RSI: Selecting the best period?

[trhudson]

Quote from EmotionBigEvel:


There are very few around that know how the RSI works! Most people think below 30 is oversold and above 70 is overbought, WRONG!!! You can better buy when it is above 70 and sell below 30!

And about what period you should use, the 14 and nothing else! The RSI isn't created to be used in a 5 period or a 2.

So please, no offence words about my lovely RSI anymore! You only say them because of your lack of knowledge! [/B]

Who said anything about Overbought at 70 or Oversold at 30?

With the 2 Period RSI, that's the 2 Period, not the 14 so long as you don't use extreme numbers, you can develop high probability trading situations.

As for knowledge, apparently I have opened my eyes and looked outside of what the 'books' have said, and discovered on my own, what actually will work and what will not work.

It is absolutely necessary to take the emotional bias out of the equation to discover what really works...

 

[oraclewizard77]

No indicator is perfect, nor do they work all the time. However, if you combine your interpretation of what you believe the indicator is showing after reviewing past trading charts of the instrument that you are interested in trading, and combine that with a little PA, then you might find something useful. Of course this is the difference between programing in a system and trading manually.

There are no automated RSI systems making $ millions. The systems that make $ millions everyday use the advantage of the computer to do what it does best which is crunch data quickly.

For example, a server is placed next to the NYSE server. The company that owns the server say Goldman tells it to buy as soon as price goes up 1 tick or 1 penny. Also, Goldman pays the NYSE to allow their server to have this information 1 second before anyone else.

So basically, the server sees buy orders coming in, and front runs the orders.

 

[1a2b3cppp]

Knowing ahead of time how the market is going to behave will help.

14 period will work fantastic in some conditions and will cause you to lose your money in others.

 

[pegasys1]

Having the period set to 1/2 of a "cycle" length that you are trying to trade. Ex. if it takes 50 bars on average to go from bottom, to top, to bottom, then a 25 period rsi will work best.

I use an adaptive RSI that I designed, as the cycle length is always changing. Then it becomes a game of discovering the correct period. In theory a perfect period will make the rsi signal at the same bar, or the one after it. Thats pretty damn good. I use fractals to determine the trendyness, or rangeyness of the market with the theory that more fractals, indicating swing highs and swing lows in the market in a range, more up down up down, which needs a smaller period if you are trading short term swings, or a longer period if you are trading trends. Less swing highs/lows means the market is trending, you have the same choices as with the range bound market.

I use one of each indicator, I trade both trends and short terms scalps inbetween. The trending indicator slows down during range not whipping you in and out of trades at the wrong side of the range (these are the "bad signals" from fixd period indicators), then speeds up instantly when the market breaks out, signaling the breakout with a 1 bar lag, opposed to several bars with a fixed period indicator. It is the opposite with the scalping indicator, it emphasizes the short term swings so u sell the tops of ranges and buy the bottoms.

Of course adaptive indicators aren't perfect, but they are IMO much better, some use a combination of volatility, volume, etc. One really good set of adaptive indicators i've found are the Ocean indicators. Almost prohibitively expensive tho, which is why I made my own. In the end price action is king.

 

[kut2k2]

Quote from pegasys1:

Having the period set to 1/2 of a "cycle" length that you are trying to trade. Ex. if it takes 50 bars on average to go from bottom, to top, to bottom, then a 25 period rsi will work best.

I use an adaptive RSI that I designed, as the cycle length is always changing. Then it becomes a game of discovering the correct period. In theory a perfect period will make the rsi signal at the same bar, or the one after it. Thats pretty damn good. I use fractals to determine the trendyness, or rangeyness of the market with the theory that more fractals, indicating swing highs and swing lows in the market in a range, more up down up down, which needs a smaller period if you are trading short term swings, or a longer period if you are trading trends. Less swing highs/lows means the market is trending, you have the same choices as with the range bound market.

I use one of each indicator, I trade both trends and short terms scalps inbetween. The trending indicator slows down during range not whipping you in and out of trades at the wrong side of the range (these are the "bad signals" from fixd period indicators), then speeds up instantly when the market breaks out, signaling the breakout with a 1 bar lag, opposed to several bars with a fixed period indicator. It is the opposite with the scalping indicator, it emphasizes the short term swings so u sell the tops of ranges and buy the bottoms.

Of course adaptive indicators aren't perfect, but they are IMO much better, some use a combination of volatility, volume, etc. One really good set of adaptive indicators i've found are the Ocean indicators. Almost prohibitively expensive tho, which is why I made my own. In the end price action is king.

Interesting. How would you rank the Ocean indicators against the much-cheaper Jurik adaptive indicators? Thanks in advance.

 

[pegasys1]

Similar, but superior. I don't have the newest versions, which are supposedly even better. I use Jurik in my own adaptive indicators, I try to have every input I can adaptive as accurately as possible.

Ocean has some great algos in there, complex stuff I could never write, way beyond just adaptive periods and volatility adjustments, tho it uses them too.

 

[kut2k2]

Quote from pegasys1:

Similar, but superior.

You're claiming the "Natural Moving Average" to be superior to the Jurik Moving Average. What is your basis for comparison? Thanks.

Quote from pegasys1:

I use Jurik in my own adaptive indicators, I try to have every input I can adaptive as accurately as possible.

Ocean has some great algos in there, complex stuff I could never write, way beyond just adaptive periods and volatility adjustments, tho it uses them too.

If you have the formulas, PM them to me. I can put them in a spreadsheet if nothing else and send that to you.

 

[pegasys1]

NMA is faster to react to trend change, but it doesn't get whipped around as much. Also ocean has some great occilators that stop at the 0 line during pullbacks, but not in trend change. How they do this I don't know, but they do.

Really they are just a guide, price action is the only thing you can really trade from IMO.

 

[kut2k2]

Quote from pegasys1:

NMA is faster to react to trend change, but it doesn't get whipped around as much. Also ocean has some great occilators that stop at the 0 line during pullbacks, but not in trend change. How they do this I don't know, but they do.

Really they are just a guide, price action is the only thing you can really trade from IMO.

Based on what I've read, the NMA has a fixed length parameter of 20 bars. How this beats a "conventional" adaptive MA with a flexible length is a mystery to me.

As far as price action being king, two things: indicators are based on price action so it's not like PA is being ignored, and a superior indicator will enhance PA. I've seen AMAs outperform raw price.