TradeSmith is one of the more serious fintech offerings in the direct-to-investor space, and this tradesmith review starts with the reason: the person running it is a software engineer, not a stock picker. Keith Kaplan built platforms before he built predictions. That distinction matters when you’re evaluating a product that claims to forecast stock prices 21 days ahead.

Who Is Keith Kaplan and What Is TradeSmith?

Keith Kaplan is a software architect who moonlights as a CEO. His background — computer information systems from Towson University, cloud engineering at Stansberry Research, then president and CEO of TradeSmith — reads like a DevOps resume, not a Wall Street one. That matters.

TradeSmith is a financial technology company, not a newsletter publisher. It started in 2005 with TradeStops, a portfolio-tracking tool that uses trailing stops and a proprietary volatility measurement called the Volatility Quotient (VQ). Over two decades, it grew into a suite of tools: TradeStops, Options360, Trade Cycles, Ideas by TradeSmith, and now Predictive Alpha, powered by the An-E AI engine.

TradeSmith is a wholly owned subsidiary of MarketWise (Nasdaq: MKTW), the publicly traded financial publishing company behind 25+ brands. That fact matters for trust. MarketWise reported 2.4 million active subscribers as of Q1 2026, with 381,000 paid subscribers and billings up 15% year-over-year. That public-company reporting matters. MarketWise files SEC disclosures. It’s not a random LLC in a strip mall.

The numbers on TradeSmith’s own platform: 134,000+ users tracking $29 billion in assets. Those are real people managing real money with these tools.

What Do You Actually Get?

TradeSmith is a platform ecosystem, not a single product. Here’s the breakdown:

TradeStops — The original. Portfolio risk management using VQ-based trailing stops. It tells you when to get out, not when to get in. Subscribers upload their portfolios and receive daily alerts when stop thresholds are triggered.

Options360 — Options screeners that surface trades by probability of profit and ROI. Built on the same VQ and health indicator algorithms.

Predictive Alpha — The new flagship. This is where An-E (Analytical Engine) lives. You enter a stock ticker and An-E returns a price forecast 21 trading days out — direction, magnitude, and a confidence score, the 21-day forecasting window dimensionalized in detail in the An-E AI 21-day forecasts breakdown. Subscribers get one stock search per week plus two analyst-vetted stock recommendations per month. For a deeper look at the flagship product itself, see the Predictive Alpha Review: Keith Kaplan’s AI Tool.

Predictive Alpha Prime — The premium tier. Full access to An-E’s forecasts across thousands of stocks, plus daily updates, educational materials, and analyst-vetted trade ideas. Normally $5,000/year, currently discounted to $1,799.

The An-E AI: What the Numbers Actually Say

An-E is a time-series forecasting model, not a general-purpose LLM. TradeSmith trained it on 1.3 quadrillion data points across 50,000+ backtests. The team behind it: 69 engineers, 22 machine learning experts, and 23 quantitative analysts. That’s a real R&D operation — $18 million and 50,000 man-hours of development.

The claims are specific. An-E forecasts 2,300+ stocks 21 trading days ahead. The stated directional accuracy is roughly 57.5%, with target accuracy around 60%. Those numbers are modest enough to be believable. No AI predicts the market with 90% accuracy. Anyone claiming otherwise is selling something.

Let me put those numbers in context. Academic literature on stock prediction is sobering. The FINSABER framework (2025) tested LLM-based investing strategies across two decades and 100+ symbols and found that reported advantages “deteriorate significantly under broader cross-section and over a longer-term evaluation.” Another study from the Journal of Computational Economics (2025) found that even sophisticated models rarely achieve prediction accuracy above 80% in stock forecasting. A 57.5% directional accuracy — betting on which way a stock will move — is meaningful if consistent. It’s not a crystal ball. TradeSmith itself says that.

I checked the public demo examples. An-E projected OXY would rise from $46.21 to $49.23 in 21 trading days with 70% confidence. It hit $49.19 — a 6.44% gain in 20 days. It projected Light & Wonder (LNW) would drop 13.20% from $106.60. It dropped 13.74%. The DBX forecast called for a 6.19% gain and the stock delivered 6.19% — within a rounding error.

These are cherry-picked, obviously. TradeSmith shows the hits, not the misses. But the misses are visible in the product itself — every forecast shows a confidence score. Some are low. The model doesn’t pretend to be confident on everything.

What to Consider Before You Subscribe

TradeSmith is a legitimate company with real products, a real team, and a publicly traded parent. The question is whether the tools fit your investing style.

First, these are analysis tools, not automated trading systems. You still make the decisions. Predictive Alpha gives you a forecast and a confidence score. You interpret it, you execute it, you manage the risk. If you want a robo-advisor that manages your money, this is not it.

Second, the pricing ladder is wide. Predictive Alpha starts at $49 for a year — a low-risk entry point. Predictive Alpha Prime costs $1,799. That’s a substantial investment. The 60-day and 90-day refund periods give you room to test, but you need to actually use the tools during that window to know if they work for you.

Third, AI stock forecasting has known limitations. The academic research I cited shows that short-horizon predictions look better than they actually are when you expand the test universe and time frame. An-E’s 57.5% accuracy is tested against 2,300+ stocks over backtests, but real-time performance across different market regimes — bull, bear, sideways — remains the truer test. The model launched publicly in 2024. It hasn’t been through a full bear cycle yet.

Fourth, TradeSmith’s ecosystem has a learning curve. TradeStops, Options360, and Predictive Alpha each require time to understand. The Volatility Quotient, Stock State Indicators, confidence gauges — these are real concepts with real statistical backing, but they’re not microwave-simple. You get out what you put in.

The Structural Read

TradeSmith is one of the more serious fintech offerings in the direct-to-investor space. The team is genuine. The R&D investment is real. The public-company accountability is a genuine advantage — MarketWise reports earnings, files 10-Ks, and has a fiduciary structure that private competitors don’t.

The An-E AI is impressive as a technical achievement, but its real-world value depends on how you use it. A 57.5% directional edge compounded over many trades is real alpha. But it’s an edge, not a guarantee. The confidence scores exist for a reason. Pay attention to them.

The best argument for TradeSmith is that it gives retail investors access to institutional-grade quantitative tools. The hedge fund world has been doing this kind of systematic analysis for decades. TradeSmith packages it for the individual investor at a fraction of the cost.

The structural limitation: no AI has repealed the laws of probability. Stock forecasting is hard. The academic literature is clear that even the best models struggle with regime changes, black swans, and the simple fact that markets are noisy systems. TradeSmith’s own disclaimer — “not a crystal ball” — is the most candid framing in their marketing.

If you’re a self-directed investor who wants data-driven signals and you’re willing to put in the time to learn the tools, TradeSmith is worth a look. If you’re looking for a magic button that prints money, keep looking. It doesn’t exist. For a wider view of the newsletter and service landscape, see the full Newsletter Reviews index.