The opportunity in plain terms
We separate the trading process into three layers — research, strategy development, and trade execution — and the market into three user types: retail retail, individual amateur traders, and institutional investors.
The largest investment opportunities we've observed are in individual amateur traders and institutions:
- LLMs are enabling a brand-new knowledge acquisition approach for amateur traders with prior knowledge who are eager for analysis. They benefit from LLMs' capability of context understanding and reasoning across massive data — applied to market research, trading signals, strategy recommendations, and backtesting.
- LLMs would aid institutions by assisting trading decisions, supporting research and signal generation with deeper analysis, higher data accuracy, more customization, and stricter rigor and compliance. They can also automate mid- and back-office tasks such as trade reconciliation, exception management, and report generation.
This paper primarily focuses on individual amateur traders. A separate report will be provided for the institutional investor segment.
Mapping the workflow
The trading process can be separated into three parts (simplified):
- Conduct Research. The systematic process of gathering, analyzing, and interpreting market or alternative data to form insights about asset prices, market behavior, or potential opportunities. Components include:
- Quantitative analysis — statistical studies of price patterns, order flow, volatility
- Fundamental research — macroeconomic indicators, earnings, monetary policy
- Sentiment and alternative data — news, social media, satellite imagery
- Backtesting and simulations — testing models or ideas on historical data
- Define Strategy. A trading strategy is a defined, executable plan specifying when, what, and how to trade — entry/exit rules, position sizing, risk management (stop-loss, portfolio exposure), and execution logic (market vs. limit orders, speed, slippage controls).
- Trade Execution. Transforming a trading decision into an actual transaction by placing an order through a brokerage platform, which then handles settlement, compliance, and market practices. Categories include established financial institutions (Fidelity, Charles Schwab), digital trading platforms (Robinhood, Moomoo), and alternative platforms (Acorn, Public.com).
User Segmentation
| Segment | Description | Behavior |
|---|---|---|
| Retail Retail | Everyday individuals who trade occasionally with small capital. | Casual, intuitive, often mobile-only. |
| Individual Amateur Traders | Active individuals learning or practicing trading semi-seriously. | Technical, self-directed, uses tools & forums. |
| Institutional Investors | Firms managing pooled or large-scale capital (funds, banks). | Professional, policy-bound. |
Many individual amateur traders actively engage in day or swing trading. They seek in-depth analytical tools and informed investment recommendations. LLMs offer significant potential here — providing comprehensive data analysis and transparent reasoning, empowering individual traders with enhanced professional insights.
LLM Capability in Trading Tools
| Capability | What it means | Where it shows up |
|---|---|---|
| Contextual Understanding | AI/LLMs offer significant advantages in retail investment research, processing vast amounts of financial data — particularly text and multimodal information like earnings call transcripts and analyst reports. | In fundamental analysis, LLMs analyze metrics like P/E ratio, revenue, and EPS to forecast performance. In technical analysis, they explain indicators such as RSI, MACD, Bollinger Bands, and Fibonacci retracements. They also capture event and market sentiment data in real time. |
| Reasoning | Ability to generalize across diverse data formats and topics, coupled with the capacity to articulate clear reasoning for investment decisions. Note: accuracy remains unconfirmed. | LLM reasoning provides in-depth, step-by-step analyses of strategies — factoring in risk preferences, assumptions, and historical examples — and can backtest and adjust strategies accordingly. |
| Personalization | AI agents trained on a user's financial goals, risk appetite, and life stage can deliver tailored recommendations. | Portfolio construction aligned with target risk/reward, tax-loss harvesting suggestions, rebalancing alerts based on market moves or life events, and autonomous-but-controllable agents executing rules on the investor's behalf. |
Sizing the retail surge
The volume of market data generated by retail traders — both individual amateurs and formally recognized retail entities — has demonstrably risen in recent years. Retail's share of total U.S. equities trading volume began increasing in 2024 and has gradually trended upward.
- 2021 peak: Equity allocations reached a high, driven by strong market performance and investor optimism.
- 2022 dip: A modest decrease driven by market volatility and rising interest rates.
- 2024 rebound: Allocations rose again, supported by interest-rate stabilization and the AI-led rally.
Growth in New Retail Brokerage Accounts
- 2020: Over 10 million new brokerage accounts opened, driven by pandemic lockdowns, stimulus payments, and the rise of commission-free trading platforms.
- 2021: Momentum continued — approximately 30 million new brokerage accounts opened across 2020 and 2021 combined.
- 2022–2025: Charles Schwab reported a 23% rise in new brokerage accounts in 2024, contributing $367 billion in new assets that year.
Amateur Trader Market Sizing
| Brokerage Firm | Active Accounts (2025E) | Amateur Traders |
|---|---|---|
| Fidelity | 51.5M | 20% = 3.2M |
| Charles Schwab | 37M | 20% = 2.3M |
| Robinhood | 23.6M | 10% = 2.3M |
| Interactive Brokers | 3.7M | 50% = 0.37M |
| E*TRADE | 5M | 20% = 3.1M |
| Webull + Moomoo + tastytrade + Apex | ~8M | 10% = 0.8M |
Global amateur-trader accounts: ~12.05M. The U.S. accounts for roughly 35–40% of the global retail (amateur) trader market by active brokerage accounts and trading volume.
U.S. Accounts
~4.2M
Median Spend / Mo
$200
Annual Market
$10B
Range: $100–$500/mo per amateur trader. Market sizing = $200 × 4,217,500 × 12.
Institutional Trader Market Sizing (U.S.)
We estimate institutional traders by category and typical front-office headcount (traders, portfolio managers, analysts).
| Institution Type | # Institutions | Avg Traders/PMs | Total Traders |
|---|---|---|---|
| Hedge Funds | ~3,500 | 5–15 | ~35,000 |
| Asset Managers / Mutual Funds | ~1,500 | 20–50 | ~45,000 |
| Investment Banks (front office) | ~100 | 100–300 | ~15,000 |
| Pension Funds & Endowments | ~1,000 | 5–10 | ~7,000 |
| Prop Trading Firms | ~500 | 10–30 | ~7,500 |
| Insurance Investment Arms | ~300 | 5–10 | ~2,000 |
U.S. Inst. Traders
~111K
Avg Spend / Mo
$2.5K
Annual Market
$3.3B
Range: $500–$5,000+/mo per user, depending on role and toolset. (A Bloomberg Terminal alone is ~$2,500/mo.)
Who's building what
AI Trading Features from Incumbents
While some AI/LLM features have been integrated by incumbents like Robinhood and Moomoo through acquisition or development, truly distinct product-level LLM functionalities that significantly enhance the user experience have yet to materialize.
| Target Users | Company | Strengths & Weaknesses |
|---|---|---|
| Retail retail Amateur Traders |
Robinhood Pluto (acquired by Robinhood) |
Enhanced data analysis: Pluto's analytics process market data efficiently by giving state-of-the-art LLMs access to real-time personal and global financial data. Personalized strategies: Algorithms tailor recommendations to individual profiles based on risk tolerance, goals, and historical behavior. |
| Amateur Traders Retail retail |
Moomoo | Moomoo has blended traditional ML statistical algorithms into its products over the past years. Features include AI Monitor (real-time alerts on unusual trading activity in watchlist stocks), Trend Projection (historical pattern analysis for price insights), Sentiment Analysis (multi-source signal aggregation), and Quantum AI (an AI bot that analyzes market data and executes trades). |
Pre-LLM AI Products in Retail Investment
Before LLMs, most AI products in this domain were limited by rigid pattern recognition — yielding shallow analysis and less accurate insights. They lacked the contextual intelligence and human-like interaction that LLMs now enable.
1 · Trend & Pattern Recognition Platforms
- Trade Ideas — One of the most popular AI-powered scanners; their engine "Holly" used reinforcement learning and statistical models to suggest trades.
- Tickeron — Used neural networks to identify technical patterns and rank trading opportunities by confidence score.
2 · Sentiment & News Analysis Tools
- StockTwits sentiment indicators — Tracked user sentiment trends on individual stocks.
- Accern (early retail version) — Provided filtered AI-based sentiment feeds; later refocused on institutional use.
3 · Algorithmic Trading Bots (Rule-Based / ML)
- TradingView with Pine Script — Allowed retail users to build and backtest strategies; AI was in user-created indicators.
- MetaTrader 4/5 with Expert Advisors — Rule-based bots, with some using ML for parameter tuning.
After-LLM Products in Retail Investment
| Target Users | Company | Description |
|---|---|---|
| Amateur Traders Institutions |
Scalar Field (YC X25) | AI-enabled research and analysis tool for experienced traders. |
| Retail retail Amateur Traders |
Intellectia | LLM-powered investment research platform designed to democratize access to sophisticated financial analysis tools for investors of all levels. |
| Retail retail | Rockflow | Singapore-based, serving the global market (mostly U.S. & HK). AI-driven trading application designed to simplify the investment experience. Targets Gen Z investors and cross-border investment. |
Two builders, two bets
Intellectia
Key functions
- Stock Picker — automatic AI picks; baskets of stocks for bullish or bearish markets
- Saving Trades
- Day trading
- Earnings trading
- Pattern detection
Targeted users
Both "retail retail" and individual amateur traders with trading knowledge — especially technical, swing, and day traders.
Competitiveness
- In-depth, well-rounded analysis covering technical, fundamental, event-driven, and statistical methods.
- User-friendly, hands-off recommendations — including swing-trading entry points.
- For example, it picks the top 5 stocks daily with rationales and backtests results the next day. Caveat: aligned user behavior could negatively affect market outcomes.
Concerns & risks
- No strong defensibility. Analysis is easily copied by incumbents or competitors. In side-by-side comparison, relevance and accuracy were worse than ChatGPT.
- Data and analysis accuracy. Analysis isn't deep enough — e.g., when prompted for Google's volatility structure (in the context of buy/put options), the analysis pulled from a third-party site not recognized in the industry. Some figures, such as Apple's before/after earnings closing prices, were incorrect.
Scalar Field
Target users
Individual amateur traders with trading knowledge, plus institutional investors.
Core functions
- Fundamental analysis
- Backtesting
- Earnings calendar and event analysis
- Insider trading
Competitiveness
- Starts with a reliable data provider (FactSet). Uses established industry analysis methods and makes them accessible — crucial for retail investors to understand institutional thinking.
- Excels in reasoning and plotting. Offers assumption verification before finalization, presents backtest results, and emulates seasoned-trader decision-making drawn from the founding team's expertise.
Concerns & risks
- Passive and inadequate source citation. The agent doesn't proactively provide source links — only after manual follow-up. A professional tool needs clear, one-click traceability to source documents.
- Licensing risks. Hard to obtain the necessary licenses — especially a broker-dealer license — given regulatory scrutiny, financial requirements, and compliance complexity.
What could go wrong
- Regulatory risk. Investment research, strategy, and trading require specific licenses. A CFA charter or FINRA Series 86/87 may be required to publish investment research; a Series 7 license may be required to execute client trades; and registration as an RIA (or with the CFTC/NFA) may be needed for managing assets or offering trading strategies — especially when derivatives or leverage are involved.
- Hallucinations. LLMs may generate non-existent ticker symbols, fake metrics, or misrepresent facts — damaging credibility and exposing platforms to liability.
- Security and privacy. Integrating with users' portfolios or brokerage accounts raises data-security concerns and requires specific licenses. Mishandling sensitive financial information can lead to trust loss or breaches.