Risk Is Like Energy
The first in a series on practical learnings from building and running the quant infrastructure at Vertus. With billions at stake, in the most competitive market on earth.
The only thing that matters to quants is generating superior risk-adjusted returns.
Neither risk nor return can exist in a vacuum; they are always in a dependent relationship. As developers of quantitative trading models, our main goal is to maximize return over risk. Both of these variables have unique and immovable properties. In the following series of blog posts, I will go over some of them, and let you see through my lense of building and running the quant infrastructure at Vertus.
Risk is like energy. It can be transformed, but never destroyed.
Imagine a ball of Play-Doh, representing risk, laying on a flat plane, representing the universe of investable assets. You could smash that ball of Play-Doh out as far as you wanted, the amount of Play-Doh on the table wouldn't decrease, it would only spread thinner. Risk behaves similarly, it can only be transformed, never dissolve. In practice, spreading the Play-Doh is called diversification and works best when spread as far as the yielding investable assets allow.
In most cases, such a naive, broad diversification strategy outperforms sophisticated discretionary pick-and-choose approaches. This is evident by the historical underperformance of actively managed funds vs braod market ETFs.
We never wanted to be an ordinary fund though. When building diversification models for quantitative investment strategies, atleast four dimensions of diversification have to be utilized.
The most basic two-dimensional idea of diversification is the "What" and "How much"; our Play-Doh on a flat plane. Importantly, the "What" has to encompass single assets, such as stocks, commodities, crypto, etc., but also include any vehicle that creates a tradeable equity curve (e.g. strategies, funds, derivatives).
An obious add is the "When" dimension. It encompasses the timing of diversification decisions. With the inclusion of market timing, we are already starting to fall outside of classical fund management at this point. To effectively time diversification decisions (as with any allocation decision) we require a model that takes into account more than the inherent data of the underlying asset(s) and that is able to withstand varying degrees of market efficiency.
The 4th dimension, the "How", is a bit more complex. It looks beyond underlying assets, allocation sizes and timing, to different investment approaches. For example: if your underlying is a stock, do you enter using an option, outright buy shares or use other derivatives? If your underlying is a tradeable equity curve, how and why do you enter a trade 1. on the underyling strategy and 2. on the portfolio? Two strategies trading the same asset on the same day, and in the same direction can diversify your portfolio by controlling "How" they enter and exit positions.
While these points highlight the opportunities in complex diversification techniques, they also present the necessity to think multi-dimensionally. Your Play-Doh might be well spread across the 2D plane, but does it clump in multidimensional space?