Market Regime Dashboard

The previous post, and excellent attendant reader comments, posited that effective quantitative trading needs to dynamically adapt to context. Yet, one of the most difficult systemic quant problems is identifying relevant “context”, how to measure it, and how to visualize it.

All too often, people focus exclusively on analyzing raw price charts for context (technical analysis is particular guilty). This is unfortunate, as basic time-series analysis offers to tickle out and visualize myriad richness otherwise invisible.

A series of four posts seek to build a “visual dashboard” for market regime of a single instrument (in pictures, as one is worth a 1000 words). For simplicity, each post focuses on SPY during 2007 – 2008 and explore both time domain and frequency domain. The first two posts will focus on identifying regime structure in the time domain:

• Raw structure: visualize distribution and moments
• Correlation structure: visualize measures of self-correlation

The latter two posts will focus on identifying regime structure in the frequency domain:

• Frequency structure: visualize stationary sinusoidal frequencies via Fourier analysis
• Wavelet structure: visualize non-stationary wavelets

For readers new to these techniques, they can be understood via analogy to appreciating orchestra. First, one must appreciate the progression and sequence of notes, as expressed by a sheet of music, played by each instrument; this is the time domain. Second, one must appreciate how to distinguish the melodies and harmonies played by each instrument or group of instruments; this is the frequency domain. Both are important, each providing an orthogonal (or orthonormal) perspective on the same phenomenon.

This dashboard strives to provide basic visualizations which can capture the regime of many instruments, as applicable to both trading and risk management.