural Property at the Root
    • 1.2 The Pattern Stated in One Line
    • 1.3 The Two Interfaces
    • 1.4 The Four Properties a Risk-First Interface Must Deliver: Clarity, Control, Loss Reduction, Aggregate-Risk Visibility
  • Part II — The Catalog: A Taxonomy of UX Primitives That Express the Pattern
    • 2.1 The Celebration-of-Volume Primitive
    • 2.2 The Attention-Engineering Primitive
    • 2.3 The Friction-Removal Primitive
    • 2.4 The Leverage-Pyramid Primitive
    • 2.5 The Loss-Aversion Exploitation Primitive
    • 2.6 The Opacity Primitive
    • 2.7 The Six Primitives, the One Pattern
  • Part III — The Investigative Record
  • Part IV — The Academic Record
  • Part V — Voices: Designers, Ethicists, and Investors on Record
  • Part VI — The Economics of Loss
  • Part VII — The Architectural Alternative: Alignment as Primitive
  • Sources / Bibliography

Foreword: Why This Document Exists

In June 2020, Alex Kearns, a 20-year-old college student from Naperville, Illinois, opened the Robinhood app and saw a balance of negative $730,000. The figure was wrong — a display artifact of an unsettled options spread. He had been approved for options trading by an algorithm that FINRA would later find “approved customers based on inconsistent or illogical information.” He emailed customer service three times. He received no live response. He took his life that night.

The interface that approved Kearns was the same interface that displayed the balance, which was the same interface that offered no escape channel. Frictionless to enter; friction-loaded to ask a question. The asymmetry was not a bug. It is a documented design property — and the documentary record on it is one of the largest in modern financial regulation.

The record includes a 23-page Massachusetts administrative complaint, a $70 million FINRA settlement, a 23-page Yale Law Journal Forum essay, a 25-page Management Science paper, a 56-page Federal Reserve Bank of New York staff report, three ESMA product-intervention measures, an FCA policy statement, dozens of peer-reviewed studies, and the bankruptcy filings of Celsius, Voyager, BlockFi, Genesis, and FTX.

The numbers inside that record are not subtle. A São Paulo working paper, now cited across the academic literature, found that 97% of persistent Brazilian day traders lose money. ESMA’s CFD task force documented retail account loss rates of 74% to 89% across EU jurisdictions. Barber, Huang, Odean, and Schwarz found average 20-day abnormal returns of −4.7% on the top stocks purchased each day on Robinhood. The Federal Reserve Bank of Chicago documented approximately $13 billion in customer outflows from five centralized crypto platforms in the run-up to bankruptcy.

Most of the people on that record — Munger, Buffett, Howard Marks, Tristan Harris, Harry Brignull, Brett Steenbarger, Brad Barber, Terrance Odean, Fernando Chague — are not crypto natives. They are establishment professionals describing what they observe.

What they observe, in one sentence, is this: a documented cognitive bias of retail traders — selling winners early, holding losers too long — has been converted, by interface choice, into a measurable revenue stream for the platforms. The bias is the trader’s. The conversion is the design’s.

What they observe is a pattern. Not a company. A pattern.

The pattern is the architectural property in which the platform’s revenue function is misaligned with the trader’s outcome function. The interface is the mechanism through which the misalignment is expressed. This dossier documents the pattern at the interface level across three platform categories — retail stock apps, crypto exchanges, CFD/forex brokers — and quotes the regulators, academics, and journalists who have already done the verdict-rendering.

The closing argument is architectural. The patterns documented here are not bugs of any one exchange or app; they are the designed outcome of a business model in which platform revenue grows when the audience loses. The architectural primitive that closes the gap is alignment — a fee structure in which the platform earns more when the trader survives longer. The properties a risk-first interface delivers under that alignment are four: clarity, control, loss reduction, and aggregate-risk visibility. Part I, Section 1.4 names them. Part II shows what their absence looks like. Part VII shows what their presence looks like.

Every claim in this document is sourced. The point of restraint is precision. The numbers do the work.

Part I — The Pattern

1.1 The Architectural Property at the Root

Kearns is not on the record alone. The Massachusetts Securities Division’s complaint cites “one Robinhood customer with no investment experience” who placed more than 12,700 trades in six months. The outcome studies named in the foreword — Brazil, the EU, Robinhood-era U.S., the 2022 crypto-platform runs — are the population-scale shadow of that single customer. These outcomes are not unrelated.

Most consumer-facing retail trading platforms — stock apps, crypto exchanges, CFD/forex brokers — share a structural property: the platform’s revenue function is decoupled from, and frequently inversely correlated with, the trader’s profit function. The mechanisms differ across categories. The structural conclusion is the same.

For zero-commission retail brokers, revenue is a function of trade frequency and routing economics. The U.S. Securities and Exchange Commission documented in its December 17, 2020 settlement order that Robinhood received “unusually high” payment-for-order-flow rates from its market-maker counterparties (SEC Release No. 33-10906). Therefore the platform earns more when the audience trades more, regardless of whether the audience profits.

For CFD and rolling-spot-forex brokers, the conflict is more direct. The European Securities and Markets Authority documented that the inherent features of these products include an “embedded conflict of interest between providers and their clients” — the broker is often the counterparty to the client trade, so the broker’s gross revenue is the client’s gross loss (ESMA press release, March 27, 2018).

For perpetuals exchanges, fees are charged per trade and per liquidation. Funding rates extract from leveraged longs in bull markets and leveraged shorts in bear markets. Liquidation engines convert trader equity into platform revenue at a rate determined by leverage selection.

The interfaces are the surface on which the structural property is expressed. They are designed to maximize the metric the platform monetizes. That metric, across all three categories, is volume. Not P&L. Not survival. Volume.

This is the rentability pattern. The pattern is the enemy.

1.2 The Pattern Stated in One Line

When the platform’s revenue grows in the same direction as the audience’s losses, the platform’s interface will be optimized to accelerate the audience’s losses.

This is the house edge. The phrase is not new. Charlie Munger used the more direct version — “gambling parlor” — at the 2022 Daily Journal annual meeting. Warren Buffett, at the 2021 Berkshire Hathaway annual meeting, used “casino group.” Both quotes are reported below in Part V. The structural diagnosis is therefore not novel; it is on the public record from the most senior practitioners in capital allocation.

1.3 The Two Interfaces

There are two ways to design a trading interface.

One celebrates the act of trading. It places one-tap order entry as a first-class action. It defaults to the maximum leverage the platform’s risk engine permits. It surfaces volatility — “top movers,” “biggest gainers,” “trending tokens” — to drive attention toward whatever is moving fastest right now. It pushes notifications. It awards animations. It celebrates volume.

The other celebrates the survival of the trader. It places the stop-loss as a first-class action. It treats every entry as its own object, with its own price, its own risk, and its own exit. It surfaces position size relative to account equity. It does not push notifications. It does not award animations. It celebrates survival.

The first interface optimizes for the platform’s revenue function. The second optimizes for the trader’s profit function. These are different optimizations because they are aligned with different parties.

1.4 The Four Properties a Risk-First Interface Must Deliver: Clarity, Control, Loss Reduction, Aggregate-Risk Visibility

The risk-first interface is not defined by what it removes. It is defined by what it delivers. Four properties are load-bearing. A platform aligned with trader survival delivers all four; a platform aligned with volume can deliver none.

Clarity. Every trading decision the user has made must be visible as a distinct object, with its own entry price, its own stop-loss, and its own profit-and-loss. The standard interface blends multiple fills on the same instrument into a single averaged position — and that averaging is not a neutral accounting choice. It hides the cost of every decision after the first one. A trader who entered well at $100 and then averaged down at $90 and again at $80 sees a single position with an average of $90; the bad decisions are absorbed into a friendlier-looking blended cost basis. The good entry and the bad entries become indistinguishable. A trader cannot review what they cannot see, and cannot learn from what they cannot review.

The clarity primitive treats every entry as a separate trade. Every fill is a discrete position object. The trader sees the original $100 entry at its real P&L, the $90 entry at its real P&L, and the $80 entry at its real P&L — three trades, three risk profiles, three exits. The interface refuses to merge them. Every trade tells its own story. The trader’s review of their own performance is finally honest.

Control. The averaging convention is not just a clarity failure; it is a control failure. By merging fills, the standard interface removes the friction at the exact moment a trader is most likely to act on cognitive bias — adding to a losing position. “Average down” is the disposition effect’s interface expression: it lets the trader avoid the realization of a loss by re-basing the cost. Shefrin and Statman (1985) named the bias; Odean (1998) measured it; the standard retail interface operationalizes it.

The control primitive inverts this. Each new entry is a new decision, not a modification of an old one. There is no “average down” — only a fresh trade with its own entry, its own sizing, its own stop. If a trader believes the original thesis still holds and wants to add at a worse price, they take a new position with a new stop. The platform does not silently merge the two; the platform does not hide which entry was a mistake. The cognitive load of the second decision is preserved. The trader is in control because the interface refuses to take control away.

Loss reduction. Clarity and control compound into the third property. When each entry carries its own stop-loss and its own P&L, a single bad trade cannot pull the others down with it. In the standard blended-position model, one stop-loss covers the entire averaged position; if the trader was right on the first entry and wrong on the second, both are now risked against a single exit price that no longer reflects either original thesis. A move that should have stopped out only the second entry instead stops out the entire combined position — including the entry that was, until the average corrupted it, winning.

The loss-reduction primitive prevents this contamination. Each entry has its own stop. A losing trade closes. The winning trades continue. The trader’s bad decision is contained to the bad decision. The retail-trading literature has documented for thirty years what happens when this containment is absent: the disposition effect drains 3-5% per year from individual investor returns (Odean 1998), the Brazilian day-trading study found 97% of persistent day traders lose money (Chague, De-Losso, Giovannetti 2020), and the ESMA CFD record found 74-89% of retail accounts unprofitable (ESMA 2018). The mechanism behind those numbers is not exotic. It is the same mechanism repeated thousands of times: a losing entry is hidden, then doubled, then merged with a winner, and the entire position is stopped out at once.

Aggregate-risk visibility. Clarity at the per-entry layer is necessary; it is not sufficient. A trader who can see twenty entries individually still cannot see, without computing it across N positions in their head, the dollar amount they lose if every open position moves against them and every stop and liquidation triggers in sequence. Kahneman and Lovallo, “Timid Choices and Bold Forecasts,” Management Science 39:1 (1993), named this the narrow framing problem: investors evaluate risks one at a time rather than aggregating them into a portfolio, and systematically misjudge the portfolio’s actual risk profile as a consequence. The bias has thirty years of experimental record behind it — Tversky and Kahneman (Science, 1981) on framing effects; Redelmeier and Tversky (Psychological Science, 1992) and Langer and Weber (Management Science, 2001) on aggregation effects in lab settings; Barberis and Huang (Journal of Economic Dynamics and Control, 2009) on the asset-pricing implications.

The honest accounting of the literature requires one acknowledgement. Beshears, Choi, Laibson, and Madrian, “Does Aggregated Returns Disclosure Increase Portfolio Risk Taking?” Review of Financial Studies 30:6 (2017), tested whether presenting aggregated returns disclosures in a more realistic mutual-fund-allocation setting reproduced the earlier lab findings of Gneezy and Potters (Quarterly Journal of Economics, 1997) and Thaler, Tversky, Kahneman, and Schwartz (Quarterly Journal of Economics, 1997). They did not. The behavioral effect of aggregation in lab settings does not appear to transfer cleanly to retirement-portfolio allocation decisions. The cognitive bias of narrow framing remains well-documented; whether aggregated information presentation guarantees changed downstream behavior is contested in the experimental literature. The deliverable named here therefore concerns visibility as something a risk-first interface owes the trader — the floor is shown; what the trader does with it is the trader’s decision. The position is structurally consistent with the dossier’s posture on the disposition effect: the bias is the trader’s; the conversion of the bias into platform revenue is the design’s.

The regulatory record acknowledges the aggregate-exposure problem and addressed it through broker obligation rather than trader disclosure. ESMA’s product intervention measure of March 23, 2018 imposed a margin close-out rule on a per account basis — brokers were required to auto-close retail CFD positions when account margin dropped below 50% of initial margin — and a negative balance protection on a per account basis — limiting retail aggregate liability to the total funds in the CFD trading account (ESMA-71-98-128). The FCA codified the same measures permanently in PS19/18 (July 2019). Regulators saw the aggregate-account-exposure problem clearly enough to require broker-side automated action. They did not require trader-side aggregate-risk disclosure. The architectural gap is exactly the gap between what the broker is forced to do for the trader and what the trader is shown about themselves.

The aggregate-risk-visibility primitive resolves the gap at the trader-disclosure layer. The interface surfaces, as a single visible number, the worst-case dollar outcome across every open position if every stop and liquidation triggers from the present moment forward. The number moves in real time as the trader opens positions, adjusts stops, and adds collateral. The trader does not have to compute it across N positions. The interface computes it.

The empirical absence is, as of this writing, universal in the category. A direct survey of the retail trading platforms relevant to this dossier confirms it. Centralized retail brokers: Robinhood does not surface it. Centralized perpetuals exchanges representing approximately 62% of global crypto derivatives volume in 2025 per CoinGlass aggregation — Binance, OKX, Bybit, Bitget — do not surface it. ESMA-regulated CFD brokers: Plus500, IG, eToro, and CMC Markets do not surface it. Hyperliquid’s own native interface (the leading non-custodial perpetuals protocol by 2025 open interest) does not surface it; the recently launched portfolio margin feature introduces a portfolio margin ratio (equity over risk-weighted notional value), which is a margin-health ratio, not the aggregate worst-case dollar amount. The major Hyperliquid third-party frontends — MetaMask Perps (which routes to Hyperliquid), Phantom, pvp.trade, and the third-party analytics layer at Hyperdash, HyperTracker, ASXN, and Dexly Explorer — do not surface it. The standard interface vocabulary across all surveyed platforms is per-position liquidation price, account-level equity, account-level margin ratio, and per-position unrealized P&L. None of them surfaces the aggregate worst-case dollar floor across all open positions. The primitive is, as a category property, structurally absent.

The architectural translation: isolated entries beyond isolated margin. The standard risk-first prescription names isolated margin — the convention by which each position has its own collateral. Clarity, control, and loss reduction require something stricter: each entry must have its own collateral, its own price, its own stop, its own P&L. Isolated margin protects one position from another. Isolated entries protect one decision from another. Isolated entries is the architectural object that delivers three of the four properties at once. The fourth — aggregate-risk visibility — is delivered by computing across all entries and surfacing the aggregate worst-case as a single number. Where this dossier from this point forward refers to a risk-first default at the position layer, it means isolated entries with aggregate-risk visibility — not the blended-position model dressed up with a per-position collateral wrapper.

The rest of this dossier documents what the volume-aligned interface looks like, who has documented it, and what the regulators, academics, and journalists have said about its consequences. Part VII returns to clarity, control, loss reduction, and aggregate-risk visibility as the affirmative architecture.

Part II — The Catalog: A Taxonomy of UX Primitives That Express the Pattern

This section organizes the documentary record around the design primitives — the recurring interface elements that express the rentability pattern across all three platform categories. Each primitive is documented with cross-platform evidence and closes with a frame naming the underlying architectural property.

2.1 The Celebration-of-Volume Primitive

2.1.1 Confetti Animations on Trade Execution

Robinhood’s mobile application historically rendered an animated burst of confetti each time a customer completed certain in-app actions, including the placement of a trade. The Massachusetts Securities Division documented the practice in its December 16, 2020 administrative complaint (Docket No. E-2020-0047), which alleged that Robinhood “used gamification strategies to manipulate customers into continuous interaction and constant engagement with its application.” The complaint specifically cited “celebratory imagery” tied to the frequency of trading.

In April 2021, Robinhood announced it would remove the confetti animation. InvestmentNews covered the removal under the headline “Robinhood Drops the Confetti, but Advisers Aren’t Convinced.” The Yale Law Journal Forum cited the same removal and observed that the confetti was a small slice of a larger engagement architecture.

In January 2024, Massachusetts settled with Robinhood for $7.5 million. The settlement, announced by Secretary of the Commonwealth William Galvin’s office, required Robinhood — for Massachusetts customer accounts — to “cease any future use of celebratory imagery tied to the frequency of trading, push notifications highlighting specific lists, and features that mimic games of chance” (InvestmentNews).

2.1.2 Achievement Badges and Scratch-Ticket Rewards

The Massachusetts complaint documented additional gamification primitives, including a “scratch ticket” reward system, randomized stock-share giveaways, and a “tapping” game which moved users up a referral waitlist. The Berkeley Technology Law Journal summary of the 2024 settlement specifies “confetti animations, lottery style stock rewards, curated lists of popular stocks, push notifications, and a ‘tapping’ game to climb a wait list” (Mambrini, “The Gamification of Investments,” BTLJ, November 2025).

2.1.3 Empirical Effect on Trading Volume

The peer-reviewed empirical record on these primitives is now load-bearing. Chapkovski, Khapko, and Zoican, “Trading Gamification and Investor Behavior,” published in Management Science (2026, vol. 72, issue 1, pp. 32–56), used a randomized online experiment to test the causal effect of hedonic gamification — confetti, achievement badges — on trading behavior. The headline finding: hedonic gamification increases trading volume by an average of 5.17%. Participants with lower financial literacy were measurably more likely to choose the gamified platform when given the option.

2.1.4 Cross-Platform Expression

The same primitive recurs in crypto. The Hxro–FTX product TixWix, launched October 28, 2020, was designed, in the words of Hxro CEO Dan Gunsberg, to provide retail traders “simple, fast, and intuitive ways to interact with crypto markets,” with positions tradable for as little as $1.00 (CoinDesk; CryptoNinjas). FTX CEO Sam Bankman-Fried, in the same press cycle, framed it as making trading “more fun and interactive.” A BeInCrypto analyst who tested the product wrote: “I was struck by the simplicity of the application — and how addictive it was, staring at the second by second changes in BTC price.”

The Norwegian study Rakovic and Inal, “Dark Finance: Exploring Deceptive Design in Investment Apps” (Springer, 2023, INTERACT proceedings), examined 26 mobile investment apps and found that “nearly all the studied apps incorporate dark patterns to varying degrees.”

2.2 The Attention-Engineering Primitive

2.2.1 “Top Movers” Lists and Volatility Surfacing

The Massachusetts Robinhood complaint documented Robinhood’s use of “Top Movers” and “100 Most Popular” lists alongside push notifications drawing user attention to high-volatility stocks. The Yale Law Journal Forum essay describes the same practice: “push notifications hyping short-term volatility in ‘biggest mover’ stocks” (Langvardt and Tierney, 131 Yale L.J. Forum 717, 2022).

The empirical consequences are documented in the Brad Barber–Terrance Odean line of research. Barber, Huang, Odean, and Schwarz, “Attention-Induced Trading and Returns: Evidence from Robinhood Users,” Journal of Finance, vol. 77, no. 6, pp. 3141–3190 (2022), found that “Robinhood investors engage in more attention-induced trading than other retail investors” and that “intense buying by Robinhood users forecasts negative returns. Average 20-day abnormal returns are -4.7% for the top stocks purchased each day.”

The earlier Barber–Odean paper, “All That Glitters” (Review of Financial Studies 21:2, 2008), established the foundational mechanism: individual investors are net buyers of attention-grabbing stocks, and attention-driven buying is followed by underperformance.

2.2.2 Push Notifications

The U.K. Financial Conduct Authority, in Occasional Paper 66 (“Playing the market: a behavioural data analysis of digital engagement practices and investment outcomes”), documented that push notifications increased trading volume by 11% and prize draws by 12%. Younger participants (18–34) increased their end-of-trading portfolio riskiness more than older participants across most digital engagement practice categories.

2.2.3 Leaderboards, Copy-Trading, and Social Surfacing

The CFA Institute report “Fun and Games: Investment Gamification and Implications for Capital Markets” documents the proliferation of leaderboards, copy-trading features, and social surfacing on retail platforms — and observes that these features “amplify the importance of social status” in ways that may encourage investors to weight peer behavior over fundamentals.

2.2.4 The Mechanism Stated in the Designer’s Voice

Tristan Harris, co-founder of the Center for Humane Technology and the technology ethicist most identified with the critique of attention engineering, has framed the mechanism as follows: “I have asymmetric access to know how to do a trade to get the outcome that I want from your behavior, whether it’s a habit formation or a belief shift or infinite scroll.” The Center’s framing — that the products are not neutral and that the business model is the design — is the closest available source for the designer-side critique of the pattern.

2.3 The Friction-Removal Primitive

2.3.1 One-Tap Orders, Default Max Buttons, Autofill Leverage

The Massachusetts complaint documented that Robinhood permitted “one Robinhood customer with no investment experience [to make] more than 12,700 trades in just over six months.” The complaint framed this as a consequence of Robinhood having “lacked adequate infrastructure” combined with “gamification strategies.”

The friction-removal primitive is not specific to stock apps. On crypto perpetuals exchanges, the default leverage setting at the position-opening interface is frequently set to 10× or higher, with sliders extending to 100× or 125× depending on the venue. The Federal Reserve Bank of New York staff report “The Financial Stability Implications of Digital Assets” (Economic Policy Review 30:2, November 2024) describes how “exchanges provide leverage in the crypto ecosystem through their facilitation of margin lending and their offering of ‘leveraged tokens,’ a type of derivative product that gives consumers leveraged exposure.”

A specific and under-discussed expression of the friction-removal primitive is fill-blending: when a trader adds to an existing position, most retail interfaces silently merge the new fill into a single position with a re-computed average entry price. The friction belonging to the second decision — the friction the trader would feel if the interface said “you are adding to a losing position; the new stop will now also expose your winning entry” — is erased. The interface defaults to absorbing the second decision into the first. This is the structural enabler of the disposition effect at the UI layer, and it is the design pattern that isolated entries (Section 1.4) is built to refuse.

2.3.2 Options Approval Without Adequate Diligence

FINRA’s June 30, 2021 Letter of Acceptance, Waiver, and Consent No. 2020066971201 — the regulatory document underlying the largest financial penalty in FINRA history — found that Robinhood “failed to exercise due diligence” before approving customers for options trading and “relied on algorithms that often approved customers to trade options based on inconsistent or illogical information.” FINRA fined Robinhood $57 million and ordered approximately $12.6 million in restitution, for a total of approximately $70 million.

2.3.3 The Alex Kearns Case

The case of Alex Kearns, opened in the foreword of this document, is the empirical end-point of friction removal. The user was approved for options without diligence. The interface displayed a misleading balance of negative $730,000, which was the result of an options spread that had not yet settled. The friction-removed channel that was easy to enter offered no friction-removed channel to escape: Kearns attempted to reach customer service three times and received no live response. He took his life that night.

His parents filed a wrongful-death lawsuit in February 2021 (Santa Clara County Superior Court). Robinhood settled in June 2021 (terms undisclosed; disclosed in Robinhood’s IPO filing). The CBS News and CNN Business reporting are the primary record.

The Kearns case is the human cost on the record. It is one death among the broader population of retail traders whose accounts were drained more slowly. The mechanism is the same — friction removed at entry, friction installed at exit — but the time horizon differs. Most of the record that follows in this dossier is the slow version of what happened to Alex Kearns.

2.4 The Leverage-Pyramid Primitive

2.4.1 Default High Leverage on Perpetuals Exchanges

Perpetuals exchanges in the offshore retail crypto category have offered up to 100× and 125× leverage. The mathematics of these settings is well understood: at 100× leverage, a 1% adverse move liquidates the position. The Federal Reserve Bank of New York staff report cited above describes how exchanges “facilitate the use of leverage by crypto investors” in ways that “create vulnerabilities from leverage, run risk, and interconnectedness.”

2.4.2 Leveraged Tokens

Binance offers leveraged tokens (BLVTs) targeting variable leverage ranges of approximately 1.5× to 3×. Binance’s own academy material warns that “leveraged tokens can be highly volatile and are not suitable for all investors. They can experience significant losses if the price of the underlying cryptocurrencies move in an unexpected direction.” Binance’s own blog post “Why You Shouldn’t Hold Leveraged Tokens Long Term” walks through a worked example showing a 13.7% drop in the leveraged token over three days against a -5% index move, and concludes: “the more volatile the underlying index/asset, the more value the token would lose over time, even if the benchmark is flat at the end of the year.”

The product’s behavior during the May 19, 2021 crypto market crash was documented in CoinDesk, “Mayhem in Binance Leveraged Tokens During Crypto Crash Leaves Traders Fuming.” Buyers of the inverse tokens BTCDOWN and ETHDOWN, expecting gains during the price decline, experienced losses. Binance’s official explanation cited algorithmic rebalancing under extreme volatility.

2.4.3 The CFD Leverage Record

Before the 2018 ESMA intervention, retail CFD platforms in the EU offered leverage ratios up to 500:1 on FX pairs. ESMA’s product intervention measure, agreed by the Board of Supervisors on March 23, 2018, capped leverage at 30:1 on major FX and as low as 2:1 on cryptocurrencies. The U.K. Financial Conduct Authority made these measures permanent in PS19/18, effective August 1, 2019, and estimated retail consumer savings of £267 million to £451 million per year as a direct consequence.

2.5 The Loss-Aversion Exploitation Primitive

2.5.1 Color, Animation, and Averaging-Down Nudges

Loss-aversion exploitation operates at the visual layer. Red/green color coding amplifies the emotional valence of unrealized P&L. Animated price changes pull the eye toward the move rather than the position-size-relative magnitude. “Average down” prompts surface as recommended actions when a position is underwater.

The Norwegian “Dark Finance” study identified loss-aversion exploitation as one of the deceptive-design taxonomies most prevalent in non-bank investment apps in the Nordic market.

2.5.2 The Disposition Effect

The relevant academic record is the disposition effect — the documented tendency of retail investors to sell winners too soon and hold losers too long. Shefrin and Statman, “The Disposition to Sell Winners Too Early and Ride Losers Too Long,” Journal of Finance (1985), named the bias. Odean, “Are Investors Reluctant to Realize Their Losses?”, Journal of Finance (1998), measured it in the discount-brokerage data: traders sell winners at a rate roughly 50% higher than they sell losers, costing them 3-5% per year in foregone returns.

The Chapkovski, Khapko, Zoican experiment (Management Science, 2026) extended the record into the gamified era and found that price-trend notifications “enhance learning for investors with accurate beliefs, but they reinforce trading mistakes for those with incorrect beliefs.” The asymmetric impact on incorrect-belief traders is the design payload.

2.5.3 Fill-Blending: The UI Mechanism of the Disposition Effect

The disposition effect is a bias of the trader. Fill-blending is the interface mechanism that operationalizes it. When a trader holds a losing position and adds at a worse price, the standard retail interface re-computes a single averaged cost basis and presents one merged position. The bad entry is no longer visible; only the friendlier blended average. The trader has not closed the losing trade — but the interface has cosmetically improved its appearance.

This is the design pattern that converts a cognitive bias into a measurable revenue stream. A trader who can see the original bad entry is more likely to cut it. A trader who sees only a softened average is more likely to hold, add, and eventually be stopped out on the combined position. The standard interface chose averaging because averaging maximizes time-in-position; time-in-position maximizes platform fees, funding, and liquidation probability.

The risk-first inversion is isolated entries (Section 1.4). Each fill is a discrete position object with its own entry price, its own stop-loss, and its own P&L. The bad entry remains visible as a bad entry. The trader who wishes to add at a worse price takes a new position — sized fresh, stopped fresh, monitored fresh. The interface refuses to absorb the second decision into the first. Clarity is preserved (every trade tells its own story), control is preserved (each decision is its own decision), and loss reduction follows (one bad trade does not corrupt the stop on a winning trade).

2.6 The Opacity Primitive

2.6.1 PFOF Concealment

The U.S. Securities and Exchange Commission’s December 17, 2020 settlement with Robinhood (Securities Act Release No. 10906) found that Robinhood had received “unusually high” payment-for-order-flow rates and had failed to adequately disclose this to customers. The settlement totaled $65 million. The structural point is documented: zero-commission trading is not free. It is paid for by the order-routing economics, which the customer does not see at the interface layer.

2.6.2 Asymmetric Pause Mechanics

The Plus500 class-action filings in Israel (Tel Aviv District Court, 2017, plaintiff Asher Torgeman, seeking class status and ~$29 million in damages) alleged that Plus500’s CFD platform “selectively paused service customers in order to prevent them from selling options they had purchased” while leaving losing positions open for trading (CTech). The case is documentary evidence of the asymmetric-pause allegation; the court’s class-certification decisions are the primary source of record. A separate U.K. Financial Ombudsman decision (DRN-3062922) addressed a 2020 Plus500 oil option pricing error in which the firm acknowledged that “prices on three of the positions had failed to reasonably reflect the underlying market,” resulting in an incorrect margin call.

2.6.3 Withdrawal Suspensions in Centralized Crypto Platforms

The Federal Reserve Bank of Chicago’s Chicago Fed Letter No. 479, “A Retrospective on the Crypto Runs of 2022” (May 2023), documents the bankruptcies of Celsius, Voyager Digital, BlockFi, Genesis, and FTX. Total customer withdrawals from these platforms were approximately $13 billion in the run-up to bankruptcy. FTX experienced “one of the most severe episodes,” with customers withdrawing approximately a quarter of platform deposits in a single day. The same letter notes that “owners of large-sized accounts, i.e., with over $500,000 in investments, were the fastest to withdraw” and “withdrew proportionately more of their funding.”

The Chicago Fed’s framing is precise: while platforms had many retail clients, the runs were spearheaded by larger accounts, leaving retail with frozen accounts and large losses. The structural opacity — retail customers did not have visibility into the platforms’ liquidity profiles — is the mechanism through which the asymmetry was sustained until it failed catastrophically.

The New York State Attorney General’s 2023 complaint against Celsius founder Alex Mashinsky, and the U.S. Department of Justice’s prosecution of FTX founder Samuel Bankman-Fried, are the primary judicial records of the misrepresentations that operated as opacity primitives at the marketing layer.

2.7 The Six Primitives, the One Pattern

Celebration of volume. Attention engineering. Friction removal. Leverage pyramid. Loss-aversion exploitation. Opacity. Different surfaces. Same architectural property: the platform’s revenue function is decoupled from the trader’s outcome function. Every primitive is an interface expression of that decoupling. Every risk-first inversion is a re-coupling — at the data layer (clarity), at the decision layer (control), at the position layer (loss reduction), at the aggregate-visibility layer (aggregate-risk visibility), at the fee layer (alignment).

The on-chain alternative — verifiable fee records, transaction-hash-level attribution, cryptographically signed fee receipts — does not solve every form of opacity, but it makes the platform-to-builder fee record verifiable in a way the PFOF record was not. Isolated entries is the position-layer analogue: opacity at the position layer is fill-blending; clarity at the position layer is per-entry visibility.

Part III — The Investigative Record

This section preserves the original investigative-journalism record across the three platform categories, reported as documentary evidence.

3.1 Bloomberg Businessweek: The Confetti Was the Surface

Misyrlena Egkolfopoulou, Annie Massa, and Anders Melin reported the foundational Bloomberg piece on Robinhood’s design lineage, “Has Robinhood Made Day Trading Too Irresistible?” (April 21, 2021). The article reports that Nir Eyal, author of Hooked: How to Build Habit-Forming Products, was approached in a Palo Alto coffee shop by Robinhood co-founder Baiju Bhatt before the company’s launch, and that Eyal “can see in the app many of the successful design features he’s identified,” though Robinhood disputes that Hooked influenced the design.

The article quotes Brett Steenbarger, professor of psychiatry and behavioral sciences at SUNY Upstate Medical University: “It’s a different technology of high stimulation and high distractibility. With the traditional brokers, you haven’t had that.”

The article also quotes Eyal himself, by then increasingly preoccupied with the negative consequences of the techniques he had documented, from his 2019 book Indistractable: “Companies making their products more engaging isn’t necessarily a problem—it’s progress. But there’s also a dark side. For many people, these distractions can get out of hand, leaving us with a feeling that our decisions are not our own.”

The Bloomberg framing was load-bearing for the regulatory record that followed: confetti was the most visible primitive, but confetti was a symptom, not the disease.

3.2 The Massachusetts Administrative Record

The Massachusetts Securities Division’s complaint (Docket No. E-2020-0047, December 2020) is the most-cited regulatory document in the gamification debate. The original 23-page filing and the amended complaint are public.

The complaint quotes Secretary Galvin: “Treating this like a game and luring young and inexperienced customers to make more and more trades is not only unethical, but also falls far short of the standards we require in Massachusetts.” The January 2024 settlement, $7.5 million, required Robinhood to overhaul its digital engagement practices for Massachusetts customer accounts.

3.3 The FINRA Record

FINRA’s June 30, 2021 announcement: “FINRA announced today that it has fined Robinhood Financial LLC $57 million and ordered the firm to pay approximately $12.6 million in restitution, plus interest, to thousands of harmed customers. The sanctions represent the largest financial penalty ever ordered by FINRA and reflect the scope and seriousness of the violations.”

Jessica Hopper, then FINRA executive vice president and head of enforcement, stated: “The fine imposed in this matter, the highest ever levied by FINRA, reflects the scope and seriousness of Robinhood’s violations, including FINRA’s finding that Robinhood communicated false and misleading information to millions of its customers.”

3.4 The Crypto Bankruptcy Record

The Federal Reserve Bank of Chicago Letter No. 479 (May 2023) is the cleanest single source on the 2022 crypto-platform failures. The figures: FTX had 9.7 million customers at peak per its bankruptcy lawyers, but only 1.9 million with positive balances at the bankruptcy filing. Voyager Digital had reported 3.5 million customer registrations and 1.19 million funded accounts at its peak, yet 975,000 customers held positive balances at bankruptcy. Celsius cumulatively had approximately 1.8 million customers.

The U.S. Department of Justice’s prosecution of Samuel Bankman-Fried, the New York Attorney General’s complaint against Alex Mashinsky, and bank-regulator actions documenting Voyager Digital’s misrepresentations regarding deposit insurance (“USD held with Voyager is FDIC insured up to $250K,” per the Voyager Twitter post cited in the Chicago Fed letter) are the primary judicial records.

3.5 The CFD/Forex Investigative Record

The CFD class-action and complaint record is documented across multiple jurisdictions. The Plus500 CTech filing (January 2017) and the Giambrone-coordinated complaint pool (2017–2018) are the most-cited cases. The Plus500 U.K. account-freeze episode (May 2015) — when the FCA’s Plus500UK Limited supervisory action required the firm to freeze approximately 55% of its U.K. customer accounts pending enhanced AML procedures — is the regulatory-action equivalent.

3.6 FTX TixWix and the Gamified-Options Lineage

The October 28, 2020 Hxro–FTX TixWix announcement is worth quoting in full because it is the on-record self-description of the gamification design intent at a major derivatives venue. From Sam Bankman-Fried, then CEO of FTX: “FTX’s motto is ‘by traders, for traders.’ Part of that means constantly finding new ways to make trading more fun and interactive, and giving users better ways to express their market beliefs.” FTX subsequently filed for bankruptcy in November 2022.

Part IV — The Academic Record

This section consolidates the peer-reviewed empirical record. Where the investigative record establishes that the patterns exist, the academic record establishes their consequences.

4.1 Day Trading Outcome Studies

Brazil. Chague, F., De-Losso, R., and Giovannetti, B., “Day Trading for a Living?” Working Paper 2019_47, University of São Paulo Department of Economics (June 11, 2020), available at SSRN.

The headline numbers from this study are the most-cited empirical finding in the entire retail-trading literature. Of all individuals who began day trading the Brazilian equity index futures market between 2013 and 2015 and persisted for at least 300 days, 97% lost money. Only 1.1% (17 individuals out of 1,551) earned more than the Brazilian minimum wage. Only 0.5% (8 individuals) earned more than the starting salary of a bank teller (~$54/day). The top individual earner averaged $310/day with a standard deviation of $2,560. The authors found “no evidence of learning by day trading.”

The authors’ direct conclusion, quoted from the paper: “In aggregate, day trading is a losing proposition; day trading is an industry that consistently and reliably loses money. From an industrial organization perspective, it is difficult to understand how such an industry survives.”

Taiwan. Barber, B. M., Lee, Y. T., Liu, Y. J., and Odean, T., “Just How Much Do Individual Investors Lose by Trading?” Review of Financial Studies 22:2, 609–632 (2009). The study analyzed every trade on the Taiwan Stock Exchange from 1992 to 2006 and found that less than 1% of day traders earned persistent positive returns net of fees.

4.2 The Disposition Effect

Shefrin, H., and Statman, M., “The Disposition to Sell Winners Too Early and Ride Losers Too Long: Theory and Evidence,” Journal of Finance 40:3, 777–790 (1985), named the bias. The framing — that traders prefer to realize gains and defer losses — gave subsequent decades of empirical work a clean theoretical anchor.

Odean, T., “Are Investors Reluctant to Realize Their Losses?”, Journal of Finance 53:5, 1775–1798 (1998), provided the foundational measurement. Using a sample of 10,000 accounts at a large discount brokerage, Odean found that investors sell winners at a rate approximately 50% higher than they sell losers, and that this asymmetry costs them several percentage points per year in foregone returns. The bias is robust across tax-motivated and non-tax-motivated periods, which excludes tax-loss harvesting as the explanation. The Odean paper is the most-cited single source in the disposition-effect literature.

The mechanism documented by Shefrin-Statman and Odean is the cognitive bias the trader brings to the interface. The interface design choice that converts the bias into an extraction mechanism — fill-blending into a single averaged position, rather than per-entry isolation — is documented in Section 2.5.3 and named as the risk-first inversion in Section 1.4.

4.3 Attention-Induced Trading

Barber, Huang, Odean, and Schwarz (Journal of Finance 77:6, 2022) found average 20-day abnormal returns of -4.7% on the top stocks purchased each day on Robinhood. The mechanism — attention-induced concentration of buy orders into stocks already in the news — is the empirical link between the attention-engineering primitive and realized retail underperformance.

The earlier Barber–Odean paper, “All That Glitters” (RFS 21:2, 2008), is the foundational reference.

A subsequent working paper, Barber, Lin, and Odean, “Resolving a Paradox: Retail Trades Positively Predict Returns but Are Not Profitable” (2023), reconciles the apparent contradiction in the order-imbalance literature. The paradox: order-imbalance studies appear to show retail trades positively predict returns, while account-level studies show retail traders lose money. The resolution: retail buying concentrates in attention-grabbing stocks that subsequently underperform. Long-short strategies based on extreme quintiles of retail order imbalance earn -15.3% annualized among heavy-retail-trading stocks, versus +6.8% among other stocks.

4.4 The Causal Gamification Experiment

Chapkovski, Khapko, and Zoican, “Trading Gamification and Investor Behavior,” Management Science 72:1, 32–56 (2026), is the first published randomized experiment isolating the causal effect of hedonic gamification on retail trader behavior. The headline findings, in the authors’ phrasing: “Participants with lower financial literacy prefer platforms with hedonic gamification elements, such as confetti and achievement badges. On average, hedonic gamification increases trading volume by 5.17%.”

A follow-up paper, “Gamified Risk-Taking” in Journal of Behavioral and Experimental Finance 46 (June 2025), extended the experimental protocol to 605 participants across four countries.

4.5 The CFD/Forex Empirical Record

ESMA’s CFD Task Force documented that across EU jurisdictions, 74% to 89% of retail CFD trader accounts lose money, with average losses per client ranging from €1,600 to €29,000 (ESMA-71-98-128, March 27, 2018). The supplementary technical paper (ESMA-35-43-1000) provides the methodology.

The Spanish CNMV’s January 2015 – September 2016 study: 82% of Spanish CFD clients suffered losses totaling €142 million including fees.

The U.K. FCA: approximately 80% of CFD-trading retail accounts unprofitable, per FCA statements over multiple years (FCA PS19/18 and the 2018 announcement materials).

The Australian ASIC: prior to its 2021 leverage reforms, retail traders in CFDs lost more than AUD $770 million in a single observed period (ASIC product-intervention order materials).

4.6 The Yale Law Journal Forum Essay

Langvardt, K., and Tierney, J. F., “On ‘Confetti Regulation’: The Wrong Way to Regulate Gamified Investing,” 131 Yale L.J. Forum 717 (January 17, 2022).

The essay’s framing of the underlying business model is the cleanest source for the structural argument: “Zero-commission investing apps like Robinhood have a business model that requires clients to trade as much as possible. To that end, these apps incorporate design features that are sometimes called ‘gamification’: behavioral prompts and flashy casino-like design elements that encourage unreflective or unconsidered decision making based on cognitive bias, imperfect rationality, and impulse.”

The essay argues that direct regulation of the aesthetic features alone — what the authors term “confetti regulation” — is constitutionally fraught and unlikely to address the structural cause. The implicit argument: if the business model requires the engagement, removing the surface symbols does not remove the underlying design pressure.

4.7 The Springer “Dark Finance” Paper

Rakovic, I., and Inal, Y., “Dark Finance: Exploring Deceptive Design in Investment Apps,” in J. Abdelnour Nocera et al. (eds.), INTERACT 2023, Springer LNCS. The study examined 26 investment apps available in Norway and found that “nearly all the studied apps incorporate dark patterns to varying degrees, and the manipulation level using these practices differs between bank and non-bank apps. Banks have more transparent apps with fewer dark patterns.”

4.8 The Federal Reserve Crypto Stability Record

Federal Reserve Bank of Chicago, Chicago Fed Letter No. 479, “A Retrospective on the Crypto Runs of 2022” (May 2023).

Federal Reserve Bank of New York, “The Financial Stability Implications of Digital Assets,” Economic Policy Review 30:2 (November 2024).

4.9 The Crypto-Gambling Public-Health Literature

Mills, D. J., and Nower, L., “Preliminary findings on cryptocurrency trading among regular gamblers: A new risk for problem gambling?” Addictive Behaviors 92, 136–140 (2019).

Delfabbro, P., King, D., and Williams, J., “Cryptocurrency trading, gambling and problem gambling,” Addictive Behaviors 122, 107021 (2021).

Oksanen, A., Mantere, E., Vuorinen, I., et al., “Gambling and online trading: emerging risks of real-time stock and cryptocurrency trading platforms,” Public Health 205, 72–78 (2022).

Newall, P., et al., “Cryptocurrencies as Gamblified Financial Assets and Cryptocasinos: Novel Risks for a Public Health Approach to Gambling,” Risks 11:3, 49 (2023).

The scoping review by Johnson et al., “Cryptocurrency trading and its associations with gambling and mental health,” summarizes the field: “Our scoping review indicates a likely relationship between problem gambling and cryptocurrency trading. Findings also suggest overlap with high-risk stock traders, with similarities in gambling behaviors, demographics, and personality traits.”

The Lancet Public Health editorial “Cryptocurrency and new financial instruments: unquantified public health harms” frames the policy implication: “Although traditional long-term stocks and shares have underlying assets and might pay dividends, other products such as high-frequency trading have features of gamblification; they lead most investors to lose, they attract people at risk of experiencing gambling-related harm, and either encourage high frequency of use or promise sizeable imminent wins.”

Part V — Voices: Designers, Ethicists, and Investors on Record

This section preserves verbatim quotations from established figures who have placed themselves on the record about the pattern. The citation discipline is non-negotiable; these are quoted, not paraphrased.

5.1 Charlie Munger

At the Daily Journal annual meeting, February 2022:

“We have a stock market which some people use like a gambling parlor. And the transactions of the people who love the gambling parlor aspect of the business and those who want to make long term investments, to take care of their old age and so forth — I mean, muddle that in one market and it goes out of control because the stock market becomes an ideal gambling parlor activity.”

At the Berkshire Hathaway annual meeting, May 2021:

“I think it’s just God awful that something like that brought investments from civilized men and decent citizens. … It’s deeply wrong.”

At the Berkshire Hathaway annual meeting, April 2022:

“Disgusting. … short-term gambling and big commissions and hidden kickbacks and so on.”

5.2 Warren Buffett

At the Berkshire Hathaway annual meeting, May 2021:

“[Robinhood] become a very significant part of the casino aspect, the casino group, that has joined into the stock market in the last year or year and a half. … The degree to which a very rich society can reward people who now know how to take advantage essentially of the gambling instincts of not only the American public but the worldwide public, it’s not the most admirable part of the accomplishment. … There’s nothing illegal about it, there’s nothing immoral but I don’t think you’d build a society around people doing it.”

5.3 Howard Marks

Marks, H., “You Bet!” Oaktree Capital memo (January 2020). The memo’s central observation:

“Success in gambling doesn’t go to those who pick winners, but to those with the ability to identify superior propositions. The goal is to find situations where the odds are generous to one side or the other, whether favorite or underdog. In other words, a mispricing. It’s exactly the same in investing.”

5.4 Tristan Harris

Center for Humane Technology co-founder Tristan Harris, on the architecture of attention engineering:

“I think of it like insider trading on your nervous system. I have asymmetric access to know how to do a trade to get the outcome that I want from your behavior, whether it’s a habit formation or a belief shift or infinite scroll.”

5.5 Harry Brignull

Brignull, who coined the term “dark pattern” in 2010 and now uses “deceptive pattern”:

“A Dark Pattern is a manipulative or deceptive trick in software that gets users to complete an action that they would not otherwise have done, if they had understood it or had a choice at the time.”

5.6 Nir Eyal

Eyal, Indistractable (Portfolio, 2019), opening:

“Companies making their products more engaging isn’t necessarily a problem—it’s progress. But there’s also a dark side. For many people, these distractions can get out of hand, leaving us with a feeling that our decisions are not our own.”

5.7 Brett Steenbarger

Professor of psychiatry and behavioral sciences, SUNY Upstate Medical University, quoted in Bloomberg Businessweek (April 21, 2021):

“It’s a different technology of high stimulation and high distractibility. With the traditional brokers, you haven’t had that.”

5.8 Jessica Hopper, FINRA

Then-Executive Vice President and Head of Enforcement, FINRA, on the Robinhood penalty (June 30, 2021):

“The fine imposed in this matter, the highest ever levied by FINRA, reflects the scope and seriousness of Robinhood’s violations, including FINRA’s finding that Robinhood communicated false and misleading information to millions of its customers.”

5.9 William Galvin

Secretary of the Commonwealth of Massachusetts (December 2020 announcement):

“Treating this like a game and luring young and inexperienced customers to make more and more trades is not only unethical, but also falls far short of the standards we require in Massachusetts.”

5.10 Steven Maijoor, then Chair of ESMA

On the March 2018 product intervention:

“The combination of the promise of high returns, easy-to-trade digital platforms, in an environment of historical low interest rates has created an offer that appeals to retail investors. However, the inherent complexity of the products and their excessive leverage – in the case of CFDs – has resulted in significant losses for retail investors.”

Part VI — The Economics of Loss

This section consolidates the loss numbers — the regulatory disclosures, bankruptcy filings, and academic outcome studies that quantify the audience-loss side of the rentability pattern.

6.1 Retail Day Trading Outcomes

Population Outcome Source
Brazilian futures day traders, 300+ days persistence (2013–2017) 97% lost money Chague, De-Losso, Giovannetti (2020)
Brazilian futures day traders earning > Brazilian minimum wage 1.1% (17 of 1,551) Same
Brazilian futures day traders earning > bank teller starting salary 0.5% (8 of 1,551) Same
Taiwan day traders earning persistent positive net returns (1992–2006) <1% Barber, Lee, Liu, Odean (2009)

6.2 CFD Retail Outcomes

Disclosure Range Source
EU retail CFD account loss rate 74% – 89% ESMA, March 2018
Average loss per losing client, EU €1,600 – €29,000 ESMA, March 2018
Spanish CFD client loss rate, 2015–2016 82% CNMV cited in ESMA
U.K. retail CFD account unprofitable rate ~80% FCA, ongoing
Estimated annual U.K. retail savings from FCA CFD restrictions £267M – £451M FCA PS19/18

6.3 Robinhood-Era U.S. Retail Outcomes

The Barber–Huang–Odean–Schwarz finding of -4.7% average 20-day abnormal return on top Robinhood stocks is the cleanest single retail-attention number on record.

6.4 The Disposition-Effect Drag

Odean (1998): individual investors sell winners at a rate approximately 50% higher than they sell losers; the bias costs the average account several percentage points per year in foregone returns. The cost is an annuity — it compounds for as long as the trader remains in the standard blended-position interface that makes the bias frictionless to act on.

6.5 The 2022 Crypto Bankruptcy Sequence

Platform Bankruptcy Date Customers at Filing Run Outflow
Celsius July 2022 ~1.8M cumulative 35% June withdrawals from $1M+ accounts
Voyager Digital July 2022 975,000 with positive balances —
FTX November 2022 1.9M with positive balances (9.7M peak) 25% in single day, 36.7% over 5 days
BlockFi November 2022 — —
Genesis January 2023 ~340,000 Gemini-routed retail —
Total customer outflows — — ~$13B

Source: Federal Reserve Bank of Chicago Letter No. 479 (May 2023).

6.6 Regulatory Penalty Record

Action Amount Date Source
FINRA v. Robinhood (largest FINRA penalty ever) $70M (≈$57M fine + $12.6M restitution) June 30, 2021 FINRA AWC 2020066971201
SEC v. Robinhood (PFOF disclosure) $65M December 17, 2020 SEC Release 33-10906
Massachusetts v. Robinhood (gamification settlement) $7.5M January 2024 Mass. Securities Division

Part VII — The Architectural Alternative: Alignment as Primitive

The patterns documented above are not bugs of any single exchange or app. They are the designed outcome of a business model in which platform revenue grows when the audience loses.

The phrase “house edge” is on the public record from Munger. The phrase “casino group” is on the record from Buffett. The phrase “gambling parlor” is on the record from Munger. The empirical 74% – 89% retail loss rate is on the record from ESMA. The 97% Brazilian day-trading loss rate is on the record from a São Paulo working paper now cited across the academic literature. The $70 million FINRA penalty is on the record. The 23-page Massachusetts complaint is on the record. The Yale Law Journal Forum essay states the structural cause directly: “a business model that requires clients to trade as much as possible.”

The patterns at the surface — confetti, top-movers lists, default high leverage, asymmetric pause mechanics, friction-removed entry and friction-loaded exit, fill-blending into averaged positions — are the symptoms. The cause is the alignment of revenue with volume rather than with trader survival.

The architectural primitive that closes the gap is alignment. A fee structure in which the platform earns more when the trader survives longer has different defaults at the interface layer. The same designers, the same engineers, the same psychology of the user — all of it is the same. The objective function is different. When trader survival is the metric, the interface optimizes for it.

7.1 Clarity, Control, Loss Reduction, Aggregate-Risk Visibility: The Four Deliverables

The risk-first interface delivers four properties, named in Section 1.4 and repeated here because they are the affirmative architecture this dossier is built around.

Clarity. Every trade is its own object. Every entry has its own price. Every entry has its own P&L. The trader can see what they did, when they did it, and what each decision cost or earned. There is no averaged cost basis that hides the bad entry inside a friendlier blended number. The trader’s performance review is honest because the data is honest.

Control. Every decision is its own decision. Adding to a position is a new trade, not a silent edit to an old one. The trader cannot accidentally average down, because averaging-down is not a primitive the interface offers. The disposition effect (Shefrin-Statman 1985; Odean 1998) is not eliminated — it is a property of the trader, not the platform — but the interface stops actively converting it into platform revenue.

Loss reduction. Every trade has its own stop. A losing entry is contained to the losing entry. The winning trades continue. The contamination of a winning position by a subsequent losing entry — the single most expensive UX pattern in retail trading — is structurally prevented. One bad decision does not pull the others down with it.

Aggregate-risk visibility. Per-entry clarity is the most granular view of the trader’s exposure. The aggregate worst-case is the most aggregate view. Between them, every dollar of risk the trader has on is visible. The number — the dollar amount lost if every stop and every liquidation triggers from this moment forward — is surfaced in real time, in a single visible field, across all open positions simultaneously. The trader does not compute it. The interface computes it. Narrow framing (Kahneman and Lovallo, Management Science 1993) is the cognitive bias the absence of this metric exploits; aggregate-risk visibility is the interface property that refuses the exploitation.

These four properties are the affirmative case for alignment. They are what a trader gets in exchange for the absence of confetti.

7.2 The Architectural Object: Isolated Entries

Isolated entries is the design pattern that delivers the first three properties at once. Each fill is a discrete position object. Each fill has its own collateral, its own entry price, its own stop-loss, and its own P&L. The interface refuses to merge fills into a single averaged position. There is no “average down” button. There is no blended cost basis. There is no single liquidation price that exposes a winning entry to a losing entry’s risk.

Isolated entries is stricter than isolated margin. Isolated margin protects one position from another. Isolated entries protects one decision from another. The architectural object is finer-grained; the trader’s protection is correspondingly tighter.

Isolated entries is also the object that makes the per-decision review of trader performance possible. A trader who can see twenty separate entries can identify which entries worked and which did not. A trader who sees one blended position can only identify the blend. The first trader can learn. The second trader cannot — and the Chague-De-Losso-Giovannetti finding of “no evidence of learning by day trading” should be read in part as a finding about the interfaces those day traders used, not only about the traders themselves.

The fourth property — aggregate-risk visibility — is computed across the isolated entries. Where isolated entries makes every individual exposure honest, the aggregate-risk-visibility computation makes the sum of those exposures honest. The two architectural objects are complementary: the per-entry container preserves the granular view; the aggregate-risk computation preserves the portfolio view. Without the second, the trader who has been protected from per-entry contamination still does not see, at any single moment, the dollar amount they would lose if every one of those independent entries went against them simultaneously. The first three properties protect each decision from the others. The fourth makes the sum of decisions legible.

7.3 The Full Risk-First Stack

What that looks like, primitive by primitive:

House Edge Primitive Documented Effect Risk-First Alternative
Confetti on trade execution +5.17% trading volume (Chapkovski et al., 2026) No celebratory imagery on execution
“Top movers” lists, push notifications -4.7% 20-day abnormal returns on top-purchased stocks (Barber et al., 2022) No leaderboards. No attention-engineering surfaces. No ads. No pop-ups.
One-tap order entry as first-class action 12,700 trades by single inexperienced customer (Mass. complaint) 2-click stop-loss as first-class action
Default high leverage (10× to 125×) Liquidation cascades; 74% – 89% CFD retail loss rate (ESMA) Isolated entries with conservative per-entry leverage defaults; position-sizing guidance at order entry
Fill-blending into averaged positions Disposition-effect amplification (Shefrin-Statman 1985; Odean 1998 — 3-5% annual return drag) Isolated entries: each fill is a discrete position with its own entry price, stop-loss, and P&L
Red/green color, averaging-down nudges Disposition-effect amplification at the visual layer Per-entry P&L visibility; stop-loss surfaced per entry, not per blended position
Concealed PFOF / counterparty-position opacity $65M SEC settlement (Robinhood) On-chain fee record. Cryptographically signed fee receipts. Verifiable transaction hash. Permanent ledger.
Aggregate-risk opacity (no surveyed retail platform surfaces the total worst-case dollar floor across all open positions — Binance, OKX, Bybit, Bitget, Robinhood, Plus500, IG, eToro, CMC Markets, Hyperliquid native, MetaMask Perps, Phantom, pvp.trade, Hyperdash, HyperTracker, ASXN, Dexly) Narrow framing — investors evaluate risks individually and misjudge portfolio risk (Kahneman and Lovallo, Management Science 1993) Real-time aggregate worst-case dollar floor surfaced as a single visible field updating across all open positions

7.4 The Closing Argument

The architectural primitive is not the absence of design. It is design that points the other direction. The same care that goes into engineering volume can be turned to engineering clarity, control, loss reduction, and aggregate-risk visibility.

Closing

The 23-page Massachusetts complaint, the $70 million FINRA fine, the 97% Brazilian loss rate, the Yale Law Journal Forum essay that names the cause, the Munger transcript, the Buffett transcript, the Odean disposition-effect measurement — all of it is on the public record. The pattern is not in dispute.

What is in dispute is whether the people building the next interface read the record before they default the leverage slider to 100×, before they let the trader average down without friction, before they blend the next fill into a friendlier average. Alex Kearns’s options approval came from an algorithm that, FINRA later found, “approved customers based on inconsistent or illogical information.” The architecture that approved him is the architecture this dossier catalogs.

The architecture that would have refused him — clarity at the data layer, control at the decision layer, loss reduction at the position layer, aggregate-risk visibility at the portfolio layer, alignment at the fee layer — is on the public record waiting to be built. The patterns are documented. The deliverables are named. Every trade tells its own story. The economics follow.

Sources / Bibliography

Regulatory and Government

Academic

  • Shefrin, H., and Statman, M., “The Disposition to Sell Winners Too Early and Ride Losers Too Long: Theory and Evidence,” Journal of Finance 40(3):777–790 (1985).
  • Odean, T., “Are Investors Reluctant to Realize Their Losses?”, Journal of Finance 53(5):1775–1798 (1998).
  • Chague, F., De-Losso, R., and Giovannetti, B., “Day Trading for a Living?” (June 11, 2020), SSRN. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3423101
  • Barber, B. M., and Odean, T., “All That Glitters: The Effect of Attention and News on the Buying Behavior of Individual and Institutional Investors,” Review of Financial Studies 21(2):785–818 (2008). https://papers.ssrn.com/sol3/papers.cfm?abstract_id=460660
  • Barber, B. M., Lee, Y. T., Liu, Y. J., and Odean, T., “Just How Much Do Individual Investors Lose by Trading?” Review of Financial Studies 22(2):609–632 (2009).
  • Barber, B. M., Huang, X., Odean, T., and Schwarz, C., “Attention-Induced Trading and Returns: Evidence from Robinhood Users,” Journal of Finance 77(6):3141–3190 (2022). https://onlinelibrary.wiley.com/doi/abs/10.1111/jofi.13183
  • Barber, B. M., Lin, S., and Odean, T., “Resolving a Paradox: Retail Trades Positively Predict Returns but Are Not Profitable” (March 15, 2023), SSRN. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3783492
  • Chapkovski, P., Khapko, M., and Zoican, M., “Trading Gamification and Investor Behavior,” Management Science 72(1):32–56 (2026). https://pubsonline.informs.org/doi/10.1287/mnsc.2022.02650 ; SSRN https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3971868
  • Chapkovski, P., Khapko, M., and Zoican, M., “Gamified Risk-Taking,” Journal of Behavioral and Experimental Finance 46 (June 2025). https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4921361
  • Langvardt, K., and Tierney, J. F., “On ‘Confetti Regulation’: The Wrong Way to Regulate Gamified Investing,” 131 Yale L.J. Forum 717 (January 17, 2022). https://www.yalelawjournal.org/forum/on-confetti-regulation-the-wrong-way-to-regulate-gamified-investing
  • Rakovic, I., and Inal, Y., “Dark Finance: Exploring Deceptive Design in Investment Apps,” in INTERACT 2023 Proceedings, Springer LNCS. https://link.springer.com/chapter/10.1007/978-3-031-42280-5_20
  • Mills, D. J., and Nower, L., “Preliminary findings on cryptocurrency trading among regular gamblers,” Addictive Behaviors (2019).
  • Delfabbro, P., King, D., Williams, J., “Cryptocurrency trading, gambling and problem gambling,” Addictive Behaviors 122:107021 (2021). https://pubmed.ncbi.nlm.nih.gov/34171583/
  • Johnson, B. et al., “Cryptocurrency trading and its associations with gambling and mental health: A scoping review.” https://pubmed.ncbi.nlm.nih.gov/36174424/
  • Newall, P. et al., “Cryptocurrencies as Gamblified Financial Assets and Cryptocasinos,” Risks 11(3):49 (2023). https://www.mdpi.com/2227-9091/11/3/49
  • Lancet Public Health editorial, “Cryptocurrency and new financial instruments: unquantified public health harms.” https://www.thelancet.com/journals/lanpub/article/PIIS2468-2667(22)00173-6/fulltext
  • Kelleher, D. M., Grimes, J., and Chovil, A., “Democratizing Equity Markets With and Without Exploitation,” 44 W. New Eng. L. Rev. 51 (2022). https://digitalcommons.law.wne.edu/cgi/viewcontent.cgi?article=1858&context=lawreview
  • Kahneman, D., and Lovallo, D., “Timid Choices and Bold Forecasts: A Cognitive Perspective on Risk Taking,” Management Science 39(1):17–31 (1993).
  • Tversky, A., and Kahneman, D., “The Framing of Decisions and the Psychology of Choice,” Science 211(4481):453–458 (1981).
  • Barberis, N., and Huang, M., “Preferences with Frames: A New Utility Specification That Allows for the Framing of Risks,” Journal of Economic Dynamics and Control 33(8):1555–1576 (2009).
  • Beshears, J., Choi, J. J., Laibson, D., and Madrian, B. C., “Does Aggregated Returns Disclosure Increase Portfolio Risk Taking?” Review of Financial Studies 30(6):1971–2005 (2017). https://doi.org/10.1093/rfs/hhw086
  • Redelmeier, D. A., and Tversky, A., “On the Framing of Multiple Prospects,” Psychological Science 3(3):191–193 (1992).
  • Langer, T., and Weber, M., “Prospect Theory, Mental Accounting, and Differences in Aggregated and Segregated Evaluation of Lottery Portfolios,” Management Science 47(5):716–733 (2001).
  • Thaler, R. H., Tversky, A., Kahneman, D., and Schwartz, A., “The Effect of Myopia and Loss Aversion on Risk Taking: An Experimental Test,” Quarterly Journal of Economics 112(2):647–661 (1997).

Investigative Journalism

Books and Long-Form Voices

Industry/Self-Description

All claims in this document are linked or quoted to primary sources. The document is intended as a canonical reference for audience-builders, journalists, regulators, and academics on the rentability pattern in retail trading interfaces.