Blog: Correlation Cracks and Breakpoint© Volatility Signals

The last week’s equity and volatility moves have erased much of the year’s stocks gains and have signaled the end to a generation of levered volatility ETNs such as XIV (See Bloomberg's Matt Levine for a summary). There are many possible reasons that could have triggered this increased volatility, VIX short squeeze and equity market decline. Some of which being:

  • Central Bank Tightening: Signs of stronger wage growth and a new Fed Chairman led to a sell-off in rates.
  • Crowded Positioning: The extremely popular and profitable short vol, short convexity and other carry trades. See the featured article in EQ Derivatives (paywall)
  • Grab bag: White House Tweet du jour or Bitcoin price collapse.
  • Portfolio Re-risking: Rising realized volatility and correlations.  Therefore risk managers told traders to unwind positions.

Whatever the cause, this sell-off didn’t emerge overnight. Cracks emerged in the stock and bond market in the week leading up to this sell-off that produced identifiable signals for investors. These signals warned them that the volatility and correlations had changed suddenly; we call these signals Breakpoints©.


These cracks are identified in real-time as more data is released into the market. In some sense this is like the technical indicators most traders are used to (moving average crossovers, RSIs, Bollinger Bands, etc…). However this is part of a next-generation set of sophisticated and adaptive algorithms that simultaneously look at changing Factors, Trends, Volatilities and Correlations.


Let’s look at what Breakpoints© say about the correlations between Stocks and Bonds. One of the biggest assumptions in recent years has been the assumption of a negative stock/bond correlation. That’s what I was taught at MIT, and that’s the assumption that JPMorgan, Betterment, Schwab and any other asset manager. It’s a normal reasonable assumption since the realized correlation between stocks and bonds over any given month or year is usually negative.  

As a Global Macro investor by training, I’m always concerned about regime change.  A move to a positive stock-bond-correlation world is a big change. In fact as central banks become less accommodating, this regime becomes more likely and it is a reality that we have seen play-out before. During QE1 and QE2 we saw stocks and bond prices rise together. During the last two weeks we saw the opposite thing.

As good as we all are at finding patterns in charts, these subtle changes in noisy data are incredibly hard to observe with your eyes.

The Breakpoints© framework is based on probabilities, not indexes.  As a result, we can not only view the most likely correlation level, but also the probability that the correlation is positive.


Each volatility estimate below is the 1 day ahead forecast for the estimate. The grey bands represent the 25% and 75% confidence bands around them. Notice how the Breakpoint© forecasts tend to jump to higher levels of vols AND snap into lower vol levels more quickly than the moving average measures.

Estimated S&P 500 Volatility spiked to 15% (from under 10%) and Correlation went positive on Thursday Feb 1st.  This means that as of the close on Thursday investors using this signal would have had the opportunity to reduce the risk in their portfolio, in particular their equity exposure.

Here are the forecasts for correlations. Again notice how much more adaptive these estimates are than traditional moving average models. In particular notice on the last day that the correlations returned to negative levels when the stock market crashed and bonds rallied.

Astute readers (yes you) will notice the differences in the positive correlation regime this week with the one in late December last year.  That positive correlation was driven by a rise in bond volatility and a drop in equity volatility.  The December risk-on ‘melt-up’ didn’t have as large of a portfolio risk effect as this week’s move.


Later this week we’ll be diving deeper into the implications for portfolio construction and introducing other Breakpoint risk metrics that look at index sub-components and across larger numbers of securities.

Please contact us at or on Twitter: @Astrocyte_Rsrch

Thanks for geeking out with me.  It’s been a big week.

- Sean Kruzel