Markets Going Abnormal?
The behavior of the markets in 2008 forcedĂ‚Â traders to understand that markets can move much faster and much farther than they are “supposed” to. In order to avoid being caught long or short in a mechanical system trade during one of these abnormal periods, I’ve been working on abnormal filters for the systems.
An abnormal filter bascially just signals that the market environment is entering a period where it may become dislocated from what market participants would consider to be normal, based on historical standards.
Once the market enters one of these phases, it may be necessary to reduce postion sizes or even close out any position that is contrary to the filter (if the filter shows the market to be moving down abnormally fast, exit any long positions).
While I have not yet nailed down the exact parameters I intend to use in these filters, most variations of them are now showing that the market has already entered, or is close to entering an abnormal period. The previous time these filters would have signaled that danger was looming was in September of 2008.
Be careful out there!
Let me guess…. the filter’s showing markets moving down abnormally fast??
Well the down part is a gimmee.
The market can move down fast and not trigger the filter.
The market can move down far and not trigger the filter.
If it moves in one direction far enough and fast enough, yes, there could be trouble.
Wait, are you saying that we are/were in a ‘normal’ period? Or is there another state between normal & abnormal?
You should have been a nuclear physicist, given the amount of research that you like doing.
Will you be sharing the filters with us internet leeches? I could certainly make use of something that would tell me to throw the customary indicators aside and batten down the hatches.
Thanks
I think the markets are returning to “normal” for the macro environment.
I further think (with no real ability to prove it) that the hedgies were reducing the natural volatility.
Now that there are many of them blowing themselves up or dying on the vine due to redemptions, things are going to remain “abnormal” for a while.
Perhaps a good long while.
Wow, I was tired last night after the hold ‘em tournament. This post was vague and it sucked.
Yes, I plan on sharing some ideas for filters.
I don’t think the markets are ever normal. However, sometimes they are less normal than normal.
Think about applying volatility bands around the indices, maybe looking back between 10-100 days, depending on what your goals are.
If the markets have NOT been volatile for many many days, those bands will eventually shrink and tighten. When this happens, the index will not have to move very fast or very far before trading out of the bands. So while the bands may be telling you that something different is happening, what may actually be happening is the the index is resuming a more normal level of volatility.
However, if the markets have been volatile, and the bands have not shrunk tightly around the indices, and the indices can still move far enough and fast enough to break out of the bands, it may be something more abnormal.
Still vague and fairly non-helpful. I’ll try to put something quantifiable up in the future.
An abnormal filter bascially just signals that the market environment is entering a period where it may become dislocated from what market participants would consider to be normal, based on historical standards.
Are you still trying to codify a Black Swan?
How about a ratio of distance over volatility?
Meaning, the farther one is from a volatile period the greater the likelihood of a dislocation?
A rough cut (off the cuff) would be to take the mean and standard distribution of the distance between “highly volatile” days as the bands.
Example: For 2008, the average distance between days when the high/low range was over 200 points was 6 with a standard deviation of 1.2.
This would signify that between 3.6 and 8.4 days would be your distance with the likelihood being .95 @ 8.4 days and .05 @ 3.6 days.
BTW, I made those figures up for the example and there is a serious flaw with that idea in that often times there are days back to back when the volatility is extreme so you might want to start counting on the first day that the volatility is < 200 points.
Hope that made sense. It did in my head before I tried writing it out.
Cuervo, ratio of distance over volatility is interesting. I’ll have to ruminate upon that.
I’m not trying to code a Black Swan. These filters should ideally be the least restrictive possible, while not so loose that the only event they will catch is a Black Swan.
Look at the moves down in January and October. Think about a filter that would reduce size or turn off a system going into those declines while NOT turning the system off or reducing size during some of the swoons in 2007.
I would be interested in seeing that actually.