The logic of exits

How do you know when a trend has started? Ended? This forum is for discussions about trend indicators and signals.
Post Reply
Trader46
Contributing Member
Contributing Member
Posts: 7
Joined: Tue Mar 23, 2004 10:49 pm
Location: New Zealand
Contact:

The logic of exits

Post by Trader46 » Sun May 30, 2004 7:40 am

Tharp's book (Trade Your Way....) makes a good deal of exit signals and I find that a very persuasive argument, both on the logic and on my experience. As he says, there is a pretty big "cone of silence" about exits - my guess, because its so hard. I have spent many an anxious hour watching myself getting eaten to death from good strong profitable positions though and thus have spent a good while thinking about the logic of exits.

Thinking about this (and in terms of the general Turtle approach) it seems to me that the "minimum hit" you are going to take on exit is always likely to be at least 1 ATR if you exit on some signal that says a price is outside "the nosie", 2 ATR if you get stopped out by a risk control stop (as opposed to a "chosen" exit) or, even if you use some form of breakout exit (in the classic Turtle mould, say a 10 day or 20 day breakout) you will also take a hit of roughly this size - whether it be an "instant" breakout which is big, or more likely in my experience (in forex markets), the Chinese death by 1,000 cuts breakout.

Just on the logic rather than heavy back testing (which I am wary of because of curve fitting - see Kaufman, P. and numerous others), this is likely to be about 10% - 15% of the gross (i.e. before risk) profit.

If, as I have read in various places, even most strong trends do not last more than around (this is rule of thumb, not precision) 10 ATR then it is possible as a trade advances to figure roughly what might be left on the table versus what you lose by exiting. That is of course not classic turtle but seems to make sense - i.e. once the trade gets to the point where the certainty equivalent of what might be made by staying in is slightly less than the hit you take getting out, then get out.

I wonder if others have had a go at this type of logic.....

B :roll:

Hiramhon
Roundtable Fellow
Roundtable Fellow
Posts: 98
Joined: Fri May 09, 2003 12:45 am

Post by Hiramhon » Sun May 30, 2004 10:04 am

I have a different opinion. I feel that the value of testing is greater than the value of theorizing or opinion-collecting. Naturally I have tested this opinion. The test data, in my opinion, supports it. :P

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

Post by richard » Sun May 30, 2004 10:20 am

I agree that exits are hard. Here's what's hard about it for me. Say I have 3 positions. 2 are a bit down and 1 is up. The 2 that are down were trending up when I bought them but are either in a temporary reversal or else they were bought in a false breakout -- I can't tell at this point.

The winning position starts to backtrack. At a certain point I close out the winning position.

This formula guarantees that I will have mostly losing positions in my portfolio, offset partially (but not completely) by some small profits.

Alternatively, I don't take profits on the winning position and see it usually backtrack to nothing or even a loss.

I am trying to work this out. I have had some pretty significant paper profits that I have given back as they take awhile to build in a trend, but especially in the equity markets the trends reverse very very fast (an understatement).

I suppose it depends upon your timeframe and the volatility you are willing to put up with.

SL
Senior Member
Senior Member
Posts: 30
Joined: Tue Feb 10, 2004 7:02 pm
Location: New Zealand

The logic of exits

Post by SL » Sun May 30, 2004 11:58 pm

Richard,

This is a huge subject, one of which I have plenty of time for.

I get the impression from my own testing that exits cannot be looked at in isolation with the rest of the system. Exits in many respect from testing always seem to fit in the 'damned if you do and damned if you don't' category. To much and you give away to much profit, to little and you get stopped out all the time.

I have been playing around with a couple of the more tricky exit strategies I know of. One is an exit strategy that starts out wide say the lowest low of the last 30 days and pulls in gradually by two days as the trend climbs.

The other exit strategy I have been working on, is simply an exit that if the position moves 5 ATR, then pull the stop up 5 ATR's from the high of the day the position moved above 5 ATR. Thus you are trying to capture 5 ATR's once the position makes headway. Complex stops are a headache to program, but they look good if nothing else when graphed. I believe from my testing that these stops have some merit over some of the more plain vanilla stops at capturing the bigger moves.

Back to the not looking at things in isolation I mentioned earlier. An old and common rule popped in my head yesterday that had been lying dormant for a while. As is often the case you get consumed by the immediate task ahead and forget everything else you have read along the way. It touches on what you said in your post.

Here's the rule: If you have loosing position(s) and a new entry signal comes along, dump the losing position(s). That to me sounds like an exit strategy. It's not a stop-based exit, but it’s a rule about getting out when you are in. I'm not sure at the moment how I would program that one as yet in Trading Recipes, (which is what I use) to test this theory more rigorously. Perhaps Veritrader can do this?

As with the majority of cases there is a limited amount of capital in your account for margin etc. and you are faced with the question of stopping the bleeding on the losing positions and reutilising the capital resource elsewhere where you have an entry signal that may do more good than the present harm. Likewise if those previous positions get another signal they are back in the game minus costs and the drawdown previously acquired.

As I said, this is a huge subject.


Stephen

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

Post by richard » Fri Jun 25, 2004 11:28 pm

I have been using a discretionary system that seems to be working for swing trades, a few days to several weeks. But the exit is still an issue for me.

I am trading futures charts that show good solid support and I am buying into resistance, where the stop can be placed closely and the expectancy is 3:1 or better. Or I sell from resistance into support, same deal.

So this system uses a (gasp) price target. But what happens when things get close and then retrace, as they frequently do? Do I close out the trade intraday and take a smaller profit, wait and have things retrace 50%, 75% or more? It is a puzzlement.

p8
Senior Member
Senior Member
Posts: 33
Joined: Mon May 12, 2003 2:26 pm
Location: Santa Clara Valley, California

Post by p8 » Sat Jun 26, 2004 12:54 am

Great topic. I used to work very hard to search for an "optimal" exit strategy which seemed to become a never ending quest. For now I settle with the idea that anything is fine as long as it gives me a positive expectancy.

So, my exit is "discretionary" in a sense that I sometimes get out at target, sometimes let it run with a profit stop, or sometimes do a combo (ie partial exit at target and let the remaning run). It all depends on the market "landscape" and, honestly, on my mood of the day. I tend to do the combo especially for daytrading, with a bias toward letting it run.

The only thing that is *not* discretionary is the requirement that every exit rule must have a positive expectancy. So, probably my current exit strategy does not give an optimal bottom line result, but it's optimal for my current trading psyche.

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

Post by richard » Sat Jun 26, 2004 10:59 am

p8 wrote:
The only thing that is *not* discretionary is the requirement that every exit rule must have a positive expectancy. So, probably my current exit strategy does not give an optimal bottom line result, but it's optimal for my current trading psyche.
Could you elaborate on this? Wouldn't any exit have a positive expectancy provided it was not at a loss or isn't a scratch?

p8
Senior Member
Senior Member
Posts: 33
Joined: Mon May 12, 2003 2:26 pm
Location: Santa Clara Valley, California

Post by p8 » Sun Jun 27, 2004 12:14 am

richard wrote: Could you elaborate on this? Wouldn't any exit have a positive expectancy provided it was not at a loss or isn't a scratch?
Expectancy gives you an idea how much money you can expect your trade to make on the average per every dollar risked. You can use your past trade results (sample size) to calculate expectancy. The larger the sample size the better. So, naturally you can have a few losing trades in a row and still have a positive expectancy. In fact, you could have less than 50% win rate and still have a positive expectancy (eg a Turtle-type method). On the other hand, it is also possible to have a method with an 80% win rate but has a negative expectancy.

Tharp gives fairly good explanations on this expectancy stuff in his books.

raymcboyd
Contributor
Contributor
Posts: 3
Joined: Tue Jun 22, 2004 12:50 pm
Location: UK

the logic of exits

Post by raymcboyd » Mon Jun 28, 2004 6:53 pm

I have done numerous tests which confirm that exits are more critical to profitability than entries.

That having been said, I also believe that there is no magic bullet. Trend momentum varies greatly. So where an ATR-based exit will work just fine in one instance, I'll have given up a substantial part of the trend in the next.

Same goes for profit-based, time-based, retracement-based, etc exits.

In the final analysis, one has to be happy with what the market gives with whatever exit one uses. Sometimes, you get most of the trend, other times not.

It's part of life on earth.

Ray

Post Reply