Last Updated on 10 February, 2024 by Trading System
The R3 mean-reversion strategy is a strategy created by Larry Connors and his team and was published in his 2009 book called High Probability ETF Trading. Let’s take a look at the strategy and its backtested result.
The rules of the R3 mean-reversion strategy
These are the rules of the strategy:
- The close must be above the 200-day moving average.
- The 2-day RSI drops three days in a row and the first day’s drop is from a reading below 60.
- The 2-day RSI is today below 10.
- If number 1 to 3 is true, then enter at today’s close.
- Exit on today’s close if the 2-day RSI is above 70.
The Amibroker code for the strategy
In Amibroker, the code for the strategy looks like this:
Buy= RSI(2)<10 AND RSI(2)<Ref(RSI(2),-1) AND Ref(RSI(2),-1)<Ref(RSI(2),-2) AND Ref(RSI(2),-2)<Ref(RSI(2),-3) AND Ref(RSI(2),-3)<60 AND C>MA(C,200);
buyPrice=Close;
Sell= RSI(2)>70;
sellPrice=Close ;
The results of the R3 mean-reversion strategy
The strategy was tested on dividend-adjusted data set that covers the whole period the ETF has been listed, from inception until December 2020 (but commissions and slippage weren’t included). Here’s a table that summarizes the result and compares them with Connors’ results:
Result by Connors The average gain since Profit
The average gain in % inception to Nov.2020 factor Difference
DIA 0.17 0.65 2.48 0.48
EEM 0.91 0.45 1.4 -0.46
EFA 0.85 1.13 6.34 0.28
EWH 1.18 0.31 1.17 -0.87
EWJ 0.9 0.51 1.78 -0.39
EWT 0.85 0.31 1.29 -0.54
EWZ 1.94 1.99 4.83 0.05
FXI 1.19 1.32 2.98 0.13
GLD 1.08 0.76 2.54 -0.32
ILF 1.78 1.77 6.17 -0.01
IWM 1 0.74 2.23 -0.26
IYR 0.17 0.45 1.67 0.28
QQQ 1.24 1.04 2.46 -0.2
SPY 1.16 1.28 7.17 0.12
XHB 0.22 1.99 9.53 1.77
XLB 0.24 0.42 1.54 0.18
XLE 1.2 0.72 1.77 -0.48
XLF 0.46 0.26 1.11 -0.2
XLI 0.8 1.02 2.89 0.22
XLV 0.63 0.63 2.46 0
ETFs not included in Connors tests:
GDX 1.68 3.11
GDXJ 2.66 6.69
TLT 0.13 1.24
XLP 0.44 2.38
XME 0.97 1.86
The test showed that about 50% of the strategies performed worse since after the strategy was published 12 years ago. Although the five ETFs (GDX, GDXJ, TLT, XLP, and XME) at the bottom of the test were not part of Connors’ test, Quantified Strategies added them in their test to show how they performed with the strategy.
For the S&P 500 (SPY), the win-ratio for SPY is 90%, but there are only 79 trades since 1993. Here’s how the compounded equity curve looks like:
For the Nasdaq (QQQ), the compounded equity curve looks like this:
How the R3 strategy performs for a portfolio of all ETFs
It may be good to know how the strategy performs in a portfolio of all ETFs. Simulating the results on all 25 ETFs, from the year 2000 until December 2020, showed a good performance. The buy and sell criteria remained the same as above, but the system was set to make a maximum of five open positions at any time and risk no more than 20% of equity on each position.
Here’s how the equity curve looks like:
The results are as follows:
- The number of trades is 992
- The win-ratio is 75%
- The average gain per trade is 0.68%
- The profit factor is 2.08
The portfolio didn’t do well during the sovereign debt crisis in 2011, as it had a maximum drawdown of -16% in August 2011. However, the CAGR is 6.47%, so the strategy performed well but doesn’t seem to fit all markets.
The performance of the R3 strategy on a portfolio of SPY and QQQ
You may be wondering how the strategy performs on a portfolio of just the Nasdaq and the S&P 500. Here’s the equity curve of a portfolio that allocates 50% of the equity to each trade in QQQ or SPY and can hold 1 or 2 positions at any time:
As you can see, while the profit factor is high — 3.37 — the CAGR is quite low: 2.69%. However, time spent in the market is quite small: 4.5%.
Read More:
Larry Connors’ R3 Strategy (It Still Works) | Trading Strategies Explained