Testing the Cumulative RSI(2) on Index ETFs

Joe Marwood has recently posted an article which includes a variety of tests of the RSI(2) when applied to the SPY etf or individual stocks. You can see the original article here.

A while back I also tested the RSI(2) on a variety of Equity Index ETFs and I figured that readers might be interested in the results.

The strategy which I tested uses the same rules that Joe used in Test 2 of his article.

To remind you, the rules are as follows…

  1. Security is above its 200-day MA
  2. Use 2-period RSI
  3. Add up past two days of the 2-period RSI
  4. Buy if the cumulative RSI is below 35
  5. Exit when the 2-period RSI closes above 65

For comparison purposes, I’ll use the same account and commission size used by Joe Marwood (Starting balance of $100,000, $0.01 commission per share).

If you’ve read Joe’s article you will already know that the strategy shows a profit when applied to the SPY, but that the edge appears to have degraded since 2008.

The most successful years seem to have been between 1993 and 2005. For this reason, the following test periods will be between 01/01/2005 and 01/01/2016.

Cumulative RSI(2) Strategy on QQQ.

When testing the cumulative RSI(2) strategy on the Q’s etf between 01/01/2005 and 01/01/16 we get the following results…

  • No. of trades = 108
  • No. of Winners = 69.44%
  • Average P/L% per trade = 0.25%
  • Average hold time = 4 days
  • Annualised return = 2.54%
  • Maximum drawdown = -13.48%
  • CAR/MDD = 0.19

The monthly breakdown of returns is as follows…

QQQ

Cumulative RSI(2) Strategy on DIA.

When testing the cumulative RSI(2) strategy on the diamond etf between 01/01/2005 and 01/01/16 we get the following results…

  • No. of trades = 92
  • No. of Winners = 66.30%
  • Average P/L% per trade = 0.12%
  • Average hold time = 4 days
  • Annualised return = 1.15%
  • Maximum drawdown = -15.21%
  • CAR/MDD = 0.08

The monthly breakdown of returns is as follows…

DIA

Cumulative RSI(2) Strategy on IWM.

When testing the cumulative RSI(2) strategy on the IWM etf between 01/01/2005 and 01/01/16 we get the following results…

  • No. of trades = 97
  • No. of Winners = 68.04%
  • Average P/L% per trade = 0.24%
  • Average hold time = 4 days
  • Annualised return = 2.19%
  • Maximum drawdown = -21.36%
  • CAR/MDD = 0.10

The monthly breakdown of returns is as follows…

IWM

The above results inspire little confidence in this particular strategy.

With that said, what if we removed some of the complexity?

The following tests use the same period of data and ETFs as above, except this time I will remove the 200 day MA market timing filter from the strategy and instead of exiting positions when the Cumulative RSI(2) rises above 65, the exit rule will simply be a close above the previous day high.

Simplified Cumulative RSI(2) Strategy on QQQ.

When testing the simplified version of the cumulative RSI(2) strategy on the Q’s etf between 01/01/2005 and 01/01/16 we get the following results…

  • No. of trades = 141
  • No. of Winners = 78.01%
  • Average P/L% per trade = 0.65%
  • Average hold time = 4 days
  • Annualised return = 8.72%
  • Maximum drawdown = -17.17%
  • CAR/MDD = 0.51

The monthly breakdown of returns is as follows…

ModQQQ

Simplified Cumulative RSI(2) Strategy on DIA.

When testing the simplified version of the cumulative RSI(2) strategy on the diamond etf between 01/01/2005 and 01/01/16 we get the following results…

  • No. of trades = 128
  • No. of Winners = 68.75%
  • Average P/L% per trade = 0.48%
  • Average hold time = 5 days
  • Annualised return = 5.69%
  • Maximum drawdown = -18.86%
  • CAR/MDD = 0.30

The monthly breakdown of returns is as follows…

ModDIA

Simplified Cumulative RSI(2) Strategy on IWM.

When testing the cumulative RSI(2) strategy on the IWM etf between 01/01/2005 and 01/01/16 we get the following results…

  • No. of trades = 141
  • No. of Winners = 70.21%
  • Average P/L% per trade = 0.53%
  • Average hold time = 5 days
  • Annualised return = 6.80%
  • Maximum drawdown = -25.93%
  • CAR/MDD = 0.10

The monthly breakdown of returns is as follows…

ModIWM

The above results show that the simplified strategy produces superior returns to the original strategy.

Finally, I  tested the simplified strategy on a database which includes all four index ETFs. That is SPY, DIA, QQQ and IWM.

The rules are the same as above…

  1. Use 2-period RSI
  2. Add up past two days of the 2-period RSI
  3. Buy if the cumulative RSI is below 35
  4. Exit when the close is higher than the previous day high.
  5. The maximum number of open positions allowed is one.

Because the database now includes 4 ETFs (which are all highly correlated), we need a method for choosing which ETF to trade when there are multiple signals on a single day.

To keep things simple, when more than one ETF produces a signal on the same day, the ETF with the lowest cumulative RSI(2) reading will be bought.

Simplified Cumulative RSI(2) Strategy on SPY, IWM, QQQ and DIA database.

When testing the cumulative RSI(2) strategy on a database which includes all 4 index ETFs between 01/01/2005 and 01/01/16, we get the following results…

  • No. of trades = 185
  • No. of Winners = 70.27%
  • Average P/L% per trade = 0.45%
  • Average hold time = 5 days
  • Annualised return = 7.70%
  • Maximum drawdown = -18.59%
  • CAR/MDD = 0.41

The monthly breakdown of returns is as follows…

ComboETF

One last thing…You’ll have noticed each of the above tests does not include the data between 01/01/2016 and today.

Seeing that the markets have been in almost free-fall, I think it necessary to look at how the strategy has performed during the start of 2016.

On the close of the 4th Jan 2016, all 4 ETFs produced a buy signal. The etf with the lowest cumulative RSI(2) reading was IWM….

IWM2

This position is still open, and it clearly demonstrates that to trade this type of mean-reversion strategy you need a pretty strong stomach!

To finish on a more positive note…the last time that the strategy produced a single trade drawdown of over 10% was during July 2011.

IWM2011

 

How is this finishing on a more positive note? Well, the following 30 trades produced 21 winners and a net profit of 24.45% .

 

Leave A Response

* Denotes Required Field