ETF Rotation Strategy using Sharpe Ratio

Rotation strategies, also know as momentum strategies or sometimes sector rotation strategies, operate on a simple premise.  The premise is we will buy the fund (ETF in this case) that has done the best (is the strongest) over the last few months or weeks or years, then we will hold it for a specific length of time then we will do our calculations again to see which fund is the strongest then will we switch our investment from the old fund to the new better performing fund.  The purpose of this strategy is to follow a trending market up, and rotate into a stronger fund as that market is getting weaker.

Strength is often measure as momentum only (or relative strength), and sometimes volatility is factored in, instead will build a rotation strategy using ETFs (exchange trade funds) that uses the Sharpe Ratio as the deciding factor as to which assets will be invested in.  The Sharpe Ratio is basically a composite of the momentum and volatility, it is refered to as the risk adjusted performance, that is how much performance do you get for a unit or risk.  A Sharpe Ratio of 1 means you get the same amount of return as the benchmark (typically a low duration bond at around 2.5%) for a fixed amount of risk.  So a portfolio with a Sharpe Ratio of 1 may have very different amounts of total return but the amount of risk for the higher return portfolios is more.

When selecting funds for a strategy keep in mind the funds goal, and its volatility and general performance.  For example mixing a short term bond fund like SHY and a high volatility fund such as ZIV or EDV would not work well together if all you looked at was volatility in the backtest.  Since SHY is very low volatility and the other fund is much higher results using volatility alone would just result in SHY being chosen almost 100% of the time.  Instead you could calculate the strength for the rotation based on momentum or use the Sharpe ratio with a low volatility factor (sometimes called F Factor).  This volatility factor modifies the traditional calculation of the Sharpe Ratio and is only used to adjust the position scoring inside the backtesting tool, all Sharpe ratios posted are calculated using traditional methods.

We are going to choose to rotate between SPY, EFA, and GLD.  This isn't the highest performing option you can choose, it just intends to show off how using the Sharpe Ratio as the scoring technique, and using market filters or inversely correlated asset classes can improve results substantially.

If we select 3 month lookback for the Sharpe Ratio we get the following chart, it gives 15.8% annual return, 0.86 Sharpe Ratio, 19.24% volatility, and a 46.26% draw down.  The purple line below shows the effect of choosing the top 1 for rotation, and the light blue line shows the result of an equal weighted portfolio of SPY, EFA and GLD.

This is a major improvement as far as return %, Sharpe ratio, drawdown and volatility as compared to the S&P 500 index or even an equally weighted portfolio.  For example over the same time period the S&P 500 (SPY) including dividends had a performance of 8.45% annually, a 0.51 Sharpe ratio, 19.57% volatility and a 55.44% draw down.

The big thing you will notice is the Rotation Strategy still is subject to a drawdown during bear markets.  Lets examine two ways of solving this issue.

Idea #1 is to switch to a cash fund or bond fund when the market is dropping, I am calling this a cash filter.

Idea #2 is to add a few bonds or cash like ETFs to the portfolio so the rotation can rotate into those ETFs if the others are weak.  We will choose US Treasury bonds as the cash ETF since they usually have a negative correlation to stocks.

First lets just test Idea #1 (Cash Filter) with SHY as a cash filter using the 150 day Simple Moving Average:

A big improvement, no more large drawdown and an increase in overall performance.  18.24% annual return, 1.07 Sharpe Ratio, 17.09% volatility, and a draw down of 24.87%.

Now lets look at the performance of Idea #1 (adding bond as cash filter) and Idea #2 (adding bond into rotation) for a few different bond funds, specifically SHY, IEF and TLT:

Idea #1 - Cash Filter - Results Testing with Different ETFs

Idea #2 - Adding Bond Funds to the Rotation - Results Testing with Different ETFs

It looks like adding that cash filter really improved results, and adding an ETF to the rotation helped some and hurt some.  Since TLT is the more volatile fund I think it was a better match to the other funds thus it performed better when added to the rotation.  In this case the cash filter did better than just adding another fund to the rotation, but if you can match the volatility of the funds (such as adding TLT) the results of adding the fund to the rotation can be good.

Below are the results for the winning cash filter of TLT at 150 period Simple Moving Average.

Final Sharpe Rotation Strategy:

Rotating Funds:  SPY, EFA, GLD
Cash Filter:  TLT
Rules:  Invest in the top 1 fund based on the 3 month Sharpe Ratio (volatility factor of 0.5)
Cash Filter Rules:  If the top fund is below it's 150 day moving average then invest in the bond ETF (TLT)



Annual Performance of our Final Strategy

Robustness
Are the values I choose above just flukes or is the strategy going to work well regardless of how exactly we choose the values presented above?

Sharpe Ratio Length (months)

Sharpe Ratio Volatility Factor (F Factor) [Using 3 Month Sharpe]

Cash Filter Length (Days) [Using 3 Month Sharpe]

The results show a wide range of inputs working well for the strategy, all results yielded better returns, Sharpe Ratios, volatility and drawdown numbers than the S&P 500 and an equally weighted portfolio.  We could choose better settings, like lowering the Volatility factor (thus making it closer to a momentum only strategy), and reduced the cash filter moving average length.


Conclusion
This is an example of how using a simple rotation strategy and adding in a moving average cash filter or inversely correlated assets to the rotation can help reduce draw downs, increase returns and eliminate the guesswork of investing.  Rotation strategies can reduce your risk while providing much higher returns than the market and high Sharpe Ratios (reward to risk).  The strategy presented here certainly isn't the end all be all of rotational strategies, but it shows how rotating funds into the strongest fund and avoiding market corrections with either inversely correlated asset classes (bond fund in this case) mixed into the rotation or by using a simple moving average filter to detect market corrections can result in a robust and realistic strategy.  Sharpe Ratio is another way to measure the strength of a fund in addition to purely using momentum and/or volatility, Sharpe ratio rotation can add value in comparing funds risk adjusted return against each other instead of just comparing their momentum or volatility alone.

2018 UPDATE:
​Many rotation based strategies suffered in 2015/2016 years, this strategy was no exception.  The 2015 year ended up at -2.4% with a 12.5% draw down, and 2016 ended up at -5.7% with a 10.8% draw down.  However in 2017 and 2018 there was a sharp turn around, 23% (with only 2.2% drawdown) returns in 2017 and 11.8% (with 10.1% drawdown) so far in 2018.