4 min. read
Systematic (algorithmic) funds structure global financial markets. Most of the most profitable funds in the world are systematic funds. The scale and sophistication of their backend systems is commensurate with that status.
The nature of IP in an industry where code bases cannot meaningfully be patented has led to a perverse set of labor agreements: The most qualified quants are often the least mobile, being tied up by half decade-long non-compete clauses and other draconian contractual arrangements.
This legacy model of organizing a systematic fund has never suffered a serious challenge.
There have been a number of Web2.0 attempts to crowdsource market forecasts but they have not succeeded:
Quantopian - Steve Cohen’s Point72 backed Quantopian with promises to put $500m into the best of a crowdsourced set of systematic strategies. It closed down in 2020 due to underperformance.
Numerai - An equity market forecasting competition whose payouts are in the form of a token, Numerai was promoted by Howard Morgan, an early employee at Renaissance Technologies. Numerai has existed for more than six years and currently enjoys less than $30m AUM.
The dev cycle behind systematic strategies is multi-year. To amateurs with a data science background, the notion of working on a forecast for a few weeks and then striking it rich is attractive. Market experts, however, do not crowdsource years of their professional life: Knowing the true length of the dev cycle, they stay away these models.
Forecast crowdsourcing models have therefore suffered from selection effects towards amateurism.
A real challenge to the legacy systematic fund model would focus on the nature of IP in the industry.
Small quant funds are often built by experts who, after years at top funds, try to capture the value of their own expertise. They build an admirable code base that focuses on a few niches but cannot compete at scale with the dev teams of the largest funds.
Were small quant funds to work together, the scale of their tech would outstrip that of their largest competitors. They of course do not coordinate, though, because to do so would reveal their IP.
That is a funny reason not to work together, however. Most market forecasting algorithms don’t know the backstory & codebase of the features they take in. Does coordination among experts really necessitate sharing IP?
We have built a statistical protocol that allows quant funds to work together without revealing their IP. The related tech - an AI, OEMS and staking system - greatly accelerates the dev cycle of systematic strategy-building, and can be used in a number of ways.
We change the unit economics by shrinking the unit down multiple orders of magnitude.
Users contribute components of forecasts - features and algorithms. These components require less build and are less prone to overfitting than forecasts. Our AI is responsible for the parts of quantitative development that are so challenging even seasoned experts regularly fail.
We use the now-proven power of decentralized networks to scale information, and a powerful AI to put that information into action, along with a token-staking system that rewards the builders of that network.