In the first part of this series, I described the construction of the “Flash Crash Oscillator,” in which we considered cross-correlations among a broad spectrum of market sectors using twelve common ETFs as time series proxies. Part 1 may be found here.
This post will focus on two proposed schemas to either trade the oscillator or alternatively use it as an effective directional market filter for other systems, as well as discuss the intuition behind these proposed systematic approaches.
Two Systematic Approaches
In the charts below, I propose two related methods of trading the oscillator. In the first version (indicated by the light red hypothetical equity curve in the second table), we simply trade crossovers where one is long rises in oscillator, and short declines in the indicator.
In the second version (indicated in blue), we follow the first system, but anticipate turning points at far the reaches, fading the extremes before the cross under the expectation that a rapid reversal is soon to ensue.
[Please Click to Enlarge]
Over the last three years, the first method produced a frictionless, non-compounded return using the S&P 500 Spider ETF (SPY) of 28% per annum with a curve R2 of 91%.
The second anticipatory method performed slightly better with a 33% average growth rate and a linearity measure of 96%. It is not difficult to imagine simple improvements using underlying volatility and equity variance adjustments to further improve this performance.
Schema Intuition
The intuition behind these models is simple. When inter-market correlations are on the rise, it is typically a sign of a fast moving, volatile market where equities are being sold in tandem. When those correlations disperse, it is a sign of rising stability and market normality.
By percent ranking our absolute measures, we normalize readings for the current market environment. Nevertheless, as in the second model, extreme readings represent effective statistical tails, which may thus be faded immediately.
Certainly these returns would be difficult to achieve with trading frictions; however, although the oscillator is marginally computationally intensive, it is likewise highly intuitive and simple to replicate, possibly making it an excellent candidate for a joint trend confirmation and turning point indicator.
Never Investment Advice
Related posts:
