Thursday, September 07, 2006

Why Is It So Difficult to Change Time Frames?

When professional traders find that the patterns they've been trading no longer confer an edge in the marketplace, they try to make changes. Typically, they either try to change their time frame (go from shorter-term trading to a longer timeframe), change the patterns they look at, or change the markets they trade.

As I mentioned recently on my Trader Performance page, all such changes are difficult. That is because the changes entail unlearning old patterns--not just acquiring new ones.

In all my work with traders, I would say that the most difficult transition they've attempted is a transition from a short time frame to a longer one within the same trading instrument.

To trade a longer time frame means that you must let winners ride longer than you ordinarily would. It also means tolerating larger drawdowns and living with uncertainty longer. The instincts that help a scalper--getting out of losers quickly, for example--wreak havoc with efforts to trade longer time frames.

Once you have a "feel" for a market--which, really, is also a feel for a time frame--it is very difficult to unlearn that feel and develop a new one. The majority of traders, I find, do not navigate that change successfully. The rate of success is greater in applying the old time frame to a new instrument than in finding a new time frame in a familiar market.

Perhaps, however, there's another reason that it's so difficult to change time frames: Patterns that exist on an intraday time frame may be very different from those over longer periods.

Here's an example. As I mentioned recently in the Weblog and my market update yesterday, I've been playing with a trade setup in the Odds Maker program from Trade Ideas that involves buying pullbacks from upthrusts and selling bounces from downthrusts. The pattern tests out quite well on intraday data across a variety of stocks and markets. When I tested a similar pullback pattern over multiday periods, however, the buy setup actually yielded a negative edge!

Specifically, I went back to 2004 and looked at four-day periods in SPY that rose more than 1.5%. I then identified four-day periods that retraced a portion of that gain, but not all of it. Over the following four days, SPY did not resume its upthrust. Indeed, the average change over the next four days was -.30% (19 up, 19 down), weaker than the average four-day change over the 2004-2006 period.

Could it be that markets are more mean-reverting over short time frames than over longer periods? Might this help to explain why the transition from daytrading to swing trading is so difficult for many traders? As we extend the time frame, we go from markets in which locals are dominant to ones in which institutions determine the big moves.

Perhaps who dominates markets determines how markets move.

Changing from intraday to multiday trading might entail shifting from thinking like locals to thinking like institutions.

And that's very difficult for market makers.