Skip to main content

What is SignalNet?

· 3 min read
SignalNet Team
SignalNet Team
Building the Signal Network

SignalNet is a decentralized platform that turns the world's best quantitative minds into a collective intelligence engine. Instead of one hedge fund hiring 50 quants, SignalNet lets thousands of independent contributors compete, collaborate, and earn — all without ever revealing their models.

The Core Idea​

Every week, SignalNet publishes an encrypted dataset of financial features covering the S&P 500. Contributors — quants, data scientists, ML engineers, hobbyists — download the data, build predictive models, and submit their stock rankings.

These individual signals are then aggregated into a meta-signal that consistently outperforms any single contributor. Institutional funds subscribe to this meta-signal. Contributors earn rewards based on how much their signal helped.

Why This Matters​

The Talent Problem​

The quant finance industry has a structural inefficiency. The top 20 hedge funds employ roughly 5,000 quantitative researchers. Meanwhile, there are an estimated 500,000+ people globally with the skills to build competitive trading models — data scientists, ML engineers, physics PhDs, Kaggle grandmasters.

Most of them will never work at a hedge fund. Their alpha dies on the vine.

SignalNet gives them a path to monetize their skills without relocating to Greenwich, passing a Series 65, or raising a fund.

The Diversity Advantage​

The mathematical insight behind SignalNet is simple: diverse, weakly correlated signals combine into something far stronger than any individual signal.

If you have 50 contributors, each with an information coefficient (IC) of 0.02 and average pairwise correlation of 0.15, the ensemble IC is approximately:

IC_ensemble ≈ IC_individual × √(N / (1 + (N-1) × ρ))
IC_ensemble ≈ 0.02 × √(50 / (1 + 49 × 0.15))
IC_ensemble ≈ 0.02 × √(50 / 8.35)
IC_ensemble ≈ 0.02 × 2.45
IC_ensemble ≈ 0.049

That's a 145% improvement over any individual. And it gets better as contributors become more diverse.

The Trust Problem​

How do you trust a platform with your alpha? You don't have to. SignalNet's encrypted features mean:

  • You never share your model — only your predictions
  • Features are obfuscated, so you can't reverse-engineer the data pipeline
  • Signal hashes are committed on-chain before resolution — no retroactive changes
  • Payouts are verifiable via merkle proofs

How It Works in Practice​

  1. Monday: New round opens. Download encrypted features for ~503 S&P 500 stocks.
  2. Friday: Submission deadline. Upload your stock rankings (0.0 to 1.0) and stake SIGNAL tokens.
  3. Next 20 trading days: Daily provisional scores show how your signal is tracking.
  4. Resolution: Final scores published. Rewards distributed based on IC (accuracy), TC (uniqueness), and MMC (ensemble contribution).

Who Is SignalNet For?​

  • Quants who want to monetize alpha without running a fund
  • Data scientists looking for real-world ML competitions with real stakes
  • ML engineers who want to apply their skills to finance
  • Students who want to build a verifiable track record
  • Funds who want institutional-grade signals without building an in-house team

What's Next​

We're launching our first tournament rounds soon. Early contributors will receive token grants and priority leaderboard placement.

Read the docs → Launch the app →