Live Dashboard

Prediction Markets,
Outperformed.

AI-powered forecasting engine that consistently beats market consensus across politics, economics, sports, and more.

See Live Predictions → View on GitHub
Prediction Accuracy
Orcetra vs Market Consensus
Market Baseline
Orcetra (66.6%)
57.1%
3,698+
24/7
57.1%
Win Rate
Beats AutoGluon & FLAML on 513 datasets
3,698+
Predictions
Resolved and verified
91.1%
Golf
Top performing category
72.9%
Politics
Election & policy accuracy
24/7
Monitoring
Continuous analysis

See How It Works

Watch our AI system analyze prediction markets in real-time

How It Works

Our three-step intelligence system identifies prediction opportunities through systematic market analysis

📡

Scan

Continuously monitors prediction markets for arbitrage opportunities and mispricing signals

⚖️

Calibrate

Multi-agent AI system analyzes market data, news sentiment, and historical patterns

✅

Verify

Generates calibrated predictions and tracks accuracy against resolved outcomes

Benchmark Results

Tested on 513 OpenML datasets with a 60-second budget per dataset. Three-way comparison against leading AutoML frameworks.

Overall — 513 Datasets

Orcetra 57.1%
AutoGluon 21.6%
FLAML 10.9%
Tie: 10.3% • 60-second budget • CPU only

Classification (382)

58.4%
vs AutoGluon 17.5% · FLAML 11.3%

Regression (131)

53.4%
vs AutoGluon 33.6% · FLAML 9.9%

Quick Start

Give it any dataset. It finds the best prediction.

$ pip install orcetra
$ orcetra predict housing.csv --target price --budget 45s
🎯 Orcetra v0.1.0
Step 1: Analyzing data... regression, 8 features
Step 2: Baselines... RF: 0.2554 ⭐
Step 3: AutoResearch loop...
Agent: LLM-guided (Groq)
#9 🎯 StandardScaler+HistGBM: 0.2403 (+5.9%!)
✅ LLM-guided search: +5.9% over baseline in 45 seconds
Two modes: 🤖 LLM-guided (analyzes history, proposes smart next steps)  |  🎲 Random search (no API key needed)

Why Orcetra

Superior prediction intelligence through advanced AI and systematic market analysis

vs. Prediction Markets

Markets reflect wisdom of crowds, but suffer from systematic biases like longshot overpricing and recency effects. Orcetra adds AI calibration on top, correcting these biases with data-driven analysis.

vs. Traditional Forecasting

No manual research. No expert surveys. No subjective judgment calls. Fully automated pipeline running 24/7 with transparent, verifiable results and real-time market integration.

Team

Led by experts in AI research and prediction systems

BL
Beibei Li
Chief Scientist
AI Professor at Carnegie Mellon University with expertise in machine learning and decision systems.

Faculty Profile
KZ
Kai Zhang
Researcher
Researcher focused on automated ML and prediction systems. Open-source enthusiast.
GZ
Guilin Zhang
Researcher
Building the prediction engine. Open-source contributor and ML engineer.
XH
Xiyang Hu
Researcher
Researcher at Carnegie Mellon University with expertise in machine learning, anomaly detection, and responsible AI.

CMU Profile
LM
Luyi Ma
Researcher
ML engineer focused on model optimization and ensemble methods. Core contributor to regression improvements.

See It In Action

Explore our live dashboard with 3,698+ verified predictions and real-time analysis

Live Dashboard →