Technical Models:
The following
text was published as part of the comments I wrote on August 24, 2006. They are
reproduced here because they are worth referring to – especially when the going
gets rough.
My Comments after Thursday (August 24):
Let me digress
and talk about the model. I have not discussed the record of the model very
often, even though I have been publishing daily results for the past 5 months.
For 4 issues (DIA, AKAM, AAPL and GOOG), tabulations for the year 2006 are posted.
So when you look at the model’s record, what do you think? In fact, what do I
think? I know the model’s record is good but, until recently, I couldn’t be
specific. A few days ago, I came across a short article that is very
informative – especially, if one follows trends. The article is on the
‘tradingmarkets’ web site. The address of the article in their archives is:
Allow me to
summarize a few points from the article. The author, also a professor, reports
on 12 technical indicators that ‘Barchart’ tracks. He has applied the 12
indicators to look at SPY (the S&P500 ETF) which is one of my favorite
options trading tools. He has analyzed the signals for SPY as produced by the
12 indicators for 2 years (Aug 2004 – Aug 2006). The results are not what I
expected and, indeed, they are surprising.
Of the 12
indicators, only one made money in the past 2 years. Seven of the remaining
eleven actually lost more than 15 SPY points (that’s 150 S&P500 points
which at a conversion of 8 to 1 is equivalent to 1,200 Dow points). The
remaining 4 indicators lost between 0 and 15 SPY points. This overall
performance is incredibly poor.
One of the
indicators, that he highlights the results for, generated 49 moves in the 2
year span. Of these, only 8 moves were profitable. That means that the
indicator was wrong 84% of the time. It was right only 1 in 6 moves. The moves
averaged a length of 7 days. This is all quite a shock.
How does the
model stack up? Well, yesterday I reported that this year alone based on the
DIA, the model has been right 10 out of 12 moves (5 out of 6) and it has
accumulated a net profit of about 1,200 Dow points for 2006. The length of an
average DIA move was 12 days. The model’s record for DIA is comparable to that
for SPY. So the reality is that the model is substantially ahead of the
indicators on Barchart that were analyzed. I am relieved to hear that. Good
luck to all.