|
PureBytes Links
Trading Reference Links
|
On Sun, 18 Jul 1999 00:17:52 -0500 "charles meyer" <chmeyer@xxxxxxxx>
writes:
>From what you have written am I correct to infer that because your 20-day
coorelation study is currently above .9 the VIX is indicating that the
next directional move will be to the downside--at least for the short
term?
Charles,
the correlation idea was just a way of seeing if the VIX and OEX are in
sync with each other. The idea about when the VIX goes up(down) the OEX
goes down(up) gets confusing because it doesn't always work. To simplify
this I turned the VIX upside down so its relationship to the OEX is more
visual (the division by 10 was just to get it to scale with the rest of
the chart data). (The current actual 20 day correlation between the VIX
and OEX is minus .908, the inversion process turns it around to a
positive .908.) So what this all means is that the direction of the VIX
(up/down) will relate to a corresponding movement in the OEX (down/up) at
least for the short term, or as long as the correlation reading remains
high. I am not saying that this is indicating a move down on the OEX from
just looking at this correlation. But if you think that the VIX is going
to reverse from its current low level reading (and I do, but not for this
reason) then the correlation reading of .908 is telling you that using
the VIX to time the OEX is more reliable now then when the correlation
reading is neutral (.5) to negative (.5 or less). I do not believe the
idea that the VIX "cannot" tell you what direction the OEX will go when
it changes direction. If the correlation between the VIX and the OEX is
high and positive, you can determine with reasonable accuracy what the
direction of the OEX will be. If the correlation is neutral the VIX will
not help you to much. If it is negative, you might be looking at a move
in the opposite direction then what you think the VIX is predicting. (of
course this is all further complicated by the fact that I inverted the
VIX, otherwise all the negatives become positives and so on). I find
these correlation studies very useful in determining what inter market
relationships are strong, weak, or contrary indicators. An idea that I
have been working on for awhile is to use a correlation matrix of various
markets and indicators to help in finding more reliable trading signals
(ie: a stochastic was working real well a few months ago, but now a
moving average crossover is getting better results, only I have lost to
much money by the time I found out, maybe you have been here before) I
also wanted to add that the 20 day time period was arbitrary. The time
period issue is also subject to change. One way to address this is with
the use of "auto-correlations". I have also had success with using
Rescaled Range Analysis to determine useful trading time periods. Hope
this answers some of your questions.
Ron McEwan
|