Mode Switching

How do you know when a trend has started? Ended? This forum is for discussions about trend indicators and signals.
Post Reply
Roscoe
Roundtable Knight
Roundtable Knight
Posts: 250
Joined: Sat Jan 24, 2004 2:06 am
Location: Houston TX

Mode Switching

Post by Roscoe » Thu May 06, 2004 8:43 pm

I keep coming back to this basic question: how can I tell if a market is in a trending mode or in a non-trending/sideways mode? If I could answer this question with any degree of reliability then I could match the system to the current market mode.

Would anyone care to offer suggestions on how I might approach defining market mode? My preferred timeframe is daily bars, but the concept would presumably be scaleable to shorter or longer timeframes.

ksberg
Roundtable Knight
Roundtable Knight
Posts: 208
Joined: Fri Jan 23, 2004 1:39 am
Location: San Diego

Eternal Question

Post by ksberg » Thu May 06, 2004 9:50 pm

Ah Ross, the eternal question! If I only knew the mode :!:

Lot's of people have tried, Wells Wilder with ADX, Adam White with Vertical Horizontal Filter (VHF), even Sweeney in Campaign Trading.

I noted a link on the forum to papers by Charlie Wright, one of which bears some pertinence here Markets, Strategies, and Time Frames. He talks about 3 modes: trending, directionless, and volatile markets. I bring this up to note that maybe, just maybe two modes aren't sufficient to capture the different kinds of price action that foils mechanical systems.

Worth a thought.

Kevin

Forum Mgmnt
Roundtable Knight
Roundtable Knight
Posts: 1842
Joined: Tue Apr 15, 2003 11:02 am
Contact:

Post by Forum Mgmnt » Fri May 07, 2004 6:23 am

:twisted:

You can tell if a market is in a trending mode when your trend-following systems are profitable and you are on equity highs.

If your trend-following systems are losing money and you are in a drawdown, then the market is consolidating.

That's the best way to tell what "mode" the market is in. Unfortunately, that's not what you really want to know. What you really want to know is what the future mode will be. A much harder question, and one that I don't have an answer to.

I haven't found a way to improve my trading using this line of reasoning.

One of the major problems is that great trends often, perhaps most often, follow periods of nasty consolidation. So if you aren't trading trends because the market is in a consolidation mode, you won't be in the market when they develop.

The best substitute I've found is simply to trade multiple systems at the same time, and systems that don't lose too much when the market is not in the ideal mode for that system. That way, no matter what mode the future brings, you will have some systems making good money and some losing a bit or breaking even.

- Forum Mgmnt

ksberg
Roundtable Knight
Roundtable Knight
Posts: 208
Joined: Fri Jan 23, 2004 1:39 am
Location: San Diego

Multiple Systems

Post by ksberg » Fri May 07, 2004 10:53 am

Forum Mgmnt,

Agreed, the "future mode" problem is precisely why ADX, VHF, and similar techniques work poorly: they let you know what you should have been trading in the recent past. Existing predictive modes in trading include Elliott wave, fibonacci retracements, cycles, and fourier/wavelet-related methods (several separate topics unto themselves!).

You made mention a while back about introducing a counter-trend system based on the performance of the trending system. I've been experimenting with trading the equity curve in this manner. So far I've learned that I have fewer counter-trend techniques in my toolbox :oops:, but otherwise it looks feasible.

I think there's also something to be said for implied structure. Turtle rules include one such assumption, the "Last Profitable Trade" rule. This assumes some consolidation may occur after a profitable run (which often happens). The point most people miss is that the trade really isn't skipped ... the rules get shifted to give more breathing room. So, some "mode" along these lines probably has a decent chance of succeeding.

Any thoughts?

Cheers,

Kevin

Jwebster
Roundtable Fellow
Roundtable Fellow
Posts: 52
Joined: Fri Jul 04, 2003 10:23 am
Location: London, UK

Post by Jwebster » Mon May 10, 2004 8:59 am

Roscoe,

Go to this site http://www.ta-lib.org/ and download their add-in for excel. Within that you'll find a function that computes a trendmode or cyclemode (trending/ranging) by indicting 1 (trending) or 0 (ranging). The function is based on the hilbert transform/MESA equations used by John Ehlers. You can use various prices for your input, but I believe the original price used was the median of the daily high/low. You enter the function as you would an array formula. Not the be all and end all but gives you a reference. See what you think!

Chris67
Roundtable Knight
Roundtable Knight
Posts: 1046
Joined: Tue Dec 16, 2003 2:12 pm
Location: London

Post by Chris67 » Tue May 11, 2004 7:18 am

Roscoe,

I agree with c.f. that the best trends develop in mkts that have not been profitable for a while so unless your in you wont know or be in when the trend occurs .. HOWEVER .. I also firmly believe in Ed Seykota's theory ..and it works real well for me ..

Take a daily chart going back one year .. stand 10 feet away from the chart and if from that distance nothing is screamingly obvious keep out of it ...

All the other mathematical theories are utter B/S .. if they werent then we all be millionaires by now ..

richard
Roundtable Fellow
Roundtable Fellow
Posts: 94
Joined: Tue May 18, 2004 10:17 pm

Post by richard » Tue May 18, 2004 10:20 pm

Hello, this is my first post here.

One assist if you are involved in equities is Dow Theory and its four phases of the market. This probably applies to other markets as well. There are two trendless phases and two trending phases of a market.

At my stage in thinking, it pays to be aware of fundamentals in order to deterimine which phase the market you are looking at is in.

If you study financial history and make yourself a student of the fundamentals, perhaps you can learn to gain a feel for the type of market you are considering?

Post Reply