Somewhere right now, a trader is staring at an ad that promises a “revolutionary AI trading bot” capable of generating “consistent 90%+ returns on autopilot.” The ad looks slick. The testimonials are glowing. And the pitch is irresistible—just plug in your account, let the AI do its thing, and watch the money roll in.
Our team has been reviewing trading tools for years, and we can tell you this: that ad is almost certainly lying to you. Not about everything—but about the part that matters most.
Here’s the uncomfortable reality about AI trading bots. The technology is real. Machine learning can genuinely improve certain aspects of trading. But the vast majority of products marketed as “AI-powered” to retail traders are about as intelligent as a kitchen timer with a fancy label. The U.S. Commodity Futures Trading Commission put it bluntly in their official customer advisory: “AI technology can’t predict the future or sudden market changes.”
That’s not our opinion. That’s the federal regulator responsible for overseeing commodity and futures markets telling you, in plain English, that the marketing doesn’t match reality.
So how do you separate the genuine AI trading tools from the hype machines and outright scams? That’s exactly what this article is for. We’ve built a framework—what we call the 4-Level AI Framework—that lets you classify any “AI” trading bot you encounter and determine what you’re actually getting. We’ll show you what real AI trading looks like, walk through the SEC’s recent crackdown on “AI washing,” and give you a practical checklist for evaluating any tool before you hand over your money.
No hype. No sales pitch. Just the truth.
The Great AI Trading Bot Marketing Machine
The word “AI” has become the most overused term in trading software marketing. And frankly, it’s not hard to see why. Slap “AI-powered” on any product and it immediately sounds cutting-edge, sophisticated, and trustworthy. The problem? There’s no regulatory standard defining what qualifies as “AI” in a trading product. Any company can call their basic if/then alert system “artificial intelligence” and face zero consequences—unless, as we’ll cover shortly, regulators catch them making specific false claims.
The marketing playbook is predictable. Terms like “machine learning,” “neural networks,” “deep learning,” and “proprietary algorithm” get thrown around in product descriptions with the same casual abandon as “artisanal” on a $12 loaf of bread. They sound impressive. They create an aura of technological superiority. And most retail traders don’t have the technical background to question whether these terms are being used accurately.
The result is a market flooded with products that are, at best, basic rules-based trading systems dressed up in AI marketing—and at worst, outright scams designed to steal your money.
The CFTC’s customer advisory is remarkably direct about this. The agency warns that scammers are “exploiting public interest in artificial intelligence to tout automated trading algorithms, trade signal strategies, and crypto-asset trading schemes that promise unreasonably high or guaranteed returns.” They go on to note that these fraudulent products claim AI algorithms can generate “tens of thousands of percent” returns or “100 percent win rates.”
Quick reality check: if anyone had a system capable of 100% win rates, they wouldn’t be selling it to you for $49.99 a month.
Understanding what algorithmic trading actually involves—the infrastructure, the data requirements, the ongoing maintenance—is the first step in seeing through the marketing fog. But to truly evaluate any product’s AI claims, you need a structured framework. Which brings us to the tool we built for exactly this purpose.
Our 4-Level Framework for Evaluating “AI” Trading Bots
After spending years testing and reviewing trading tools, our team developed a simple classification system that cuts through the noise. We call it the 4-Level AI Framework, and it works for any product—stocks, crypto, forex, futures—that claims to use artificial intelligence.
Think of it like this. Imagine someone tries to sell you a “self-driving car.” Level 1 would be a car with cruise control. Level 2 adds lane assist. Level 3 is a Tesla with Autopilot—genuinely impressive, but still requiring human oversight. Level 4 is a fully autonomous vehicle that doesn’t need a driver at all. Each level is a real technology. But marketing them all as “self-driving” would be wildly misleading.
The same thing is happening in trading.
| Level | Classification | What It Actually Does | Retail Examples |
|---|---|---|---|
| 1 | Rules-Based | Executes pre-programmed if/then logic. “If RSI drops below 30, buy.” No learning, no adaptation. | Most marketed “AI bots” |
| 2 | Indicator-Based | Combines multiple technical indicator triggers with conditional logic. Slightly more complex rules, but still static. | “Smart” scanner alerts, basic signal services |
| 3 | Machine Learning | Learns from historical data, optimizes parameters, adapts strategies based on performance. Requires significant computational resources. | Trade Ideas Holly AI, some TrendSpider features |
| 4 | Advanced AI (Deep Learning/LLMs) | Neural networks, natural language processing, sentiment analysis, multi-agent reasoning systems. | Primarily institutional; emerging in retail at basic levels |
Here’s the kicker: the overwhelming majority of retail “AI trading bots” are Level 1 or Level 2. They’re rules engines with a marketing department. The “AI” label is cosmetic.
The critical difference between Level 1-2 and Level 3-4 is whether the system learns and adapts. A Level 1 bot executes the same rules forever, regardless of whether market conditions have changed. A Level 3 system—like Trade Ideas’ Holly—reruns its entire strategy library through backtesting every single night and adjusts which strategies it deploys based on current market conditions. That’s a fundamentally different technology.
Does that mean Level 1-2 tools are useless? Not necessarily. A well-designed rules-based scanner can absolutely help you find trade candidates faster. But you should know what you’re buying. If someone is charging you $200 a month for a Level 1 product while marketing it as “cutting-edge AI,” you’re overpaying for the technology and being misled about what it does.
For a deeper look at the different categories of automated trading systems—from grid bots and DCA bots to trend followers and arbitrage systems—our guide to trading bot types breaks down how each one works mechanically.
What Genuine AI Trading Actually Looks Like
Alright, so most “AI” bots aren’t really AI. But genuine AI trading tools do exist for retail traders. Understanding what they look like—and what they can realistically do—helps you calibrate your expectations.
Trade Ideas Holly AI is probably the closest thing to genuine machine learning available to retail day traders. Here’s how it actually works, stripped of the marketing language:
Every night after the market closes, Holly runs over 60 different trading strategies through a backtesting engine using machine learning optimization. The system evaluates each strategy’s performance against recent market data, then selects and deploys only the strategies that meet strict criteria—specifically, a 60%+ historical win rate with a minimum 2:1 risk-reward ratio. Holly comes in three variants (Holly Grail, Holly 2.0, and Holly Neo), each targeting different risk profiles and trade types.
This is legitimately impressive technology. It’s adaptive—the strategy selection changes based on current market conditions. It’s data-driven—decisions are based on quantitative backtesting, not gut feelings. And it’s been around since 2016, which gives it a meaningful track record.
But—and this matters—it’s not magic. Independent testing by StockBrokers.com notes that some of Holly’s displayed results appear cherry-picked. And a 60% win rate, while solid, means 4 out of every 10 Holly trades lose money. For a comprehensive breakdown of the platform, pricing (currently ~$254/month for Premium or ~$178/month billed annually), and our team’s hands-on experience, read our full Trade Ideas review.
TrendSpider represents another legitimate application of AI in retail trading. Its AI Strategy Lab uses pattern recognition algorithms and machine learning to identify chart patterns and optimize strategy parameters automatically. It’s not generating trade signals from thin air—it’s enhancing technical analysis that traders are already doing. Our TrendSpider review covers its AI features in detail.
What the academic research says: A 2024 study published in the Journal of Financial Economics by Cao, Jiang, Wang, and Yang found something fascinating. AI did outperform human analysts in raw prediction accuracy for stock analysis. But here’s the real takeaway—the hybrid “Man + Machine” approach, where human analysts used AI as a tool rather than a replacement, outperformed both humans alone and AI alone. That finding aligns perfectly with our team’s philosophy: AI is a co-pilot, not the pilot.
Where AI actually adds value (the unsexy truth): Professional quant traders—the ones who actually make money with AI—don’t primarily use it for predicting price direction. They use it for operational advantages: dynamic position sizing based on volatility regimes, execution optimization to reduce slippage, regime detection to know when to turn strategies on or off, and portfolio heat distribution across sectors. These applications compound small advantages over thousands of trades. It’s not glamorous. But it works.
The GAO confirmed this in their May 2025 report on AI in financial services (GAO-25-107197), noting that financial institutions use AI across automated trading, risk management, credit decisions, and customer service—but the report also flagged significant risks around bias, hallucinations, data quality, and conflicts of interest. The technology is powerful. It’s also imperfect.
The SEC Is Cracking Down: Real AI Washing Enforcement Cases
If you needed proof that “AI washing” is a real and serious problem, look no further than the SEC’s enforcement division. The regulator has made it clear that companies making false claims about their AI capabilities will face consequences—and they’ve started backing that up with action.
March 18, 2024: The First-Ever AI Washing Cases
The SEC announced settled charges against two investment advisers in what became the first enforcement actions specifically targeting AI-related misrepresentations.
Delphia (USA) Inc. was a Toronto-based robo-adviser that told its clients—in SEC filings, press releases, and on its website—that it used a “predictive algorithmic model” powered by client data. Delphia claimed it “puts collective data to work to make our artificial intelligence smarter so it can predict which companies and trends are about to make it big.” The problem? During a 2021 SEC examination, Delphia admitted it had never actually built the algorithm. It never used client data in any AI model. The whole thing was marketing fiction. Even after admitting this to regulators, Delphia continued making similar false claims for nearly two more years. Penalty: $225,000.
Global Predictions Inc., a San Francisco-based firm, falsely claimed to be the “first regulated AI financial advisor” and marketed “expert AI-driven forecasts” on its website and social media. None of it was true. Penalty: $175,000.
SEC Chair Gary Gensler’s response was unambiguous: “Investment advisers should not mislead the public by saying they are using an AI model when they are not. Such AI washing hurts investors.”
SEC Enforcement Director Gurbir Grewal went further, identifying five distinct categories of AI-related risk: AI washing, using AI to facilitate fraud or market manipulation, conflicts of interest from AI prioritizing firm interests over clients, hallucinations in AI-generated investment materials, and systemic risk from too many firms using the same AI models.
The message to the industry couldn’t be clearer. And it’s backed by the SEC’s creation of the Cybersecurity and Emerging Technologies Unit (CETU) in February 2025, with AI washing designated as an “immediate priority.”
In January 2024, the SEC, FINRA, and NASAA also issued a joint investor alert specifically about AI and investment fraud, warning that “bad actors are using the growing popularity and complexity of AI to lure victims into scams.” That joint alert—from three separate regulatory bodies—should tell you everything about how seriously regulators view this problem.
For retail traders, the lesson is straightforward: if a company’s AI claims sound too good to be true, they probably are. And now there’s regulatory precedent showing that fraudulent AI claims carry real penalties.
7 Red Flags That Expose Fake “AI” Trading Bots
Our team has reviewed dozens of trading tools over the years, and certain patterns emerge with remarkable consistency among the fraudulent or misleading ones. Here are the warning signs that should immediately trigger your skepticism.
1. “Set It and Forget It” Promises
Real AI trading requires constant monitoring, parameter adjustment, and human oversight. Even Trade Ideas Holly—one of the most sophisticated retail AI tools available—resets and re-optimizes its strategies nightly. Any product promising you can activate a bot and walk away is either lying about its capabilities or setting you up for significant losses. The GAO’s 2025 report specifically flagged the risk of “AI agents acting autonomously without human validation and approval.”
2. Guaranteed Returns or Impossibly High Win Rates
The CFTC warns specifically about products claiming “100 percent win rates” or returns of “tens of thousands of percent.” No legitimate trading system—AI or otherwise—can guarantee returns. Markets are inherently unpredictable. If someone promises you consistent 10%+ monthly returns with no risk, they’re running a scam.
3. “Secret Algorithm” or No Methodology Explanation
Legitimate AI companies can explain—at least at a high level—how their technology works. Trade Ideas publishes information about Holly’s nightly backtesting process. TrendSpider explains its pattern recognition methodology. If a company hides behind “proprietary secret algorithms” and refuses to explain even the basic approach, that’s a Level 1 bot wearing a Level 4 costume.
4. Deepfake Celebrity Endorsements
This is the 2024-2025 evolution of trading scams. Fraudsters now use AI-generated deepfake videos of public figures like tech CEOs and investment legends to endorse fake trading platforms. The joint SEC/FINRA/NASAA investor alert specifically warns about AI-generated videos used to “spread false or misleading information.” If a trading bot’s marketing features a celebrity endorsement—especially one shared via social media—verify it independently before believing a word of it.
5. Anonymous or Unverifiable Team
Who built this bot? Can you find them on LinkedIn? Is the company registered with any regulatory body? Legitimate companies have real people behind them with verifiable track records. Trade Ideas was founded in 2003. Its team is publicly identified. If you can’t find a single real human behind an “AI trading bot,” that’s your answer.
6. High-Pressure Tactics and Artificial Urgency
“Only 50 spots left!” “Price doubles at midnight!” “Join now or miss out forever!” These are the hallmarks of scams, not legitimate technology companies. Cognitive biases—particularly fear of missing out and urgency bias—make these tactics devastatingly effective. Understanding how cognitive biases affect your trading decisions is one of the best defenses against manipulation.
7. You Can Deposit but Can’t Withdraw
This is the ultimate red flag, and by the time you encounter it, it’s usually too late. Fraudulent platforms make deposits seamless but create endless obstacles—”verification requirements,” “tax fees,” “processing delays”—when you try to withdraw. If you encounter withdrawal problems with any trading platform, contact the CFTC, SEC, or your state securities regulator immediately.
The hidden costs of automated trading are real even with legitimate tools—subscription fees, data costs, slippage, and execution gaps all eat into performance. But with fraudulent bots, the cost isn’t just fees. It’s your entire deposit.
The 5-Question Due Diligence Framework for Any AI Trading Bot
Red flags tell you when to run. But what about products that seem reasonable? Maybe the marketing isn’t outrageous. The price seems fair. The website looks professional. How do you dig deeper?
Our team uses these five questions to evaluate any AI trading tool. They’re designed to be practical—you don’t need a computer science degree to use them.
Question 1: “What level of AI is this, really?”
Apply our 4-Level Framework. Read the product’s documentation carefully. Does it describe actual machine learning processes (data training, model optimization, adaptive parameters)? Or does it describe rules and indicators dressed up in AI language? If the documentation says things like “our AI monitors RSI and MACD crossovers,” that’s a Level 2 indicator-based system—useful, but not AI in any meaningful sense.
Question 2: “Can they explain HOW the AI makes decisions?”
Ask directly. Email support. Read the FAQ. A legitimate AI company will explain—even in simplified terms—what data the model uses, how it’s trained, and what criteria it uses to generate signals. “Proprietary” is fine; “we can’t tell you anything” is a problem. The FINRA 2026 Regulatory Oversight Report specifically flags “auditability and transparency” as a key AI risk, noting that “complicated, multi-step agent reasoning tasks can make outcomes difficult to trace or explain.”
Question 3: “Is there independent, verifiable performance data?”
Company-published results can be cherry-picked. Look for independent reviews, third-party backtesting, or verified track records on platforms like Collective2 or Myfxbook. Any legitimate tool should encourage you to backtest its strategies yourself using historical data before risking real capital.
Question 4: “Is this company registered and regulated?”
This is non-negotiable. Use FINRA BrokerCheck (brokercheck.finra.org) and the SEC’s EDGAR database (sec.gov/edgar) to verify registration. Check Investor.gov’s free search tool. If a company claiming to manage or advise on investments isn’t registered with appropriate regulatory bodies, treat it as a major red flag. The joint SEC/FINRA/NASAA alert explicitly recommends this step.
Question 5: “What happens when the AI is wrong?”
Every system loses. What matters is how the system—and the company—handles losses. Does the bot have built-in risk management? Maximum drawdown limits? Position sizing controls? Can you set your own stop-losses? If the product doesn’t address what happens when trades go wrong, it’s either poorly designed or intentionally hiding its downside.
And one more thing—regardless of how good any AI tool looks on paper, always paper trade it first. Practice accounts exist for exactly this reason. If you’re interested in building your own automated systems, our trading bot blueprint walks through the entire process step-by-step.
Frequently Asked Questions About AI Trading Bots
Do AI trading bots actually work?
Quick Answer: Some do, within realistic expectations. Genuine machine learning tools like Trade Ideas Holly can provide a statistical edge, but no AI bot delivers guaranteed or effortless profits.
The answer depends entirely on what “work” means. If you expect a bot to generate passive income while you sleep, no—nothing works like that. If you mean a tool that can process more data than a human, identify patterns faster, and remove some emotional decision-making from your trading, then yes—legitimate Level 3+ tools can add value. The key finding from academic research (Cao et al., 2024) is that AI combined with human judgment outperforms either alone. AI works best as a tool that enhances your existing trading skill, not as a replacement for it.
Key Takeaway: AI bots work as tools, not money machines. Use them to augment your strategy, not replace your judgment.
Are AI trading bots a scam?
Quick Answer: Some are outright scams. Others are legitimate tools with overhyped marketing. The 4-Level Framework helps you distinguish between them.
The SEC’s AI washing enforcement actions prove that fraud exists in this space. Delphia and Global Predictions were literally charged with lying about their AI capabilities. But products like Trade Ideas Holly and TrendSpider are legitimate, well-established platforms backed by real technology. The challenge for retail traders is that scams and legitimate tools can look remarkably similar from the outside—which is why the 5-Question Due Diligence Framework exists.
Key Takeaway: Not all AI bots are scams, but many are overhyped. Always verify claims using our 5-Question Framework before investing.
What is the best AI trading bot in 2026?
Quick Answer: There’s no universal “best.” Trade Ideas Holly is the most established Level 3 ML system for retail day traders, but the right tool depends on your strategy, market, and budget.
Our team considers Trade Ideas Holly the gold standard for genuine machine learning in retail day trading—it’s been operating since 2016, uses real nightly backtesting across 60+ strategies, and has the longest track record in the space. But “best” is subjective. A crypto trader needs different tools than an equity day trader. Check our full Trade Ideas review for a detailed breakdown, including pricing and honest limitations.
Key Takeaway: Focus less on finding “the best” bot and more on finding a legitimate tool that fits your specific trading style and budget.
Can AI trading bots make you money?
Quick Answer: They can contribute to a profitable trading operation, but they cannot generate profits independently. The trader’s skill, risk management, and expectations matter more than the tool itself.
Here’s an analogy: a high-end set of kitchen knives can help a skilled chef cook better. But they won’t turn someone who’s never cooked into Gordon Ramsay. AI trading tools work the same way. They amplify what you already have—which means if your risk management is poor and your strategy is untested, AI will just help you lose money faster and more efficiently.
Key Takeaway: AI bots can contribute to profitability only when paired with solid trading fundamentals and realistic expectations.
What is “AI washing” in trading?
Quick Answer: AI washing is the practice of making false, misleading, or exaggerated claims about a product’s use of artificial intelligence. The SEC coined the term and has begun enforcing against it.
The term was popularized by former SEC Chair Gary Gensler, who compared it to “greenwashing”—companies exaggerating environmental claims. In trading, AI washing ranges from mild exaggeration (calling a rules-based system “AI-powered”) to outright fraud (claiming to use machine learning that doesn’t exist). The SEC’s March 2024 enforcement actions against Delphia and Global Predictions established clear legal precedent that AI washing violates securities laws.
Key Takeaway: AI washing is a recognized form of securities fraud. If a tool’s AI claims seem vague or unverifiable, skepticism is warranted.
What does the SEC say about AI trading bots?
Quick Answer: The SEC, along with FINRA and NASAA, has issued formal warnings about AI-related investment fraud and has brought enforcement actions against companies making false AI claims.
The regulatory stance is multi-pronged. In January 2024, the SEC, FINRA, and NASAA issued a joint investor alert warning about AI-fueled investment scams. In March 2024, the SEC brought its first AI washing enforcement actions. The CFTC issued its own customer advisory stating explicitly that “AI technology can’t predict the future or sudden market changes.” The message from regulators is consistent: AI is a legitimate technology, but false claims about AI capabilities are fraud.
Key Takeaway: Regulators are actively targeting AI-related fraud. Use their free tools (Investor.gov, FINRA BrokerCheck) to verify any company’s claims.
How do you know if an AI trading bot is legitimate?
Quick Answer: Apply the 5-Question Due Diligence Framework: classify its AI level, demand methodology transparency, seek independent performance data, verify regulatory registration, and assess risk management protocols.
Beyond our framework, look for longevity and reputation. Trade Ideas has been operating since 2003. TrendSpider has years of public reviews. Legitimate tools have independent reviews from trusted sources like StockBrokers.com, and they don’t rely on social media testimonials from anonymous accounts. If a product appeared last week with no verifiable history and promises extraordinary returns, treat it with extreme skepticism.
Key Takeaway: Legitimacy leaves a trail—registration records, independent reviews, transparent methodology, and a real team. Scams don’t.
Is Trade Ideas Holly a real AI?
Quick Answer: Yes. Holly is a Level 3 machine learning system that runs nightly backtests across 60+ strategies and deploys only those meeting strict performance criteria. It’s one of the few retail tools that qualifies as genuine machine learning.
Holly uses machine learning optimization to evaluate and select strategies each night, deploying only those with 60%+ historical win rates and minimum 2:1 risk-reward ratios. The three variants—Holly Grail, Holly 2.0, and Holly Neo—target different risk profiles. This is fundamentally different from a static rules-based system. That said, Holly isn’t infallible. It loses roughly 4 out of every 10 trades, and results can vary significantly based on market conditions. For the full breakdown of how Holly works in practice, see our in-depth Trade Ideas review.
Key Takeaway: Holly is genuine machine learning—not marketing hype—but it’s still a tool with real limitations, not a money printer.
Why do most AI trading bots fail?
Quick Answer: Most fail because they’re not actually AI (Level 1-2 rules-based), and even legitimate AI systems struggle with overfitting, data quality issues, and the fundamental non-stationarity of financial markets.
Markets change constantly. A strategy that worked perfectly last quarter may fail this quarter because the underlying market dynamics shifted. This is called non-stationarity, and it’s the core reason why backtests lie—a model trained on historical data is essentially studying for last year’s exam. A 2025 study published in Springer’s International Journal of Data Science and Analytics tested ML algorithms across 55 markets and found that accuracy varied dramatically based on market efficiency. Walk-forward testing reveals that most models have zero predictive power out-of-sample. Add in data quality issues (the GAO’s 2025 report flagged “incomplete, erroneous, unsuitable, or outdated data” as a key AI risk), and you start to see why even well-designed systems frequently underperform their backtested results.
Key Takeaway: The “failure” often isn’t the technology—it’s unrealistic expectations. Markets are hard, and AI doesn’t change that fundamental truth.
Can AI predict the stock market?
Quick Answer: Not reliably. The CFTC states explicitly that “AI technology can’t predict the future or sudden market changes.” AI can identify patterns and probabilities, but consistent, accurate market prediction remains beyond current technology.
If AI could reliably predict market direction, the firms with the best AI—Renaissance Technologies, Citadel, Two Sigma—would never have losing periods. They do. These firms have armies of PhDs, decades of proprietary data, and billions in infrastructure, and they still can’t predict markets consistently. Your retail AI bot, working off publicly available price data, is operating at a profound competitive disadvantage. The real value of AI in trading isn’t prediction—it’s processing speed, pattern identification at scale, and removing emotional bias from specific decisions.
Key Takeaway: AI can improve your analysis. It cannot predict the future. Anyone claiming otherwise is either misinformed or trying to sell you something.
Disclaimer
The information provided in this article is for educational purposes only and should not be considered financial advice. Day trading involves substantial risk and is not suitable for every investor. Past performance is not indicative of future results.
For our complete disclaimer, please visit: https://daytradingtoolkit.com/disclaimer/
Article Sources
This article’s claims and data are supported by the following authoritative sources. We encourage you to review them directly for additional context.
- SEC Press Release 2024-36: SEC Charges Two Investment Advisers with Making False and Misleading Statements About Their Use of Artificial Intelligence — Primary source for the Delphia and Global Predictions AI washing enforcement cases, including penalties and SEC Chair Gensler’s statement.
- CFTC Customer Advisory: AI Won’t Turn Trading Bots into Money Machines — The Commodity Futures Trading Commission’s official warning about AI-related trading scams and the limitations of AI in trading.
- SEC, FINRA, & NASAA Joint Investor Alert: Artificial Intelligence and Investment Fraud — Joint alert from three regulatory bodies warning investors about AI-fueled fraud, deepfake technology, and verification steps.
- GAO Report GAO-25-107197: Artificial Intelligence — Use and Oversight in Financial Services (May 2025) — The Government Accountability Office’s comprehensive analysis of AI use, risks, and regulatory oversight gaps in the financial services industry.
- FINRA 2026 Annual Regulatory Oversight Report: GenAI Continuing and Emerging Trends — FINRA’s most recent regulatory guidance on AI risks including agent autonomy, auditability, and data sensitivity concerns.
- Cao, Jiang, Wang & Yang (2024): “From Man vs. Machine to Man + Machine: The Art and AI of Stock Analyses” — Journal of Financial Economics — Peer-reviewed research finding that AI outperforms human analysts but the hybrid “Man + Machine” approach outperforms both.
