Profit Target exits improve this system's performance

Discussions about the testing and simulation of mechanical trading systems using historical data and other methods. Trading Blox Customers should post Trading Blox specific questions in the Customer Support forum.
rabidric
Roundtable Knight
Roundtable Knight
Posts: 243
Joined: Mon Jan 09, 2006 7:45 am

Post by rabidric » Mon Feb 28, 2011 3:16 am

cliffg wrote: Seems like when the rate of change in p/l is rapid a chandelier reacts quicker. Is there such a thing as a dynamic chandelier?
Yes. Have a think about donchian channels. By fixing time sensitivity, they are dynamically sensitive to volatility. If you view the atr parameter of chandelier as more of a scale parameter, then you see volatility is pegged but time sensitivity is dynamic. Chandelier is still sort of a halfway house though. True fixed price sensitivity would be more like a point and figure "bar".

Whilst chandelier sounds good in that it trails quickly behind a fast break out, in the long run donchian comes out better in most tests. Less degrees of freedom is a small part of the answer. You can figure out the rest for yourself.
8)
[edited for spelling]
Last edited by rabidric on Mon Feb 28, 2011 11:50 am, edited 1 time in total.

Eventhorizon
Roundtable Knight
Roundtable Knight
Posts: 229
Joined: Thu Jul 08, 2010 2:36 pm
Location: Boulder, CO
Contact:

Post by Eventhorizon » Mon Feb 28, 2011 11:20 am

stopsareforwimps ...

Google + "Hakansson" gets me stuff about cosmetics and some steel mill dude!

Google + Hakansson + Bermuda Triangle gets me your post! LOL.

I am curious to follow up, can you be more specific about Hakansson, or am I being dense (it happens often)?

stopsareforwimps
Roundtable Knight
Roundtable Knight
Posts: 199
Joined: Sun Oct 10, 2010 1:47 am
Location: Melbourne Australia

Hakansson

Post by stopsareforwimps » Mon Feb 28, 2011 4:04 pm

Eventhorizon wrote:stopsareforwimps ...

Google + "Hakansson" gets me stuff about cosmetics and some steel mill dude!

Google + Hakansson + Bermuda Triangle gets me your post! LOL.

I am curious to follow up, can you be more specific about Hakansson, or am I being dense (it happens often)?
Hakansson pointed out the important fact that Markowicz single period optimization is not optimal for longer term investing. For example it ignores volatility drag, as recently discussed here.

Unfortunately, while his work was path-breaking and useful, he attacked some sacred cows in the theory of finance, and made some errors in his papers. This resulted in an academic feeding frenzy at his expense. Partly as a result, multi-period optimization has been shamefully neglected in academia since.

Full details in "Investing by the numbers" by Jarrod Wilcox chapter 6. Also

Nils Hakansson "Multi-period mean-variance analysis: Toward a general theory of portfolio choice" Journal of Finance (1971) pp 857-884.

Richard Roll "Evidence on the 'growth-optimum' model" Journal of Finance (1973) pp 551-567

It is more important to be roughly right than precisely wrong, except in academia.

SimJimons
Senior Member
Senior Member
Posts: 31
Joined: Mon Jun 20, 2011 9:59 am

Post by SimJimons » Fri Jul 01, 2011 10:12 am

Hi guys, great forum with a lot of interesting topics!

It's been very interesting to read your thoughts on the "target exit topic", something I've had a very painful experience with. That is, in relative terms. We run a fairly large CTA in the Nordic region which up until last year showed relatively strong numbers compared to many of our competitors, a fact that slowely allowed our sales people to gain traction with institutional clients. However, early 2010 we introduced a new set of models in our portfolio, of which about 40-50% carried some form of target related exit. We felt pretty confident that they wouldn't be activated at the same time because they were so distant in time-frame and profit level. However, that was before the mother of all bulls showed up in bonds, effectively triggering profit taking in most of the systems...before the last leg of the rally began. Hence, we were running at half speed when most of our competitors were running at full throttle. We left about $100 million of our clients money on the table, basically giving up (and then some) the slight edge we had created over competition, in two months! For those of you that haven't really felt, or bothered about, relative pain I can tell you that it hurts almost as much as absolute pain :) I would imagine that most people don't really pay attention to the above issues when simulating historic performance of a system. However, in the real world they may actually be of greater importance than the Sharpe.

A second issue I have with target exits (that is in a trendfollowing context, not when it comes to mean reverting stuff) is the fact that we introduce one or more degrees of freedom, everything else being equal. This obviously means that we should be able to extract higher Sharpe ratios in a backtest (we did, but actually diversification was our key motivator, anyway). But this doesn't necessarily mean that they will repeat in real life, or actually it's very likely that they will not. Hence, a backtest doesn't really tell us if the system has merit or not, if we don't adjust for the increase in datamining (sorry if I'm stating the obvious).

Lastly, which again has to do with datamining, is the extreme sensitivity to certain settings. Let's say that we use ATR to take profit and the level we use is 3.00. For obvious reasons a system like the one above will perform vastly different if one or more trades reach 2.99 ATRs versus 3.01 ATRs. Hence, a very slight change may render our system more or less useless, or at least the possibility exists. Compare this with the much less curve-fit moving average system. In that case it doesn't matter much if we pick 299, 300 or 301 days as the lookback (under most circumstances).

Just my 2 cents...

/Sim

AFJ Garner
Roundtable Knight
Roundtable Knight
Posts: 2040
Joined: Fri Apr 25, 2003 3:33 pm
Location: London
Contact:

Post by AFJ Garner » Fri Jul 01, 2011 10:31 am

SimJimons wrote: We run a fairly large CTA in the Nordic region
Lynx? Estlander?

LeviF
Roundtable Knight
Roundtable Knight
Posts: 1428
Joined: Mon Dec 22, 2003 12:24 pm
Location: Des Moines, IA
Contact:

Post by LeviF » Fri Jul 01, 2011 11:33 am

I appreciate your input, but one recent example hardly makes a case against using profit targets. What would you be saying if bonds reversed shortly after hitting your targets and your competitors went into steep draw downs?

longmemory
Roundtable Knight
Roundtable Knight
Posts: 128
Joined: Tue Apr 12, 2011 11:46 am

Post by longmemory » Fri Jul 01, 2011 12:50 pm

Levi wrote: "For me, the conflict is that getting out of 99% of the position at 1R (leaving 1% in so the system doesnt jump back in the trade) is better than letting it ride. Again, this goes against all the theory of why LTTF works: returns arent normally distributed - you need to capture the 20+R fat tails, most people arent disciplined enough to cut losers and run winners or deal with lots of losing trades, yada yada. So then what is my edge?..."

Levi, look at it this way: What matters is NOT the shape of the distribution PDF (Probability Density Function) but the the CDF (Commutative Density Function). 'Eyeballing' the PDF is misleading. Need to look at the CDF.

How to prove this to yourself: write out the % daily change for every trade of a NON-compounded back test. Integrate the positive daily changes and negative daily changes. When you move to the RIGHT of the CDF, the CDF Y value will continue to rise in response to your 10 bagger or 20 baggers, but, this is the key. The rate of change will TAPER OFF. The Law of Diminishing Returns.

Now integrate the negative surface area and, lo & behold, the CDF rate of change goes through an inflection point and the Y value at the right edge begins to rise MORE RAPIDLY at some point. (NOt a surprise since best fit line is generally Cubic or higher order polynomial.)

Now, compute the Reward to Risk Ratio in terms of the XY values of the CDFs.

Once you've computed the Ratio, plot it, and you will see the wisdom of Sluggo's advice.

L

SimJimons
Senior Member
Senior Member
Posts: 31
Joined: Mon Jun 20, 2011 9:59 am

Post by SimJimons » Fri Jul 01, 2011 5:32 pm

AFJ Garner: I'm afraid I've already said too much, sorry...

LeviF: Obviously I'm not suggesting that our less fortunate outcome last year provides proof that profit targets (in a trendfollowing context) don't work. What I was trying to illustrate was that even if they do work in the long run, delivering a higher Sharpe, they may hurt your business in a big way, at least temporarily.

However, I do believe that my other statements have merit, and may help explain why profit targets can be a dangerous road to take if you are a trendfollower. In addition, even if you can control for datamining, etc, it is highly likely that the higher Sharpe you experience is nothing but a mirage, appearing because of the change in the distribution of returns. As we all know the Sharpe is only a meaningful measure of risk adjusted return if returns are normally distributed (i.i.d), which they clearly are not. And if this is the case we have to look at moments beyond mean and variance, i.e. skew and kurtosis. And profit targets will have a worsening impact on both. Hence, what you may view as a superior system looking from the right may actually be inferior if you look from both sides.

Finally, if you evaluate a system from a portfolio enhancement point of view it is my belief that you will find a non-profit target system more diversifying, in general. But then again, we all have our own reasons for coming up with our systems, meaning that there probably isn't a right or wrong, it's more about tastes...

/Sim

sluggo
Roundtable Knight
Roundtable Knight
Posts: 2986
Joined: Fri Jun 11, 2004 2:50 pm

Post by sluggo » Sat Jul 02, 2011 8:21 am

Profit Target exits chop off the right hand fat tail, thus they reduce kurtosis of individual trade outcomes.

PT exits also reduce variance of individual trade outcomes, a result which Ralph Vince has been praising for more than 20 years in his "Fundamental Equation of Trading".

Whether Profit Target exits increase or decrease skew of individual trade outcomes, is not so clear. PT exits transform some big winning trades into medium winning trades (by exiting early at the target), which suggests smaller positive skew. However PT exits also transform some LOSING trades into medium winning trades (Trades that achieve the target then reverse and become losses), which suggest larger positive skew.

LeviF
Roundtable Knight
Roundtable Knight
Posts: 1428
Joined: Mon Dec 22, 2003 12:24 pm
Location: Des Moines, IA
Contact:

Post by LeviF » Sat Jul 02, 2011 9:02 am

Dont forget you dont have to get out of your entire position at a profit target. You can leave in 1% to 99% of your original size.

Chelonia
Roundtable Knight
Roundtable Knight
Posts: 497
Joined: Mon Apr 30, 2007 3:37 pm

Post by Chelonia » Sun Jul 03, 2011 3:08 pm

Depending on how much contracts your unit exists of trading futures.

SimJimons
Senior Member
Senior Member
Posts: 31
Joined: Mon Jun 20, 2011 9:59 am

Post by SimJimons » Mon Jul 04, 2011 4:30 am

Sluggo, I meant that the combined effect on the portfolio likely will be worse if you take higher moments into account, my bad. However, I have a problem seeing how PTs (and I don't mean Personal Trainers :) ) can increase skew?! Yes, they may turn losers into winners, but constraining the right hand tail will not increase skew, as far as I can see. Anyway...

LeviF, while it's true that you don't have to exit your entire position, I fail to see the logic behind reducing a position at all if the signal to do it is flaw?! That is, if you believe it's flaw. But I guess what you mean is that if you like your signals equally much it is prudent to act on both from a diversification point of view?!

/Sim

stopsareforwimps
Roundtable Knight
Roundtable Knight
Posts: 199
Joined: Sun Oct 10, 2010 1:47 am
Location: Melbourne Australia

Post by stopsareforwimps » Mon Jul 04, 2011 5:40 pm

[quote="SimJimons"]... constraining the right hand tail will not increase skew, as far as I can see. Anyway...
/Sim[/quote]

Positive skew is a good thing. I don't think you would want to reduce that. In fact that is the whole rationale of letting winners run and cutting losers short - it increases skew.

I suspect the idea of profit targets is that once a trade has run a long way it is likely to go strongly in the other direction at some point. By having profit targets you may make yourself less likely to suffer these massive reversals. Such reversals produce negative skew.

Similar techniques include detecting parabolic moves, detecting Sornette type bubbles, and shortening time parameters when markets have moved a long way in a short time.

File under "picking the top/bottom".

LeviF
Roundtable Knight
Roundtable Knight
Posts: 1428
Joined: Mon Dec 22, 2003 12:24 pm
Location: Des Moines, IA
Contact:

Post by LeviF » Mon Jul 04, 2011 6:19 pm

Keep in mind if you hold a position a very long time, if your equity has been increasing and you size based on x% of equity, the old position will have less of an impact on your equity than a more recent entry.

Chelonia
Roundtable Knight
Roundtable Knight
Posts: 497
Joined: Mon Apr 30, 2007 3:37 pm

Post by Chelonia » Tue Jul 05, 2011 2:24 am

I think that´s true if volatility remains more or less the same, if however volatility increases substantially the impact of that will neutralize the (dis)advantage you thought you had on x% against your higher equity.
Last edited by Chelonia on Tue Jul 05, 2011 2:47 am, edited 1 time in total.

rabidric
Roundtable Knight
Roundtable Knight
Posts: 243
Joined: Mon Jan 09, 2006 7:45 am

Post by rabidric » Tue Jul 05, 2011 2:37 am

which is why shorts after blowouts are so depressing-> the declining volatility works against you and the slow collapse inevitably lasts so long you end up with pathetic delta exposure compared to newer positions in vega growth contracts.

one would think therefore that playing around with position-size re-normalization would be the solution. It appears good in principle theory, but in practice it has it's own issues to do with unwanted side-effects that compromise your leverage adjusted CAGR, i find. Some forum-users seem to have success with frequent rebalancing(in order to maintain constant portfolio heat). I personally have found that the increased complexity is not really that beneficial, though i think there is something to be said for simple rebalancing e.g. quarterly or annual.

The reason i bring up the rebalancing thing is that it is actually a parallel idea to Prf Tgts , but expressed in a wider portfolio manner. targets and rebalancing are both about the same thing:

existing position exposure management.

SimJimons
Senior Member
Senior Member
Posts: 31
Joined: Mon Jun 20, 2011 9:59 am

Post by SimJimons » Tue Jul 05, 2011 3:07 am

Stopsareforwimps, I think you misread my post! I was referring to sluggo's reply in which he was saying that targets may increase skew (which is a good thing, yes). I had a problem seeing that they would, however. Positive skew is definitely good, which is why I don't like profit targets in trendfollowing strategies.

Obviously it is very attractive to be able to predict reversals. But since trends tend to be longer than what one would expect from chance, timing reversals is a difficult task. Hence, you may leave a lot of money on the table, and if you do that more often than not you are worse off. In addition, if you believe you have an edge in the targets you set you shouldn't settle for neutralizing your position, you should reverse it (from long to short, for example)!

What I really don't like with profit targets is that they have no natural way of re-entering a trade. So, instead of using a profit target in a trend system I would rather use a countertrend system as an overlay, letting each system live its own life. May seem as semantics, but it helps you to really drill down on the reversal signal's merit, while alleviating the re-entry issue.

LeviF, we re-size positions every day (more or less, after adjusting for noise) so old vs new position isn't really an issue. However, for managers that can't do this because of size of AuM, for example, you may have a point.

/Sim

stopsareforwimps
Roundtable Knight
Roundtable Knight
Posts: 199
Joined: Sun Oct 10, 2010 1:47 am
Location: Melbourne Australia

Post by stopsareforwimps » Tue Jul 05, 2011 4:45 am

SimJimons wrote:Stopsareforwimps, I think you misread my post!

/Sim
Yes - now I see. In my defence I had been coding all day.

I think we agree.

I am attracted also to the idea someone expressed that profit targets are another parameter in your system, thus an opportunity to curve-fit the past.

Tim Josling

liaojing
Contributor
Contributor
Posts: 3
Joined: Tue Mar 08, 2011 3:23 pm

Post by liaojing » Wed Jul 06, 2011 12:23 pm

Maybe tightening the trailing stop can help? On the one hand, tightening trailing stop can help protect more open profit, on the other hand, it still give trend some room to develop. And after stopped out, if the trend resume, it can easily reenter. So maybe tightening trailing stop is better than profit target.
Take simple channel breakout for example (80-day high enter, 20-day low exit), suppose its original exit is 20-day low. Now change it to after make 3 times ATR profit, use 15-day low as exit, after make 6 times ATR, use 10-day low as exit. It seems this stop tightening can improve MAR.

rabidric
Roundtable Knight
Roundtable Knight
Posts: 243
Joined: Mon Jan 09, 2006 7:45 am

Post by rabidric » Thu Jul 07, 2011 7:18 am

good first post liaojing, welcome to the forums.

The thing with your idea, is that it basically acts like a "pseudo tgt" anyway. i.e. by tightening the trailing stop, you end up tripping over the typical volatility at a small pullback, which often turns out to be a poor location.

It sounds nice in theory, and some backtests may show it to be ok, but it isn't robust really. Your basic trailing stop is as big as it is for a reason. Tinkering with it too much is counterproductive 9/10 times. The standard target also has the advantage that it triggers on favourable price movement, rather than unfavourable price movement. Within a portfolio this has anti-correlated effects that the others have been discussing all thread as you have read.

Post Reply