NEW: Retirement Planner — household-level, tax-aware, end-to-end encrypted. See the full suite →
Stock picks · wealth preservation · retirement planning

Intelligent investing — from single picks to your complete retirement plan.

A probability-calibrated ML engine for stock selection, rigorous dividend screening for income portfolios, and end-to-end encrypted household retirement planning. Honest numbers, live track record, and zero-knowledge privacy on your financial data — all in one tool.

Beat S&P 500
Win rate
Matured positions
Live performance — matured positions only, verified daily against source database.

Three models for three timeframes.

Not one signal. Three independent rankings calibrated for different holding periods — so you can size your portfolio across conviction levels and time horizons.

💎
Investment Grade LOW RISK
180-day horizon
Quality-screened companies with strong earnings, insider conviction, and the highest probability of beating the S&P 500 over six months. Built for larger position sizes and longer holds.
  • Institutional-quality fundamentals
  • Composite quality + growth + momentum scoring
  • Tight stop losses (~10%) with ratcheting targets
🚀
Breakout Candidates MED RISK
90-day horizon
Growth-momentum stocks at technical breakout points, backed by fundamental support. Medium-term swings where the model's edge is sharpest.
  • Relative strength + growth acceleration
  • Technical breakout pattern recognition
  • Balanced risk / reward — core swing capital
Catalyst Momentum HIGH RISK
60-day horizon
Options-confirmed swing trades. Triple-verified setups where ML probability, options flow, and technical momentum must all agree before a stock makes the list.
  • Insider activity + recent catalyst detection
  • Options chain confirmation (IV rank, P/C, OI)
  • Highest edge when signals align — size accordingly
Retirement Suite · Phases 1–4 Live

Beyond stock picks — a complete retirement platform.

The same data science approach extended across the whole retirement lifecycle: dividend-stock screening, payout modeling, and household-level retirement planning with tax-aware withdrawals and end-to-end encrypted storage. Built for retirees, near-retirees, and anyone serious about their financial future.

Live sample Top 3 A+ wealth preservation picks right now
Loading universe...

Ticker names hidden — sign in to see the full ranked list of all A+ picks plus every stock that passed our screening.

Retirement Suite · Product Roadmap
Phase 1
Wealth Preservation Scoring
Dividend stocks ranked by Safety, Growth, Quality, Value.
Live
Phase 2
Income Monitoring
Ex-dividend calendar, cut warnings, portfolio integration.
Live
Phase 3
Payout Modeling
Monte Carlo simulator — how long your dividends sustain withdrawals.
Live
Phase 4
Retirement Planner
Household-level, tax-aware, end-to-end encrypted — accumulation through decumulation.
Live
Phase 5
Deep Analysis Tools
Historical backtesting, sector rotation, tax-loss harvesting, rebalancing.
Coming

Every recommendation, decoded.

Every pick includes probability, expected value, uncertainty bands, a stop loss, a sell target, and broker-ready copy. No guessing what the model thinks.

RANK #3 · 90D BREAKOUT
NVDA A+
NVIDIA Corporation · Semiconductors
+14%
Expected Value
73%
ML Prob ↑
+22%
range −6% to +48%
Target
↓8% / ↑22%
2.7:1 ratio
Risk / Reward

Illustrative example. Real picks are updated from the nightly pipeline and include live prices, stop-loss ratcheting, and historical re-recommendation tracking.

Built for people who want to evaluate the model.

No black box, no "proprietary AI" hand-waving. Here's what's actually happening under the hood.

Stacked ensemble classifier CORE
Four base learners (LightGBM, XGBoost, CatBoost, L1 Logistic Regression) combined by an L1-LogReg meta-learner. Trained on ~280 features across fundamentals, technical signals, options flow, insider activity, and cross-asset momentum.
Walk-forward validation HONEST
Models are validated on future data the training set never sees. A horizon-sized embargo prevents lookahead bias. A 4-way train / meta / calibration / test split separates model selection from final evaluation.
Isotonic probability calibration CALIBRATED
When the model says "70% probability," stocks go up approximately 70% of the time — by design, not hope. Calibration is verified empirically and displayed in the app.
Quantile regression uncertainty bands P5/P95
Every prediction includes a 90% interval. You don't just see "+22% upside" — you see the range of plausible outcomes, from the 5th to 95th percentile of forecasted returns.
Bidirectional expected value EV
Expected value is P(up) · r_up + (1−P) · r_down — the probability-weighted return across both outcomes. Not the degenerate formula that ignores downside.
Regime detection + drift monitoring LIVE
Feature distributions are checked daily via Population Stability Index. When features drift, the model is flagged for review. When regimes shift, the pipeline adapts.
The daily pipeline
Data acquisition
Prices, fundamentals, corporate actions, options chains, macro data — fetched nightly from our institutional data provider.
Feature engineering
~280 features built from raw data. Strict point-in-time joins. Nothing leaks from the future.
Train & validate
Models retrain nightly with horizon-sized embargo and 4-way splits. Bayesian hyperparameter tuning.
Predict & rank
Thousands of tickers scored. Top picks selected per model. Every prediction carries its uncertainty band.
Track performance
Every pick is logged. Win rate, alpha, calibration — verified against live market outcomes.
Deliver
Results ready in the app before market open. Broker orders copy-ready.

Ready to see what the model is picking today?

Access is currently invite-only. Request access to receive a code when a spot opens, or sign in if you already have one.

Built to avoid false precision.

The platform now frames each model output as historical setup reliability, forecast range, factor attribution, regime context, and portfolio risk — not as an oracle prediction.

Reliability, not certainty
Signals are presented as historical setup quality with sample-size context where available.
Uncertainty bands
Users see bear/base/bull return ranges instead of a single upside number.
Portfolio-aware risk
Position and sector concentration warnings reduce the risk of stacking correlated picks.