TA: Tricks and Traps
MM, I'm not necessary responding to your post, but as a newbie, I could not figure out how to post a message anew. I liked your 4th of July message, and so I picked you to respond to in order to post. It's nice to see someone like yourself with such a positive message.
I've been lurking here about a month or so and have been impressed with the good vibs and talent on this thread. FYI I am a pharmaceutical scientist with a good background in curve fitting and mathematics.
I've been a student of TA for about 5 years. About 3-4 years ago, I tested all the Morris TA indicators, and found none worked consistently. I also developed several systems myself, and none of these worked well. I also used a program that recommended buys and sells based on a vote of its indicators. Only one system in this program worked well, but it only gave a signal willy-nilly, making it un-useful.
One problem with TA most practitioners don't recognize is that many indicators (and all momentum oscillators) assume and require the stock is non-trending, i.e., it's price-time curve exhibits a sort of a pseudo sin wave. So no single TA momentum indicator will work well in a trending stock. For example, in an upward trending stock, stochastic will go up and stay there. It way dip below the 80% level, giving a false sell signal. This is a major problem, for if you are long, you want an up-trending stock. This is not to suggest or imply that a combination of TA indicators will not work. Suffice it to say, I could never develop a series of indicators that worked well. Others on this thread seem to have a combination of indicators that work.
Consequently, I became a chartist, using simple regression lines with 2 standard deviations for price over-extension and sell signals.
One indicator posted by a technical analyst tested best in my evaluation of all the systems. It gives a buy signal the day that both these conditions are simultaneously satisfied: 5-day EMA of OBV exceeds 21-day EMA of OBV; and 21-day EMA of OBV exceeds 55-day EMA. A sell signal is given when the oppostive is true.
Now I found my source of price data to be a problem (many people use them -- they advertise heavily), and I quit them. If you know someone who provides inexpensive data, I'd be interested. I don't like the TA systems of providers, especially the proprietary systems (you cannot test the indicator/system). I use the DOS version of Advance GET. My sell signal is usually a break below the bottom 2 SD regression trend line. So I'd like a provider whose data will work with the DOS verion, not the windows version.
With respect to the Elliott Wave principle, I found it quite faulty. For example, it was not uncommon to find a stock in a minor uptrend of a bear pattern switch the next day to a bull uptrend. I found you could get a variety of characterizations depending on what time-frame you set the system to. Elliott wave theory works well in retrospect, but I never sound it useful prospectively.
Now I am an avid believer in TA, but only used in combination with fundamental analysis. I like the fundamental system of Motely Fool. I hold two positions recommended by them: PLCM and CREE.
I'd appreciate help on a data provider -- not the one that is all over the pages of IBD). Also, would anyone be interested in running the OBV scan I noted above (I can't do it now) and report stocks that did not meet the criteria yesterday, but meet it today.
For those who disagree with me, that's great. As one great scientist said, a clash of doctrines is not a disaster, it's an opportunity.
If any of you need help with pharmaceutical drugs you may be taking, write me an e-mail (I'm a clinical pharmacologist). For example, if you are taking Zocor for cholesterol, you may be in danger if you drink grapefruit juice. This is no joke.
My post is running too long, so good-bye. I hope to hear from any of you on this thread or via email (shelbyb@ix.netcom.com). Best wishes. Harold Boxenbaum
|
