The default time setting for RSI on my Scottrade Elite platform is 14 (13 smoothing). I'm pretty sure this is the more popular setting. However, I just read an article where the author recommends using 21 to eliminate some false signals. IMO, technical analysis works because so many people follow it and react to it at the same time, right? So would you be better off keeping it at 14? The above assumes using a 3 to 12 month, daily chart. I don't use RSI on my 5 minute, one day charts but if I did would the time setting have to be changed? If so, what's the best for a 5 minute chart? Thanks.
I use a (14) on the 5 minute. Should track the previous 14 bars. I could be way off base, I use is to add strength to my decisions only.
RSI should be set based on the dominant cycles of whatever you trade. So experimentation will result in a currently best number. Smoothing is just that, an attempt to reduce noise (at the expense of lag). Technical analysis can be helped by multiple users but this is not essential if the underlying thing that the indicator is trying to do is valid when no one is using it. Sort of like the sound of one hand clapping.
If you were to trade for 6 months using only RSI(14) as an indicator and basing your trade decisions on a combination of it and your sense of good entries based on what you see in the chart, and if you were net positive at the end of that 6 months, I don't think your gains would be the result of 'so many people following it and reacting to it at the same time'.
RSI is a static indicator. The market doesn't always cycle from short term tops to bottoms in the same number of bars. This is what makes the market dynamic. To trade effectively and consisitently you have to find a way to compensate for this.
It helps to know where the 14 comes from before you put too much faith in it. J. Welles Wilder, the inventor of RSI, suggested 14 because it is half of the lunar cycle and he placed great significance in including the lunar cycle as a part of trading. Without getting into the psychodynamics of that particular stance, I'll just note that the lunar cycle is 28 calendar days, not 28 trading days. I'll also note that 21 is the average number of trading days in a month, perhaps making it more appropriate for a lookback period than a pseudo-astrological component. As always, YMMV.
bingo. When I use RSI on the indices, i have to increase the noise a bit to create signals. Your setting will depend on the volatility of the instrument. I use bloomberg...so i cant say what brokers or software allow you to manipulate the settings.
Indicators www.amazon.com/gp/product/customer-.../102-3046001-6164908?ie=UTF8&n=283155&s=books I believe this book by Constance Brown can be very helpful for most reasonably experienced traders. She gives some good info on indicator settings and continued calibration. In addition she gets into splitting certain cycles for some indicators. I think one of her more helpful suggestions is to use different settings for the same indicators depending if you are looking to BUY or SELL. This being a factor of the number of bars generally for a rally as opposed to a faster correction. I believe this to be a fairly thought provoking book for the seasoned trader. Good Luck Nutsneal
Nothing beats RSI(2)... NOTHING!! RSI(2) < 1 on the daily/weekly chart signals bottoms. If followed by POSITIVE DIVERGENCE, load the boat. RSI(2) > 99 on the daily/weekly chart signals tops. If followed by NEGATIVE DIVERGENCE, get your shorts on. Notice the attached chart. On July 21, the QQQQ bottomed out and the RSI(2) showed POSITIVE DIVERGENGE... a lower price made a higher RSI(2) reading. Had you loaded the boat, you could have sold for over $3 profit on the stock. Think QQQQ CALLS/PUTS!!