Monday, October 19, 2009

Two Simple Relative Strength Rotation Models

With markets roaring back and the SPY having broken $110 for the first time in over a year just moments ago, it seems that Relative Strength Rotation methods have come back into vogue.

There are literally dozens of ways to define relative strength, and it is important to recognize that different computations will serially and stably outperform their peers during various behavioral epochs. Furthermore, inflection points can be harsh using this method. Various smoothing methods, use of multiple time frame references, a modicum of forced diversification, and standard money management techniques can all be a big help with that. All in all, it's a hard strategy to beat over time for those interested in always being invested.

Below I present two very basic mechanical trading systems employing rankings of current price divided by variously weighted simple moving averages among the selected ETFs. (The specific relative strength readings for which are provided every night in ETF Rewind* under the "Weighted Strength" column for nearly 200 tracking ETFs.)

Asset Class Rotation

The chart below indicates the equity curve that would have resulted from rotating into the single top performing major asset class ETF among the SPY, EFA, EPP, EEM, DBC and AGG, as ranked according to relative strength, then re-balancing weekly on a simple/ non-compounded basis, with no friction/ trading costs assessed.

The compound average annual growth rate for the nine-year study period would have been +10.7% with a simple Sharpe Ratio of +0.5 and a maximum peak-to-valley draw down of -19.4% (versus the S&P500's -51.8%). The equity curve is not optimized in any way, and involves no use of leverage or shorting: this is merely an extremely simple macro-asset-class switching method.

Currency Rotation

As currencies have been highlighted in the news lately, attached is a graphic highlighting another simple strategy rotating into the top two performing Currency ETFs among UUP, FXA, BZF, FXE and FXY, re-balancing weekly on a reinvested/compounded basis, with no friction/ trading costs assessed.

The compound average annual growth rate for the three-year study period would have been +14.2% with a simple Sharpe Ratio of 2.4 and a maximum peak to valley draw down of -8.9%.


At the very least, relative strength systems can provide natural stops for equity traders and additionally inform them as to which classes, sectors, styles and countries are running hot or cold in the current market environment. At their best, they can make for powerful trading systems in their own right.

[Past Market Rewind Trading System Posts]

Never Investment Advice


nzbryant said...

When you say 'top' performing currency etc. how did you measure 'top'?

John said...

I think the EEP symbol was a type it should be EPP.

Jeff Pietsch CFA said...

@nzbryant - I used the relative strength measure loosely described in the article.

@John - You got it, thanks.

Anonymous said...

You probably talked about it already many times but what software do you use ? I'm using Tradestation but I'm frustrated by the lack of porfolio backtesting. Thanks.

jgpietsch said...

@anon - There are many back testing platforms available AmiBroker, TradersStudio, etc... However, I most often use Excel. Best, J

Tamas said...

Do you think there might be a data-mining bias in your currency results ? You could have chosen to include GBP, CAD and MXN - all of which have GDP higher than Australia. Russia, India and South Korea are further candidates for more EM flavour though not sure if the relevant ETFs exists. As such, picking which currencies to include in your basket will influence the results.

Furthermore there is no comparison for a simple buy and hold strategy for the EUR (or Gold, or short USD basket).

jgpietsch said...

@ Tamas - Data snooping is always a risk and it is possible I have a happy sample here. While my focus is on introducing a method rather than making a recommendation, naturally, I suggest traders do their own research and testing.