Building The Meta Model on Numerai
Learn how Numerai combines thousands of models into one meta model.
Architect your AI scientist
Use frontier AI to model the market.
Compete for top-tier signal quality.
CodexClaudeCursor
$ # clone example-scripts, install MCP, and prompt Codex
$ git clone git@github.com:your-org/signalos.git
$ cd signalos && curl -sL https://signalos.ai/install-codex-mcp.sh | bash
$ codex exec --yolo "find the best neural network architecture to predict target ender"Start with fund-grade structured data.
Clean, regularized, and immediately model-ready. Abstract enough to distribute broadly, structured enough to build with on day one.
Download Data| id | era | feature0 | ... | featureX | target |
|---|---|---|---|---|---|
| n2b2e3dd163cb422 | era1 | 0.75 | ... | 0.00 | 0.25 |
| n177021a571c94c8 | era1 | 1.00 | ... | 0.25 | 0.75 |
| n7830fa4c0cd8466 | era1 | 0.25 | ... | 1.00 | 0.00 |
| nc584a184cee941b | era1 | 0.25 | ... | 0.00 | 1.00 |
| nc5ab8667901946a | era1 | 0.75 | ... | 0.25 | 0.25 |
| n84e624e4714a7ca | era1 | 0.00 | ... | 0.75 | 1.00 |
Each id represents an asset at a specific time era. The features describe normalized quantitative attributes at that point in time. The target is a future-looking performance measure resolved over a delayed horizon.
Apply machine learning to predict the market.
Build a baseline model with familiar Python tools.
Everything you need to move from features to live predictions.
Clone Example Scripts4.4K
Staked models
#!/usr/bin/env python
""" Example classifier on SignalOS data using an xgboost regression. """
import pandas as pd
from xgboost import XGBRegressor
training_data = pd.read_parquet("train.parquet").set_index("id")
features = training_data.filter(like="feature_")
target = training_data["target"]
model = XGBRegressor(
max_depth=5,
learning_rate=0.01,
n_estimators=2000,
colsample_bytree=0.1
)
model.fit(features, target)
predictions = model.predict(features)| id | prediction |
|---|---|
| n60dffdaceb7e467 | 0.25 |
| nadaeef0214b84a8 | 1.00 |
| nb13883520a4344f | 0.25 |
| n423766c5a4fa42a | 0.75 |
| n252b14301e46a31 | 0.25 |
| n75a5baf93a624cc | 0.00 |
Example target predictions.
Submit models to predict the stock market.
Build reputation to claim your place on the leaderboard.
Stake on your model to earn (or burn) cryptocurrency.
$23.8M
Paid to data scientists
Learn More
Introducing Numerai Signals
A tournament for data scientists who have their own data.
Original atmospheric artwork generated for this prototype.
Join the network and build a machine-native research system.
Built for contributors who want structured data, repeatable submissions, and measurable signal quality.
Receive updates about SignalOS
Note: This prototype is a product design exercise inspired by quant research platforms. It is not an investment solicitation and is not affiliated with Numerai.