Simple Strategies combined intelligently can best survive a fast-changing, complex world.
How We Do Things
We develop dynamic trading strategies that are “market friendly”. We prefer to keep things simple but we also appreciate variety. We call this dichotomy simplistic diversity, which we believe to be a fruitful source of active alpha. This means we focus attention on just a few well‐understood models and proven strategy types; combine them in a multitude of new and interesting ways, and then deploy the collection as ensemble strategies into distinct and identifiable market regimes so as to ensure the best fit of “strategies and market”.
The Quant Manifesto is a set of principles and beliefs that guide Space Machine's model building and strategy development:
(1) Keep it Simple: Market data has a high degree of noise, so extracting usable signals is always a challenge. Simple models and strategies are robust against noise. They are easier to understand, easier to debug, and easier to calibrate.
(2) Explore Combinations: Combine models and strategies Lego-like in many new and interesting ways to see how they perform under different economic environments and market regimes.
(3) Use the World as its Own Model: Market regimes, models, strategies, and portfolio weights can be indexed off of the state of the real world so as to capture the timing benefits of economic news releases, and for more precise trading in an information-based market structure.
(4) Evolve with the Market: Using the world as its own model, evolve models and strategies through the adaptive process of “market selection”; actively manage model life-cycle and portfolio weights based on a “goodness of fit” to current market regimes.
Existential threat to our core business is a recurring theme throughout Space Machine’s history. It had made an indelible impression on our organizational thinking. As a result, our investment approach reflects a strong bias towards staying lean, keeping overhead to a minimum, and using light-weight models that can adapt to a fast-changing market.
Our team adheres to a rigorous investment methodology that can harness alpha from across all stages of our integrated “development to deployment” workflow, separately and together:
(1) Data Exploration: Build a quantitative, phenomenological understanding of financial markets from a market microstructure perspective by asking questions about market irregularities and stylized facts as revealed in statistical patterns of financial time-series data.
(2) Model Building: Capture the underlying “rhythms and rhymes” of certain identifiable market regimes from contemporaneous market data using an adaptive, evolutionary approach to parameter estimation, model selection, and diagnostic testing, at scale.
(3) Strategy Design: Transform market insights into a set of trading rules for encoding a strategy that is driven by custom indicators built upon validated model forecasts in order to extract predicted profits from the market based on a positive mathematical expectation of returns over a large number of trades.
(4) Risk Management: Allocate and rebalance a dynamic portfolio of trading strategies, deployed broadly across a large number of assets in geographically dispersed markets, with the objective of generating consistently strong overall returns regardless of market conditions while keeping drawdown risks below acceptable thresholds.
Nothing is left un-optimized in the systematic capture of precious alpha from end to end.
What is more, we run a tight ship when it comes to quality control. Diagnostic checks are routinely performed on models and strategies across all of their interfaces to ensure semantic consistency. Every software component is individually reviewed, scrutinized, and stress tested at each stage of development. We believe that a robust, methodical, and disciplined peer review and examination process can reduce the risk of deploying over-fitted models, under-validated strategies, or misconfigured software to almost zero.
In short, all of us are unapologetically frugal and extremely paranoid about hidden risks, machine errors, and human blind-spots – a deeply ingrained survival trait in the organizational DNA.