InvestorsHub Logo

JLS

Followers 62
Posts 7863
Boards Moderated 0
Alias Born 12/14/2004

JLS

Re: TREND1 post# 764

Thursday, 02/14/2008 3:46:01 PM

Thursday, February 14, 2008 3:46:01 PM

Post# of 2987
If you look at the last two months (down market) you will see it is a good atrategy to short the market when RSI(3) drops below 70 (or you anticipate that it will based on other indicators).

For longer trends, it may be useful to use RSI(3) to generate triggers to buy, as a commodity will maintain a high RSI throughout the trend. It may not provide a better trigger than the one I have already developed (using a combination of mid-channel and ATR as foundation). FWIW, my triggers are not just a simple use of an indicator -- they have a single indicator (or pair) as a foundation, but add lots of tweaks to make them work better and also to self-correct a bad call.

Also, FWIW, if you do an RSI calculation from scratch, you will see that it uses only one input parameter, always closing price. So it should almost always follow price trend. It will always follow closing-price trend. It can -- and that's where the divergence is created -- not follow prices; and that's when the price action has greater volatility with larger than normal swings in open, high, and low prices (notice I left out closing prices). Conventional RSI is also an exponential average, so it is distorted by most-recent prices, especially on long RSIs such as 14 days.

It might be interesting to create an RSI for the other three prices -- high, low, open. Using two or three days in the calculation without filtering the result, adds insignificant delay. Then it would be interestnig to create a new, unfiltered, ATR that uses these four RSIs, then compare the result with conventional ATR which uses only two day's worth of (all four) prices and applies filtering to that result.

http://tinyurl.com/2qu2vw