Generative AI: Model as a Service in Fintech

As the manager of the Magellan Fund at Fidelity Investments between 1977 and 1990, Peter Lynch averaged a 29.2% annual return, consistently more than double the S&P 500 stock market index and making it the best-performing mutual fund in the world. His quote, “Investing in stocks is an art, not a science, and people who've been trained to rigidly quantify everything have a big disadvantage” has sparked pushback among the earliest quant traders. At the same time Medallion, Jim Simmons purely quantitive fund ran in parallel to Lynch, and outperformed his fund year over year. Decades later Medallion continues to be one of the of the best-performing asset managers in the world.

Lynch’s statement decades ago intended to clarify that the human-in-the-loop will always be necessary for asset managers. But here we are 30 years later where even art can be generated by artificially intelligent machines. This age of generative AI promotes our thesis at Crysp even further: whether investment is an art, or a science - it is ready to be displaced by machines.

Generative AI introduces a level of personalization never seen before. Users can adjust the responses, whether they are visual, audio or textual until they are satisfied. From the asset manager perspective this means the generality of rob-advisors simply won’t cut it. Crysp is more than just a hedge fund in your pocket, it’s your own personal money-manager. Crysp learns who you are, what you like, how you invest - adjusting and complementing your picks in real-time to make sure you win in the long-run.

Model-as-a-service (MaaS) is a business model in which a company offers access to its machine learning models as a service to other businesses or organizations. This allows these organizations to use the models to make predictions, classify data, or perform other tasks, without having to develop their own models from scratch.

Don’t worry, Crysp is still strictly B2C. We continue to allow our faithful users (retail investors) to run amok in the stock market while we deploy our quant algorithms to make sure they survive and thrive. Yet, in a MaaS future, we have to adress the B2B opportunity. What if the Robinhood’s or eTrade’s of the world were to plug their data into our models? What if we would sell our offering as a Model-as-a-Service allowing the dominant broker/dealers of the world to leverage our adaptive quant strategies to build real-time hedges for their own users?

Imagine a stock market where every position is countered by a thoughtful hedge that takes into account hundreds of thousands of external parameters. The race becomes - who can feed Crysp’s model with the best data? Who has the highest-quality dataset of historical performance of their users? It’s a world where users may pick one broker over another because they offer the highest level of real time personalization and protection against the unpredictability of the market.

I won’t say I’m not tempted, but the consumer opportunity is so much more exciting. With continuous positive feedback on our app the road to broker/dealer has never been more clear to me. We are sure to see a wave of model-as-a-service startups, fueled by the traditional VC fomo. But still, the pure opportunity to build a new stock market from scratch, a company that will someday soon displace even the most traditional financial vehicles such as index funds, is the world I want to live in, and a world I want to help build.

Happy new year!

Dolev