Skip to content

To Fade or To Follow

June 3, 2009

In a general context, I would prefer an account to have multiple strategies based across the flavours[trend following, trend fading, dollar neutrals], across the time frames[hourly,daily,weekly,monthly(?)], across the markets[ equities,commodities,forex]. The upside of having this, is a more smoother equity curve. The logic is: by having strategies across all these categories you can expect the individual positions to be uncorrelated with the rest.

Moreover, if one system in one timeframe enters a drawdown, it an be reasoned that until unless some huge price shock occurs, the drawdown in one of the system-timeframes-markets [say Trend Following System in hourly time frame in Corn] is compensated by other system-timeframe-market [ S-T-M], thus giving a smooth equity curve.Which equals, less drawdowns, less duration of drawdown, less psychological pain and in general more returns.

The Three Axes

The Three Axes

Having said all this, we must remember life is a tradeoff. For an account manager nothing can be more satisfying than having high recovery rates. But what is the cons of this method: High amount of equity. Exceedingly high amount of equity, to devote to all the S-T-Ms.

Given all this, I want to add another dimension to smoothening the equity curve and that forms the basis of this post.

Often, very often, we come across scrips which have shown a huge rise, sometimes triggerring the trendfollowing systems only to turn out it to be  mean reverting. Its a whipsaw in plain terms. You are expecting a breakout and a followthrough only to find it give you a loss. Can we avoid it? Perhaps.

Additionally, for those who deal with systems, there is another headache.If a scrip undergoes a substantial move, in response to which both trend following as well as trend fading systems are triggered, then which one to take?

And here comes the fourth dimension. The dimension of confidence. Just an idea, what if, we indeed run a scrips recent move through a subroutine [only in case a mean reverting as well trend following trigger in the same scrip as a result of the same event], which decides the merit/confidence of both systems. Graduating from a pure buy-hold-sell paradigm to a the grayish world of  confidence is indeed a huge leap. But thinking in terms of a score, which will decide which system wins over can be helpful.

You can call it, filter but its not exactly that, anytime a trendfollowing trigger comes up, you can calculate its score. Anytime a mean reverting trigger comes up, theoretically atleast you can calculate its score. Think the score as the probability that a scrip will follow or revert. But unlike usual mathematics its not a pure “yes-no” solution. The sum of both scores will not be cumulative to 100%.Because if it is, the variables involved in the system will not be suited for either one of them. That is consider:
Trend Following Confidence Subroutine decides its confidence based on,

  • A variable A
  • A condition B
  • A state C

For the score of Mean Reverting Confidence subroutine to be cumulative and match up to 100, its decision making conditions should be also, A’, B’ and C’ [where A’,B’ and C’ are mutually exclusive to A,B,C: that is it will be either cloudy or sunny. Hence if today is sunny, it can’t be cloudy, and if its cloudy it cant be sunny== Mutually Exclusive. So if event A= sunny day, then A’= not sunny day(=cloudy day)]

So you see, we are effectively basing our decisions on three variables only,if we are to make it cumulative. Hence we can’t really decide the merit of the two cases equally, if we are to entail variables from only process. Can we?

Hence, the scores will not/ should not add upto 100%. We can only compare the confidence levels and not numerically process them.

Given this setup and the change we bring in our mechanisms and thought process, I think we will be doing something which the junta/crowd doesnt do. And as my thumbrule goes, if the crowd didnt do it, then you must do it. Because the crowd never wins.


Jump Up!

P.S: This is not purely continous scale in scale out as talked about by other bloggers in the sphere. I am still in the binary mode, but the decision making process has become more fuzzy.

2 Comments leave one →
  1. Eric permalink
    June 11, 2009 1:26 am

    Found my way over here from Michael Stocks blog Market Sci. I am checking out the futures for the Nifty 50 (symbol IN) that trades on the SGX and it looks like the volume is really low. As of today 6-10-2009 it looks like it has traded around 11,000 contracts. Is this normal or an abbe ration? I would like to trade it but need some more info and hoping that you could help.


    • June 11, 2009 9:22 am

      Hi Eric,
      11k contracts is a healthy number from SGX standards. Right now I am staring at 294 contracts as volume traded for 11thJune, 0932 IST [GMT-5.30].

      Hence I would suggest you to trade the Indian markets if possible directly. The catchword is “if possible”

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: