↗
One-year horizon
Monthly picks are evaluated against one-year total return, matching a long-term investor's decision cycle.
StockPickAI
Long-term equity intelligence
Built for investors, not day traders
Explainable AI for better long-term equity decisions.
StockPickAI ranks public equities using fundamentals, market history, and transparent model analysis. Monthly baskets are evaluated against one-year total return, matching a long-term investor's decision cycle.
5
model-ranked stock picks each month
5,400+
US-listed stocks in historical research universe
130+
fundamental, price, and financial features
Method
Decisions that match how long-term investors actually think.
The model is trained for one-year forward return, not the next 30 seconds. Every pick comes with feature attribution so you can stress-test the thesis before you put real capital behind it.
◇
Transparent signals
Feature analysis highlights the drivers behind each pick: price momentum, earnings ratios, free cash flow, volatility, capital structure.
◷
Research-grade workflow
Rolling train, validation, and test windows reduce hindsight bias while keeping models aligned to changing market regimes.
How it works
From a 5,400-stock universe to five high-conviction picks in three steps.
STEP 01
Score every eligible US-listed stock
Each month, every stock in the research universe is scored on 130+ fundamental, momentum, and quality features. Scoring uses time-series cross-validation so a model trained on 2017 data never peeks at 2019 to evaluate 2018.
STEP 02
Rank by one-year outperformance probability
The trained XGBoost ensemble assigns each stock a calibrated probability of beating the market over the next 12 months. The top picks become the model portfolio for the period.
STEP 03
Hold, review, and explain
Picks are held for 12 months. The dashboard surfaces feature attribution, regime context, and drift signals so you can decide whether to follow the model or override it for a position.
Platform
An end-to-end research workspace, not just a feed of picks.
Model portfolios are the core, but the platform also includes the surrounding research tooling an analyst would otherwise stitch together by hand.
Model portfolios + monthly picks
Curated monthly baskets of five high-conviction stocks, ranked by calibrated probability. Every pick comes with feature attribution and a regime tag so you can see WHY the model surfaced it.
- One-year holding period, monthly rebalance
- Feature drivers visible per stock
- Walk-forward backtests vs. SPY benchmark
- Regime tags (bull / bear / high-vol / recession)
Sept 2026
Top 5
Bull
NVDA0.91
MSFT0.88
BRK.B0.84
AAPL0.82
COST0.79
Market reports + regime analysis
Regularly-generated reports on market regime, momentum continuation, and 52-week breakouts. Tickers in every report are clickable — drill into the full model view without leaving the analysis.
- Momentum continuation picks
- 52-week high / low breakout reports
- Recession-risk + market-regime dashboards
- Scheduled or on-demand delivery
📊 Momentum
📈 52-Week High
📉 Regime
Bull regime, low volactive
Recession risklow
Yield curvenormal
Latest report2 hr ago
Watchlists + paper portfolios
Track stocks you care about beyond the monthly model picks. Paper portfolios let you test strategies before committing real capital, with full cost-aware return tracking.
- Personal watchlists with CSV import
- Simulated trading per workspace
- Transaction-cost-aware returns
- Position-level realized vs. unrealized
Watching
12 positions
AAPL · +4.2%1y
GOOG · +12.8%1y
UNH · −3.1%1y
TSM · +18.4%1y
Analyst tools + sharing
Quants share completed models with analysts via the workspace. Read-only access for review; fork-grade access for analysts who want to run their own backtests against the model's predictions.
- Role-based access: investor / analyst / quant
- Per-share access level (read or fork)
- Expiring shares + revocation
- Decline reasons + accept/decline workflow
Shared with you
3 pending
Q3-2026 ensemblefork
52-week breakout v2read
Recession-tiltfork
Momentum tiltread
Access tiers
Pick the role that matches how you'll use StockPickAI.
The platform is currently invitation-only. Tier names map to feature surfaces; an admin can change the role on any user at any time.
Investor
Read-only research
Monthly picks, market signals, watchlists, paper portfolios, regime dashboard.
- Top monthly picks + regime context
- Personal watchlist + CSV import
- Paper portfolio simulator
- Market reports + clickable tickers
Analyst
Research + share workflows
Everything in Investor, plus model library, backtest lab, custom reports.
- Model library with regime + Sharpe filters
- Backtest lab — monthly + quarterly
- Receive shared models (read or fork)
- Build custom market reports
Quant
Train + share + publish
Full pipeline. Train models, run suites, publish completed experiments to analysts.
- Train your own XGBoost models
- Suite-driven A/B experimentation
- Share with workspaces or analysts
- Publish reports + paper-portfolio runs
vs. the rest
Institutional-style signals made approachable.
Most retail products optimize for engagement (buy/sell signals, daily noise). StockPickAI optimizes for the kind of slow, transparent decisions long-term investors actually need.
StockPickAI combines ranked picks, explainability, market reports, watchlists, paper portfolios, and analyst sharing in one workspace.
Historical research figures are provided for product context and are not a guarantee of future performance. StockPickAI is a research and decision-support product, not personalized investment advice.
| Dimension | StockPickAI | Traditional products |
|---|---|---|
| Output | Ranked probabilities | Buy/sell calls |
| Time horizon | One year | Intraday or opaque long-term |
| Explainability | Feature attribution | Limited transparency |
| Audience | Individuals + analysts | Institutions or active traders |
| Backtests | Walk-forward, regime-aware | Cherry-picked or absent |
| Sharing | Per-workspace, read/fork | Static newsletters |
FAQ
Common questions about how the model works and how access works.
Is StockPickAI investment advice?
No. StockPickAI is a research and decision-support tool. The platform produces ranked
probabilities, feature attribution, and historical backtests. Every output is for
informational purposes and should be evaluated alongside independent due diligence.
Nothing on the platform is personalized investment advice for your situation.
How are the monthly picks generated?
Each month, an XGBoost ensemble scores every eligible US-listed stock on
130+ features (fundamentals, momentum, valuation, capital structure, volatility,
regime). The top picks by calibrated probability of one-year outperformance form
the model portfolio. Cross-validation uses strict time-series splits so the model
never sees future data when scoring a past period.
What does "explainable" mean?
Every pick comes with feature attribution: the model exposes which inputs
contributed most to its probability score. You can drill into a stock's detail
page to see momentum / earnings / cash flow signals broken out individually.
This makes it possible to stress-test the model's thesis before acting on it.
Why is the holding period one year?
Two reasons. First, one-year horizons match how long-term investors actually
think — quarterly noise washes out, regime effects emerge. Second, one-year
holds qualify for long-term capital gains tax treatment in most jurisdictions,
which is a real component of after-tax return.
What happens after I request access?
Click "Request access" on the dashboard, sign in with Google, and your request
lands in the admin queue. An admin reviews and onboards you with a workspace,
role assignment, and an MCP/API key. You'll get an email when your workspace is
ready (if email is configured on the deployment).
Can I use StockPickAI from Claude / a CLI?
Yes. StockPickAI exposes an MCP (Model Context Protocol) endpoint at
mcp.stockpickai.com/mcp. Claude.ai, Claude Desktop, and Claude Code
can all connect with a bearer key from your dashboard. The MCP surface includes
experiment listings, predictions, paper portfolios, market reports, and more.
Is my data shared with other workspaces?
No. Every workspace is isolated end-to-end. Experiments, watchlists, paper
portfolios, market reports, and API keys are scoped per-workspace. Sharing is
explicit and opt-in — a Quant shares an experiment with an Analyst, and the
Analyst has to accept the share before the model appears in their library.
Access level (read vs. fork) is per-share and revocable.
Will there be a paid plan?
Right now the platform is invitation-only and free for approved users. We expect
to introduce tiered pricing once we're past the closed-beta phase; pricing will
reflect compute usage, not user count. Sign up for updates via the contact form
below to be notified.
Get in touch
Request access, ask about the methodology, or talk about partnership.
Thanks — your message is in. We read every one and respond within two business days.
We read every message.
If you want immediate access, the fastest path is the Request access flow on the dashboard. Use this form for anything else — methodology questions, research partnerships, press, or technical discussions about the platform.
Direct: hello@stockpickai.com
Responses typically within two business days. By submitting this form you consent to us emailing you back about your inquiry; we don't add you to a marketing list.