Many indicators are often representations of similar measures. I prefer to not get bogged down with multiple measures of a trend. To keep things simple, I typically monitor price and volume, MACD, RSI, and sometimes the Stochastics. While the MACD can help traders judge the strength of a trend and also identify when a new trend is beginning, the RSI is often used to help with timing, i.e., when to get in, and when to get out of a trade.
Because of its simplicity, the Relative Strength Index (RSI) is one indicator that has always made intuitive sense to me. The calculation is simply the average of x days up closes / x days down closes. What traders have often differed on is what average to use. The default with most charting software is a 14 day average. However, several bloggers have advocated, or at least discussed using a 2 day(period) average. IÂ know Bill Rempell, Bullish Jim, and Marlyn from Filtering Wall Street (no longer being updated) have all presented trades using the 2 day RSI.
Some new research shows the benefit of using a 2 day RSI average. In the November issue of Technical Analysis of Stocks and Commodities, an article by Larry Connors and Ashton Dorkins describes the results generated during testing more than 8,000,000 trades from January 1, 1995, to December 31, 2006. The average one week percentage gain or loss for all stocks during the period tested was +0.25%.
After quantifying overbought and oversold conditions (RSI above 98 is overbought; RSI below 2 is oversold), their research showed that stocks with a 2 period RSI below 2 averaged a gain of +0.88% one week later (beating the benchmark average by more than 3:1). Conversely, stocks that were overbought with a 2 day RSI greater than 98 lost money (-0.17%) one week later as well as underperforming the benchmark.
The implications of this research are crystal clear: Traders should use a 2 period RSI if they want the indicator to actually give them an edge.
I have been experimenting with the RSI(2) for about a month and have anecdotally observed several characteristics of the indicator.Â
- First, it seems to work better with stocks that are trending up or down and making regular peaks and valleys.Â
- When applied to a stock in an uptrend, it will do a good job of identifying a buying opportunity, but does not seem to perform as well as a sell signal because good stocks seem to stay overbought for longer than they stay oversold. Â
- It seems to work especially well when the RSI(2) is 2 or less and the stock is sitting near support, whether it be a trendline or moving average.Â
- The RSI(2) does not seem to work well for stocks that have fallen off of a cliff or are in a sharp downtrend. In these cases, stocks can stay oversold for long periods of time before bouncing.
When I find stocks which are oversold on the RSI(2) and meet the above criteria, I will post the chart so that we may examine the effectiveness of the indicator.
2006. Connors, L. and Ashton Dorkins. “Does the RSI give you an edge?” Technical Analysis of Stocks and Commodities. 48-52, November, 2007.
cool, thanks
It seems to work best when there are no major news affecting the stock. With no ‘outside’ influences affecting the price, RSI(2) is a good indicator of when to buy and when to sell. Buy below 2, sell near 100.
So that pretty much agrees with your observations…
I am glad to see someone finally expand a bit on this indicator. It is certainly a very useful tool to time trades, especially powerful in conjuncture with MACD…
Since we’re on the subject, does anyone use William’s %R or do you all stick to Stochastics?
Marty
Moving it to a 2 period RSI tells me that you all are looking for the quick short term trade. It gets dangerous using lagging indicators for short term trades. However, you can use the lagging indicators to create a forward indicator by viewing them as histograms and search for patterns. Divergences are the best. I’m just sayin.
Marty- I used to use Williams, but now I primarily use Stochs. Do you use Williams?
Alavri- Not at all just for short term trades. It works great if you’re long a trending stock and want to add on a pullback.
Also, actually getting a reading <2 on the type of chart I mentioned in the post is somewhat harder to find than you might think. When it occurs during the right technical picture, it can signal lasting bottoms.
Reference this chart of IMA
http://stockcharts.com/h-sc/ui?s=IMA&p=D&b=5&g=0&id=p82572713342
Look at the end of October.
Wood:
How about a short list of stocks that fit the criteria.
I ran the filter tonight, and there are 8 stocks that make it, but all of them are either the 25% gap down variety or on the “sub-prime lender” death slide. The indicator tends to not work very well on those.
I began my trading ‘life’ using Williams, MACD, and RSI, so I certainly haven’t had any lack of indicators. I have peeked Stochastics a few times but have not sat down and dedicated any significant time to learning/reading them.
That being said, I’m not a trader by definition, I don’t have the money for it yet. My average hold time is not measured in hours, days or even weeks for the most part. But it is just as important for me to have indicators (and settings) to find good buy-in, average-down, and sell points; I just have a lot more time to do my research!
Marty
Lagging indicators suck no matter what lookback period or filter you put on them. You need to transform the lagging indicators to forward indicators. Stop with RSI, Stoc, MACD, etc. They do not work. Forward indicators include pattern recognition, fibs, pitchforks, ABCD and pivots. If these cannot be made to work for you, then switch to fundamental analysis.
Alvari, with all due respect, I’d like to see research that pivots, ABCD, fibs, etc., “work.”
I use all of those.
I’m just saying that I’ve given you research, proven, over 8 million trades that RSI(2) provides an edge.
You cannot make a blanket statement that “they do not work,” and then list other indicators, with the implication that they do work.
Wood,
Fair enough – but they dont.