A protocol for quant funds and researchers to work together without revealing their IP.

<aside> 👉 This document is meant for professional builders of systematic strategies. A more general explainer is here. Our **site** and wiki main page are also good places to start.


Backstory And Context

We have built a tool for IP-free collaboration. It accelerates the development of systematic strategies, and makes the research that undergirds them more cumulative over time.


Our project has a typical dogfooding → SaaS origin story.

Problem: Judge Research and a quant fund wanted to partner to build a multi-hour systematic strategy focused on ETH-USD & BTC-USD. We had built useful data tools, and the fund had already engineered features that would be useful in our models.

The collaboration was hampered, however, by each party’s desire to not reveal core IP.

Statistical code cannot be effectively patented like regular software, leaving secrecy as the only way to protect IP.

Consequence: The strange nature of IP in quantitative development creates twin dynamics, squeezing the labor market in favor of the largest firms:

  1. Large funds use the nature of IP to exert labor market control via employee non-compete clauses, which might last half a decade.
  2. Small-to-medium sized funds cannot collaborate. Absent the manpower from either a large team or a collaboration, starting a new systematic fund is very risky: One has to hope some team of a half a dozen PhDs at a large fund is not coding out the same strategies.

(1) and (2) combine to disempower most quants from leaving large organizations and starting their own fund. The resulting labor market power of large funds centralizes the industry and its profits.

Solution: We built an AI on top of a massively parallel backend infrastructure. It assembles features & algorithms into forecasts, and those forecasts into systematic strategies.

In the permissionless version of the system, we use a token and a staking system to keep bad actors from participating.

How It Works: Funds can use the Judge Research SDK to reduce startup time to less than fifteen minutes. With one line of code they can pull tick-by-tick data on spot, options and futures markets, including funding rates.

Funds send features they have engineered on their own servers into our API, telling it what asset they aim to forecast and at what timeframe. Because funds send in their features - as opposed to the code that generates those features - the fund keeps its IP secret.

Our AI assembles those features into forecasts & systematic strategies.

Our OEMS, which has already processed more than $20b in real-world transactions, can execute the strategies, or funds can choose to use their own in-house systems.

Benefits: The SDK + our AI + our analytics tools empowers a fund’s researchers to:

  1. Build systematic strategies faster.
  2. Collaborate with other quant funds without revealing any IP whatsoever.

It does this without imposing any restrictions on a fund’s research & execution environments.

The SaaS → Decentralized Systematic Fund Relationship

The logic of a tool for IP-free collaboration can be extended from (permissioned) collaboration between two or three funds to collaboration between many.

With a token and a staking system added to it, it can organize a permissionless system, which can fairly be described as a decentralized systematic fund.

Users of the SaaS version will be able to collaborate with the funds/desks/individuals of their choosing. They will enjoy 90 free days and then pay a monthly subscription. The decentralized systematic fund will be always be free to participate, though require staking to send in more than a dozen features.

Like any good DeFi project, the incentive system has been designed to create massively disproportionate rewards for early participants. You can find out more in the tokenomics section.