what rsi setting is best ?

What are you looking for the RSI indicator to 'tell' you? I have it on my charts simply as a re-confirmation of all my confirming signals/indicators; macd, cci, stochastic, rsi, 3/4 moving and exponetial averages on candlestick chart... all these lagging and leading indicators do (IMHO) is help to confirm/predict where the price action will be/is at...
FWIW mine is set on 14, irrespective of whether I'm trading pairs or indices, and I 'use' it to re-confirm stochastic movement on a 15 min chart when I'm entering a trade. Occasionally it 'works' as a leading indicator (for me) but more often than not simply as part of a series helping me to pull the trigger...

Tony,
Is this view confined to RSI or does it apply to all indicators? If not, which ones pass muster and why?
Cheers,
Tim.

Price is the primary indicator, all others must be derivatives from price against time. so decide market mode and your entry/exit TA from price alone. MAs can help confirm trend, but if trend wasn't that clear without an MA to help, it probably isn't there yet.

RSI and a few other indicators can help verify turning points e.g. when you see them diverging from price trend, or when indicators reach an extreme value. These events cannot be used as entry signals (well, they can, but not with consistent profitability) but can suggest how you might size your position to limit risk, or possible stop-loss distances, that sort of thing.

The value of RSI is a bit of a blind alley. Find one, stick with it, make sure it did actually reflect market behaviour and extreme market points. 20 is commonly used and pretty good for swing trading.

e.g. See 20RSI for the FTSE100 09/03, a significant low in the market, but porice alone didn't tell us that. 20RSI was at 21.052, its lowest level for over 2 years. From my Sharescope charts back to January 1994, this indicator has only been lower than that during 4 other episodes in 15 years - 1 day in May 2006, 3 in January 2003, 2 in June 2002, and 8 in September-October 1994. 5 episodes in 15 years (and, neatly, a total of 15 days) so this is on average a once-every-three-years event. Clearly, a significant point had been reached, and price rose 25% from there. Whatever indicator you use, if it didn't show that date as significant, change it.

RSI when used on a index, composite or sector for comparison with the instrument being traded makes some sense, but few use it this way. The vast majority just stick it on a single instrument and play the same settings and levels and use the textbook interpretation.

Even with this very limited application it’s still possible to get some mileage from it if you use it primarily as a divergence indicator rather than as a vanilla OB/OS indicator.

Cardwell, Brown et al have developed rather more useful techniques in the interpretation of the indicator when considering a Bull and Bear move as different animals (LOL) and selecting an appropriate differential for each. Additionally, further work by Cardwell of positive and negative reversals of this indicator warrant some study.

I’m not for or against indicators per se, just irritated buy the blind implementation of the set of common indicators using the standard settings and parameters and the standard textbook interpretations. Irritated is probably the wrong word – please would be a better one. At least you can see everyone coming…

I have always maintained that deconstruction of an indicator to find out what it’s ‘really’ doing is the best way to understand price action. Then, use it if you wish, but with full knowledge of what it’s doing and how and what its limitations are. Somebody mentioned recently (was is steeley?) about the MACD and how the majority misinterpret decreasing size f the histogram. A case in point.

Dear BamBam,

The in-depth RSI work initiated as far as I can tell by Cardwell is not a complete piece of work so one should use knowledge and some creative mindset to be able to walk the rest of the lane. The by Brown suggested solution which caused an internet crusade by Cardwell is in my view not a solution but more something IT people would call a “work-around”. When studying the many “work-around” (buying one book was enough Brown subsidizing) interpretations around I never could find the one that exactly looked like the examples in the book. So I have created my own thing that does exactly what Brown intended to do when she introduced her solution.

Have Fun.

For a few more words a "work around"picture.

: Megamuel you really are a diamond,

that website was exactly what I was looking for. I Actually have Bulkowski's book .The encyclopedia of chart patterns ,and if anyone is into pattern trading its a "must read book" , I work 48 hours between mon and friday and get to trade on friday as a regular stint witch leaves little time to do all the research required on the markets , I know some people say "backtest it your self"but with limited research time having you guys help point me in the right directionis a big help, Thanks again hope the hangover isnt to severe ? Tell me,, what markets do you trade ?

Ive been using it set on 20 day bar look back on 5 min and daily also 30 and 15 min chart ,it helps in conjuction with other criteria ,I wouldnt personally use it as a sole indicator to unreliable. I like to trade divergrnce as well and use the stoch,etc for that and was looking for a more divergence sensetive setting yeah I know I can backtest all these things but with limited resaerch time a point in the right direction can save time searching through a million websites of sh** and luckily Megamuel pointed me straight to a webite thats exactly what I was looking for

'OFF' is the setting you should be using for all indicators.