Speaker Q&A with Revant Nayar, Principal and CIO at Princeton AI and FMI Tech.

Future Alpha: What motivated you to build a quant fund with a different approach, and how does your philosophy shape the investment process?

Revant: Our core philosophy is to use the modelling principles that laid the foundation for the most successful theories in physics- namely parsimony (use the fewest possible parameters), falsifiability (a model should be able to be falsified by the data) and the fact that a model must predict something that was not an input in the model in the first place. We also insist on model explainability, which includes both explaining who our models profit against and drawdown explainability. We are not looking for fleeting opportunities, but demand universality- alpha that works across regimes, asset classes and geographies.

Future Alpha: Running a quant fund comes with unique challenges. What have been some of the most difficult decisions or trade-offs you’ve faced in managing your portfolio?

Revant: A lot of challenges are not what you might expect. Hedging is an important challenge, especially in times of great macroeconomic uncertainty such as was 2025. As a small fund, even the cost of purchasing risk factors from Barra/Axioma can be deterring so we have to be creative or rely on AI-driven solutions. I think AI will level the playing field for small funds as it has the capacity to aggregate vast amounts of regulatory filings and news data to produce risk factors that hundreds of analysts would compile manually at the legacy providers. The other major challenge is managing idiosyncratic moves. In 2025, general uncertainty extended not just to factors but individual stocks, as single names too were asymmetrically affected by tariffs, AI waves, chip deals and general news. Single names often rallied over 10 percent intraday, which is enough to cause a significant dent in the Pnl even if one has, say, a 5 percent concentration limit. These challenges exist even at the most placid times, but macroeconomic uncertainty exacerbates them.

Future Alpha: You’ve mentioned a quantum approach to alpha. Can you explain what this entails and how it differentiates your fund from traditional quant strategies?

Revant: Most quant funds have traditionally done trend following on factors or mean reversion on residuals. These common strategy types have, therefore, become risk factors rather than alpha factors, as we saw in 2025 when quant funds as a group underperformed for weeks at a stretch. A lot of other funds have tried using AI/ML techniques to capture nonlinearity but that tends to overfit and not handle non-stationarity well. Quantum approaches on the other hand, are parsimonious, and are ideal for highly noisy, non-stationary systems like financial markets. This differentiates us from the vast majority of strategies that either repackage trend following or stat arb, or try to generate alpha using LLMs. Our fund’s differentiation does not come just from the techniques we use but from the culture itself- we try to work with non-conformists and radically independent-minded people from non-financial backgrounds, which we think is essential for alpha generation. There are very few people who are contrarian and right, and we want to tap into that talent.

Future Alpha: Portfolio optimisation is often seen as a technical exercise - how do you balance advanced modelling with real-world constraints and risk management?

Revant: There are significant problems with standard portfolio optimisation approaches. In particular, they do not account for nonlinearities, nor for the multiscale nature of betas. One might be well hedged at the daily scale, but the biweekly or monthly betas might be high, and standard single-scale hedging techniques would leave us vulnerable to those shocks. This is addressed in the multiscale portfolio optimisation tools that we use, explained in my paper with our CRO Raphael Douady last year. Also, betas used in standard optimisers are computed in a backward-looking fashion with EMAs with arbitrary lookbacks. We have techniques that can handle these issues, some of which we will publish soon.

Future Alpha: How does understanding microstructure influence your trading and execution strategy, and what lessons have you learned from applying this in practice?

Revant: Clearly, managing transaction costs is critical for any mid frequency fund. Even if one relies on broker algorithms, there are parameters that one can optimise such as the execution horizon and the speed of trading, which can produce a positive impact on the Pnl. Most academically elegant solutions do not work so well in practice because they involve far too many free parameters and are not physically motivated. The most comprehensive literature on microstructure that is the best starting point, in my view, is Sarkissian’s work on the quantum theory of price formation. It contains several very useful frameworks one can use to optimise not just execution but also risk management.

Join Revant on Day 1 of Future Alpha 2026 at 11:40 AM for an insightful panel discussion: “Re-Architecting the Edge: Building Scalable, Intelligent Front Office Infrastructure.”

Secure your pass today!