Description
The Cumulative RSI strategy comes straight out of Larry Connors’ & Cesar Alvarez’s book called “Short Term Trading Strategies that Work“. I’ve really been enjoying programming and testing some of the ideas presented in their book — a lot of which seem to have some merit — so I wanted to go ahead and share some of the work I’ve been doing with my readers. On page 67 of their book, Connors and Alvarez state that this strategy was 88% accurate on the SPY when tested from 1993 through the date of publication, earning 65.53 SPY points with an average gain of 1.26% and and average holding period of 3.7 trading days.
There are actually 2 cumulative RSI strategies in the book, and this is the first from page 67. The second one is on page 104, and that particular cumulative RSI strategy is here.
The default settings (200, 2, 2, 35, 65, no) that the trading strategy comes with are straight out of the book on page 67, and seem to perform very well in my own tests, but they can be easily customized and tested with different values using the strategy properties menu. This is useful for quickly backtesting the strategy with different instruments, time frames, and market conditions. The alternate settings the authors suggest on page 68, for reference, are as follows: (200, 2, 2, 50, 65, no). These settings resulted in more trades and larger gains during the same time period, and only slightly decreased the accuracy, so it was a beneficial trade off. The results of those settings according to the authors were 85.46% winners, 105.95 total SPY points made, with an average gain of 1.05% and and average holding period of 3.57 trading days. The authors noted that this strategy picked up about the same number of points the SPY had done in the prior 15 years while only being in the market 20% of the time, greatly reducing risk exposure.
The original strategy in the book calls for no stop to be used, but I went ahead and added the option for a percentage based stop to make the strategy more applicable to different trading styles and time frames.
Results from thinkorswim strategy backtests can easily be exported and analyzed further in Excel or other spreadsheet programs simply by right-clicking on a strategy signal on the chart and clicking “Export” in the popup menu.
The Authors’ Stats:
- Instrument: SPY
- Test period: Almost 15 years
- Win Rate: 88%
- # Trades: 50
- Avg. P&L: 1.26%
- Net points: 65.53
- Avg. Holding Time: 3.7 trading days
What You Get
- Cumulative RSIs strategy file for thinkorswim, as shown in page 67 of the book
- All parameters are customizable in the properties menu
- Option to use a stop or not to use a stop, and to set the size in %
- Customizable colors
- A custom column to highlight setups on symbols in your watchlist
- A StockHacker scan to find setups if you want to scan ETFs or stocks for setups
- A lower Cumulative RSI-2 Indicator showing overbought/oversold conditions using the Cumulative RSI function instead of the normal RSI function.
Why You Want It
- The extremely high win/loss ratio on the SPY and other instruments makes it an easy strategy to trade from a psychological standpoint
- Option to add a stop makes the strategy even more easy to trade
- Long-only strategy further makes it even easier to trade for almost anyone, regardless of what type of account they trade out of
- Ability to quantitatively backtest the strategy on multiple instruments, timeframes, and conditions affords more peace of mind and encourages traders to fully trust in their system.
Cumulative RSI : How It Works
The cumulative RSIs strategy first makes sure the instrument is in a long term uptrend (with customizable parameters), and then takes the sum of the past x-number of RSI(y) values and if the number is sufficiently oversold, the strategy issues a buy signal, and if it then becomes sufficiently overbought, it issues a sell to close signal. The overbought and oversold levels can be customized as needed. Traders can optionally add a percent stop to the strategy (simple option in the properties menu) and customize how big the percent stop should be.
