Early Access Now Open

Trading signals built
like a quant hedge fund

Most traders lose money relying on news, opinions, and gut feel. This platform delivers live signals generated by machine learning models and systematic quantitative research — the same methods used by institutional funds.

0 Sharpe — 2024 out-of-sample

0 Monthly Win Rate — 2024

0 CAGR — 2024 out-of-sample

0 Sortino — 2024 out-of-sample

⚡ Limited Early Access Spots

Be first when signals go live

Join the waitlist to get early access to live signals, research updates, and founding member pricing before public launch.

✓ You're on the list. We'll notify you the moment signals go live.

No spam. No credit card needed. Unsubscribe anytime.

First access to live signals Founding member pricing Private research updates

The Reality

Quant models vs. everyone else

Markets are too complex for opinions. Quantitative systems don't read the news. They measure what is actually happening in the data.

❌ How most traders operate

✗

"AI stocks are going up" Following narratives and hype cycles without quantitative basis

✗

Twitter, YouTube, news Reacting to opinions from unverified sources

✗

Static strategies Using fixed rules that fail when market regimes change

✗

No risk control Trading through volatile, unfavorable market conditions

✓ How quant systems operate

✓

Cross-sectional ranking ML models rank stocks by predicted excess return using 80+ features

✓

Point-in-time data No future data leakage — every signal uses only information available at the time

✓

Regime detection The system knows when market conditions are unfavorable and abstains

✓

Uncertainty quantification Each signal includes a confidence score — the model knows what it doesn't know

Documented Performance — AI Stock Forecaster Only

One strategy. Fully auditable.

The AI Stock Forecaster is the only publicly documented strategy and the only one for which performance numbers are shown here. The signal platform will include additional proprietary strategies — including BLUE and others — none of which are published. What you see below represents a fraction of the overall signal portfolio.

Sharpe Ratio

DEV period 2016–2023. S&P 500 averages ~0.5. This is 6× better.

Monthly Win Rate

85.7% of months profitable in 2024 out-of-sample. Win/loss ratio 2.46×.

CAGR (2024 OOS)

228.8% geometric CAGR in the 2024 out-of-sample holdout period.

Regime AUROC

Model accuracy predicting when to trade vs. stay out of market.

AI Stock Forecaster — Full performance breakdown (20d long/short, vol-sized LightGBM)

Metric	Full Period	DEV 2016–2023	FINAL 2024+
Sharpe (ann.)	2.73	3.12	2.34
Sortino (ann.)	6.06	5.41	9.69
Ann. return (arith.)	87.0%	79.6%	137.3%
CAGR (geometric)	122.4%	109.9%	228.8%
Max drawdown	−18.1%	−18.1%	−8.7%
Calmar ratio	6.76	6.07	26.2
Hit rate (monthly)	82.6%	82.1%	85.7%
Win / loss ratio	2.17×	2.08×	2.46×
Mean RankIC (20d)	0.064	0.072	0.010
Regime AUROC	0.72 — predicts when the strategy should trade

Walk-forward backtests · Strict point-in-time data · No survivorship bias · 2016–2024+ holdout. These numbers are for the AI Stock Forecaster only — one of the strategies in the signal portfolio. Past performance is not indicative of future results.

What You Receive

A portfolio of signals — not just equities

The platform covers multiple asset classes and strategy types. Equities are one domain. The signal portfolio spans further — across markets, instruments, and approaches that aren't publicly documented.

📊

Live Trading Signals

Ranked lists of top buy and sell opportunities generated automatically by ML models. No opinions. No human bias.

🧠

ML Equity Rankings

Cross-sectional stock rankings based on predicted excess return vs. benchmark, produced by Gradient Boosting and LightGBM models.

🌐

Market Regime Detection

Signals are filtered based on current market conditions. When the model detects an unfavorable regime, it steps aside.

🎯

Uncertainty-Filtered Signals

Every signal comes with a confidence score. High-uncertainty predictions are flagged or filtered out before delivery.

🔬

Research Updates

Access to documentation and updates on new systems being developed, validated, and integrated into the signal portfolio.

🔒

Proprietary Private Models

Several strategies are not publicly documented and remain private. Subscribers gain access to signals from these systems.

How It Works

From raw data to ranked signals

A fully automated pipeline running continuously — no human intervention in signal generation.

Data Ingestion

Price data, fundamentals, earnings events, and market microstructure features collected from 80+ assets using strict point-in-time protocols.

Model Scoring

Multiple ML models evaluate each asset independently. Gradient Boosting ranks stocks by predicted excess return. Regime models score current market conditions.

Signal Delivery

Signals pass through uncertainty filters and regime gates before delivery. Only high-confidence signals in favorable conditions reach subscribers.

Public Research

One public strategy. The signals go much further.

The AI Stock Forecaster is the only publicly documented strategy. It has a peer-reviewed research paper on Zenodo and a fully open GitHub repository — so you can audit exactly how the methodology works.

The signal platform includes additional proprietary strategies — including BLUE and others currently in validation — that cover different asset classes and approaches beyond equities. None of those are published anywhere.

AI Stock Forecaster BLUE (private) + more in validation Equities Multi-asset Regime detection DEUP uncertainty Walk-forward validated

1# Signal generation pipeline
2def generate_signals(universe, date):
3  # Regime gate — step aside if unfavorable
4  if regime_gate.G(date) < 0.2:
5    return "ABSTAIN"
6 
7  # Score all assets
8  scores = lgb_model.predict(features)
9  uncertainty = deup.e_hat(features)
10 
11  # Filter high-uncertainty predictions
12  signals = signals[uncertainty < threshold]
13 
14  # Return ranked long/short list
15  return signals.rank_by(scores)

Quant Researcher

About the Founder

Ursina Sanderink

Quantitative Analyst & ML Engineer

I've spent years building algorithmic trading systems at the intersection of machine learning and financial engineering. The AI Stock Forecaster — one of my publicly released strategies — is documented research demonstrating the methodology these signals are built on.

Most of the systems generating signals are private, undergoing validation, or too novel to publish. This platform makes it possible to benefit from systematic, quant-driven research without building the infrastructure yourself.

The methods here are the same ones used by quantitative hedge funds. The difference is access.

GitHub Repository → Research Paper (Zenodo) →

Don't miss launch

The future of trading
is quantitative

Hedge funds have used these methods for decades. Early access subscribers get in before the platform opens publicly — at founding member pricing.

⚡ Join the Waitlist