ATLANTA, April 17, 2017 — 

The best performing sector year-to-date has been utilities, a fixed income alternative with market exposure, lower beta, and a high dividend yield. The utilities sector consists of companies engaged in the production and delivery of electric power, natural gas, water and other services such as steam and cooled air. I recall that the impetus to evaluating the utilities sector came just before last year’s election when the economic and political outlooks were unclear. In particular, I was looking for individual securities that track the XLU’s low volatility, but with a strong propensity for higher alpha (relative return to the XLU). The resulting portfolio’s performance exceeded my expectations.

Image 1: Forward paper trading Utility live since November 4, 2016 – year-to-date return stands at 18.78% and max drawdown for the year is minimal at 1.9%. Excess return against the benchmark for the year 10.57%.

At Lucena, we perpetually analyze our live performance and often validate if a strongly performing portfolio such as our utility live portfolio has fundamental merit. Let’s go back and evaluate how utilities live was constructed.

Step 1 Compile a list of securities that together track the XLU: In order to find the constituents that best track the XLU, I have utilized Lucena’s portfolio replicator, a machine-learning-based pattern matching technology, to reconstruct a daily performance chart of a benchmark. In our specific case, I tasked the portfolio replicator to create an equity basket from the Russell 1K that tracks XLU. Since the XLU is based on SP 500 stocks and the resulting basket is a portfolio that is based on the Russell 1K, we ended up with a portfolio consisting of higher beta stocks (relative to XLU).

Image 2: Replicating XLU from the Russell 1K – the green line represents the XLU while the orange line represents the basket that tracks it.

Step 2 Actively rebalance the portfolio for max Sharpe:

Now that we have a basket whose individual securities are more volatile than the XLU but together track the XLU, we can continually respond to changing market conditions by re-optimizing our portfolio. By employing Lucena’s portfolio optimizer, we are able to re-adjust the allocations of the individual positions in the portfolio so that together they maintain the highest projected risk-adjusted return. In other words, by continuously overweighting the stocks projected to go higher and underweighting the stocks projected to decline, our portfolio will perform strongly even if it’s tracked benchmark, XLU, is flat. This concept can be best described by Shannon’s Daemon experiment. Shannon’s experiment is rather simple. Imagine you have a portfolio consisting of a single stock that either doubles or halves in value every day. Your portfolio is rebalanced daily by splitting your portfolio’s value in half: 50/50 between your single stock and cash. Assuming you started with $1,000 and the stock value is cut in half, your portfolio value at the end of the day would be $750. (Your cash remains at $500 while your stock was cut from $500 to $250 for a total of $750.) After rebalancing your portfolio, the next day, it would have $375 in stock and $375 in cash. Assuming the value of the stock doubles the next day, your portfolio will now be valued at $1,125 (a net profit of $125).

Image 3: As the chart demonstrates, after rebalancing only 72 times our initial $1,000 investment is now worth just under $100,000! In contrast, had you bought and held the stock without rebalancing and let it double and lose half its value every day, you would have made zero profit.

Conclusion

By applying the same concept of continuously rebalancing a portfolio consisting of highly correlated but more volatile stocks relative to a benchmark, we achieve higher performance over time. We’ve recently applied the same concept to other benchmarks (industrials and retail, for example) and discovered that the above concept still holds. In all cases, our research demonstrated stronger performance relative to the respective benchmarks over time.

Image 4: Representing a similar portfolio relative to the retail ETF benchmark (XRT) – backtest conducted since 1/1/2004 includes transactions costs and slippage. Portfolio’s performance although correlated to XRT is more than double the return and Sharpe ratio with significantly lower volatility.

Be careful! Results could be survivor biased: One word of caution, do not replicate an index based on today’s constituents and backtest them over time. You will be optimistically favoring surviving stocks. Historically, there is no way we could have known which stocks would have survived today. The proper method to apply a survivor biased free backtest is to roll back time and select constituents based on the market’s state at that time (historically).

http://www.linkedin.com