Tuesday, December 8, 2015

Using High Beta ETF (SPHB) as a Hedge

High beta stocks ETF (SPHB) looks like an interesting hedge instrument. The chart below is a pair trade SPHB vs SPY with both sides normalized on 126-days (i.e. half a year) volatility:

This chart doesn't include any cost. Clearly, the main problem with using SPHB as a hedge will be the shorting cost. In InteractiveBrokers, which usually offers affordable rates for a retail trader, the current shorting rate is 2.6%:

The logic behind this hedge lies in the "low-vol effect" anomaly: high-beta stocks underperform low-beta stocks. What's important, the main driver behind the "high-beta minus low-beta" difference is high-beta underperformance (low-beta stocks outperformance is less pronounced).


  1. Any chance you can provide a walk through on how you normalize a return series? I've been dividing the log price return by 21 day StDev, however I have no intuition on what to do with that value to create the charts you display.


    1. On any day:
      - you have a daily log price change and a 126-day log price changes stdev (volatility norm).
      - divide log price change on volatility, you've got normalized price change.
      - substract first leg normalized price change from second leg normalized price change, you've got normalized pair daily change.

      Now cumulate those daily changes to get the chart.

  2. Thanks Alexander! I get the shape of the chart correct (more or less), but the values are a bit off.

    Here's what I get for SPY for the most recent 5 days:

    Date Close RTN Norm 126 STD RTN/126STD
    12/2/2015 208.529999 -0.010 0.013 0.009 -1.142
    12/3/2015 205.610001 -0.014 0.021 0.010 -1.476
    12/4/2015 209.619995 0.019 0.021 0.010 1.903
    12/7/2015 208.350006 -0.006 0.012 0.010 -0.590
    12/8/2015 206.949997 -0.007 0.012 0.010 -0.645

    When I combine them into a pair (I tried replicating the SPY/TLT pair), I get values from 20 to -60. My high point of 20 is in Jan 2009; your graph shows a 1.

    Do you annualize your volatility measure? Thank you very much for the help!

    1. Yes, forgot to mention that, volatility is annualazed.

    2. The charts looks great, I really like the thought process behind them. Any chance you could post some sample returns, vols, & vol adjusted returns for the two ETF between Jul-2012 & Jun-2013? I am still getting higher values (ie my chart has more vol that yours). Thanks!