Neutral Set Up – August 4, 2011

[Pascal]

Pascal, Billy,

I am wondering if the market is changing.

We had a recent top in the IWM of 85.89 and most recent low is 74.88.

The DMF20 has not been able to pick up on this down move of more than 10%. That is either a sign that the move is not followed by big money, or the market is changed from the days DMF20 was developed. I note that the VIX was also late to the party, which would point to the first reason.

Considering your thoroughness I presume the 20 part of DMF was decided upon after an statistical optimization process. If you would do that optimization again over recent months - would 20 again be the optimal, or is a short average more effective.

Appreciate your insight,

Ernst

Ernst,


The 20DMF issued a short signal but it was orverruled because not confirmed by the inversed ETFs.
Just as a reminder, I modified the 20DMF rules I believe around september 2009, when I had to "hand-overrule" the 20DMF's short signals because I saw large players selling SDS. At that time, I decided to use an SDS confirmation in the programmed tools. Later on, I moved to an average of the four most traded ETFs, to have sometning that followed the global market more closely.

The question is: now that we are out of the POMO days again, isn't it better to respect a 20DMF short signal without waiting for a confirmation?

This is what i am studying right now. moving from a 20DMF to a 10DMF would shorten the cycles and maybe push us to overtrade. That however is another study that I could make later. I remember that I did some tests in 2009 and the 20D was what worked best, compared to 10D and 15D.


Pascal

[lulzasaur]

Hi Pascal,

going along those lines of thinking, I was wondering if it might be beneficial to have multiple robots instead of one that would rotate depending on what current market conditions were. For example, in turbulent markets such as this, we would switch to a robot that would be more sensitive with different rules; whereas in a "normal" market we would switch to a slower robot. An idea to measure whether or not we are in a "normal" market could be determined by a myriad of factors including but not limited to readings on the $tick, price volatility, etc etc. I think part of the problem with models (by other people) in the past is trying to put a "one size fits all" strategy which does not work very well.

[Pascal]

Hi Pascal,

going along those lines of thinking, I was wondering if it might be beneficial to have multiple robots instead of one that would rotate depending on what current market conditions were. For example, in turbulent markets such as this, we would switch to a robot that would be more sensitive with different rules; whereas in a "normal" market we would switch to a slower robot. An idea to measure whether or not we are in a "normal" market could be determined by a myriad of factors including but not limited to readings on the $tick, price volatility, etc etc. I think part of the problem with models (by other people) in the past is trying to put a "one size fits all" strategy which does not work very well.

Today's market action does not request specific explanation. Money management and stop loss is what is important.
Even if the IWM robot is in a long and probably stopped out position today, the signal was very weak and let's hope that everyone followed our advises not to trade weak signals. The string of losses on the robot came from the 20DMF's miss on the short signal about 10 days ago. For now, the 20DMF is in oversold, waiting for a buy signal which will surely come, even if we are now in a sell-all situation.

What we might do in the future is run two separate versions of the robot: a "trade all" signals and a "trade strong signals" IWM robot.

I have however to study more on how to manage the last 20DMF short miss. There is nothing urgent here, as we are in a "buy wait" mode. The "buy" side works as before. Simply be aware that without QE3 there is little chance that this market comes back in a bull mode very quickly. But buyable opportunities and shortable bounces will come along the way.


Pascal

[Adriano]

...
I moved to an average of the four most traded ETFs, to have sometning that followed the global market more closely.
...
Pascal

Pascal, I was wondering if a weighted average of some sort would have given different results, instead of a pure average. Just an idea. At any rate, I fully trust you'll find the best solution.