EXIT signals ...

How do you know when a trend has started? Ended? This forum is for discussions about trend indicators and signals.
Robert Nio
Senior Member
Senior Member
Posts: 41
Joined: Thu Apr 17, 2003 11:34 am
Location: Palo Alto, California

EXIT signals ...

Post by Robert Nio »

Hi There,

I think I have a unique problem. I have an outstanding entry signal generator finding me the best trends in the making and keeping me away from "non-movers" BUT I am missing a good exit strategy.

Now I wonder if anyone of you did any research in this area and has some empirical proof that the turtle rules are good enough to indicate a folding trend or maybe found other good indicators for a stopping trend.

Thanks for any feedback.

Robert
damian
Roundtable Knight
Roundtable Knight
Posts: 815
Joined: Tue Apr 15, 2003 8:43 pm
Location: dusseldorf

Post by damian »

g'day Robert.

Not quite an answer to your question, but you may find something of use in the rather long spiel that follows.

I am not one for ‘inventing’ new indicators, but sometimes I get bored and come up with some strange and mostly useless ways of representing info that is already contained in the price data itself.

Anyhow, here is one indicator that is far from perfect and I never bothered to take it any further. The maths is mostly redundant and the whole mess can be boiled down to a far more basic set of operations. Please do so, sorry for being lazy.

I came up with this indicator some time ago when I developed another entry/exit system similar to Aberration in concept. I discussed the original system on a web forum in Australia perhaps 2 years ago. I think I may recognise one or two people on this forum from that old forum. They may recognise this part of what follows. Due to laziness (again) I will put the code here in Behold! Format. It should be pretty simple to decipher. Please ask Q’s if needed, I am not that lazy.

I final point. By no means do I claim that this set of indicators is amazing or even that useful. I just noted that it gave the ability to see when a trend was strong and when it was weakening. It can be improved upon 100 times over. I recall that the system did test ok on a 20 market portfolio using fixed fractional position size.


Indicators

high average:
HAV = @average((@highest(h[1],50) +@highest(H[1],25)
+@highest(H[1],10))/3,50)

Low average:
LAV = @average((@Lowest(L[1],50) +@Lowest(L[1],25)
+@Lowest(L[1],10))/3,50)

Moving average:
MA50 = @average(C,50)

Tree:
TREE = (HHV-MA50)/(MA50-LLV)

Tree stump:
TREESTUMP=(MA50-LLV)/(HHV-MA50)

Branch:
BRANCH = TREE- TREESTUMP

Ok, I often (always) use tree parts for field/indicator names when I cant think of anything meaningful. HAV and LAV form a band around the MA50. If you plot the three, note the position of MA50 within the band with respect to the strength of the trend. BRANCH is a plot that shows the behaviour of MA50 in the band. It is largely redundant but if you like visuals then plot it. Generally, BRANCH is pretty smooth and gently curving but when a trend is about to end it usually makes a peak for a bull trend and a dip for a bear trend. Sometimes the peak/dip is obvious, sometimes not so. A fair bit of discretionary interpretation is required. Often the trend is obviously over on the chart when BRANCH dips/peaks. Other times it is slightly pre-emptive… perhaps a time to scale out, if that is your thing. There are also instances where BRANCH told you nothing about a trend ending. This whole set of indicators is not much, but it does seem to show trend strength ok. I used a value of 50 for the look-back for no reason at all. I do not remember/know if it was the best, worst or in the middle of a wide set of good values. I do recall that on CL, a value of 80 gave 50% W/L and profit factor of 2.5-ish. I only remember this as that is what is scribbled in on my notes. Also, the highesthigh/lowestlow look-backs are also very arbitary (50,25,10). In fact, that is how I came up with this idea in teh first place. I couldn't decide on what value to use for a lookback on a breakout so I used the 'vertical' average of a few that covered a wide range. This way the breakout level will be influenced by recent market action as well as distant. I played around with this idea for a while and then added the idea of using teh moving average of the vertical average (I totally just made up teh term 'vertical average' to differentiate it from a moving average. The vertical average is teh average of a set of data at one point in time ie HLOC. And a moving average is.... you know).

While I am here I may as well explain the entry/exit rules. Note that they do not incorporate the BRANCH indicator. BRANCH was just an off-shoot (almost a pun) of my fiddling around. BRANCH could be improved upon 10 fold.

BUY if close is greater than HAV and MA50 is greater than it was about 1 week ago and some measure of very short term volatility is rising (ie the market is expanding in the direction of what could be an up trend)

SELL if close is less than LAV and MA50 is less than it was about 1 week ago and some measure of very short term volatility is rising.

Exit position at MA50

Initial risk for longs = HAV-MA50 on the day of the signal

Trade size (longs) = (capital*X%)/((HAV-MA50)*bigpointvalue)

The entry/exit rules you will note are similar to a few other systems that you have probably seen. I have always used bands around the MA for long term trading. The system I trade now is very similar to this one in that there are bands around a MA.

Ummm, that is about all. Please remember, I am not trying to impress, just giving something for people to play with if they have the time. Plus to some degree the HAVLAV-MA combo does seem to indicate trend strength. But you can see the trend strength just by looking at the chart.
Robert Nio
Senior Member
Senior Member
Posts: 41
Joined: Thu Apr 17, 2003 11:34 am
Location: Palo Alto, California

Post by Robert Nio »

Damian,

thanks for the reply. The results I get are not better than my current approach (opposing signal / stop of NEW signals).

Still lokking ...;-)

Robert
Kiwi
Roundtable Knight
Roundtable Knight
Posts: 513
Joined: Wed Apr 16, 2003 1:18 am
Location: Nowhere near

Post by Kiwi »

Robert, Have you read Stridsman's book.

One of the points he makes is that with some types of entry you find it difficult to improve the obvious results by adding a stop especially a trailing stop. This is also my experience.

What the stops can do is reduce the standard deviation of returns.

They also define maximum risk at that time - which, if you define your risk for new positions based on a protected equity may improve your overall results compared with a system which carries a huge open risk of retracement.
Bondtrader
Roundtable Knight
Roundtable Knight
Posts: 101
Joined: Mon Apr 28, 2003 11:07 am
Location: Zimbabwe

Post by Bondtrader »

It doesn't matter what anybody's view is, concerning a rule for mechanical systems trading. It only matters what the test results say. To find about about a rule, run some tests.
ES
Roundtable Fellow
Roundtable Fellow
Posts: 97
Joined: Mon May 05, 2003 1:02 am

EXITS

Post by ES »

OK IV'E STUCK TO USING 2 OR 3 ATR'S FOR STOPS OR EXITS AND THIS HAS WORKED FINE. A TREND FOLLOWER HAS OPPORTUNITES EVERY SINGLE MOMENT OF THE DAY FROM NEW ZEALAND TO THE LAST MARKET AND THE CURRENCY MARKET NEVER REALLY SHUTS.

TAKING PROFITS OFF THE TABLE IS QUINTESSENTIAL. THEN TAKE THE MONEY AND BUY OR SHORT SOMETHING ELSE. IF YOUR MANAGING SEVERAL HUNDRED MILLION OR A BILLION THEN YOUR QUESTION REQUIRES DEEPER RESEARCH OTHERWISE
eminime
Full Member
Full Member
Posts: 20
Joined: Mon Dec 01, 2003 4:05 pm
Location: Minneapolis, MN

Post by eminime »

So, if I understand what you are saying Kiwi; you may end up having more losing trades (protecting equity) but they could be smaller which overall will help your system performance?

Kiwi wrote:Robert, Have you read Stridsman's book.

One of the points he makes is that with some types of entry you find it difficult to improve the obvious results by adding a stop especially a trailing stop. This is also my experience.

What the stops can do is reduce the standard deviation of returns.

They also define maximum risk at that time - which, if you define your risk for new positions based on a protected equity may improve your overall results compared with a system which carries a huge open risk of retracement.
Kiwi
Roundtable Knight
Roundtable Knight
Posts: 513
Joined: Wed Apr 16, 2003 1:18 am
Location: Nowhere near

Post by Kiwi »

Nearly eminime.

Because the stop smooths out the results stridsman says that the system is better. I am a bit too lazy to figure out why again at the moment as I have different focusses at the moment. But in very simple terms if you max loss without a sl was 5000 and your max with a sl was 2500 then you could bet twice as hard with the stop loss.

John
kianti
Roundtable Knight
Roundtable Knight
Posts: 335
Joined: Fri May 02, 2003 6:10 am
Location: Florence - Italy

Nassim Taleb on StdDev and Skewness

Post by kianti »

In some strategies and life situations, it is said, one gambles dollars to win a succession of pennies. In others one risks a succession of pennies to win dollars. While one would think that the second category would be more appealing to investors and economic agents, we have an overwhelming evidence of the popularity of the first. A popular illustration of such asymmetry in returns is presented to us with the story of the Long Term Capital Management hedge fund. The fund derived steady returns over a dozen quarters then lost all of them in addition to almost all its capital in a single observation....
Belief in the Law of Small Numbers and Overconfidence
The first hint of an explanation for the neglect of the small risks of large losses comes from the early literature on behavior under uncertainty. Tversky and Kahneman (1971) writes “We submit that people view a sample randomly drawn from a population as highly representative, that is, similar to a population in all essential characteristicsâ€
Forum Mgmnt
Roundtable Knight
Roundtable Knight
Posts: 1842
Joined: Tue Apr 15, 2003 11:02 am
Contact:

Post by Forum Mgmnt »

As with most things, there's much truth in both sides of the debate:
  1. All else being equal, I'd much rather have my returns come in evenly. So lower standard deviations are better whether on the upside or on the downside.
  2. If the wins come only every once in a while, I'll still take them, so having volatile upside returns is better than not having upside returns.
  3. Long-Term-Trend-Following Systems tend to have high volatility that is more concentrated on the upside than the downside. There is still a significant amount of downside volatility (just ask a recent Aberration trader) with this approach due to give back at the end of trends and long-series of losing trends in choppy abortive markets.
I have found that mixing systems improves MAR significantly but it lowers volatility, and improves the standard deviation-based performance measures even more dramatically than it does MAR, especially if the systems are of different timeframes and type.

The Sortino Ratio is like the Sharpe Ratio but uses the semi-standard deviation instead of the standard deviation. It is much less commonly used as a measure of performance for professional money management. (PLUG WARNING: VeriTrader includes the Sortino Ratio :) )
kianti
Roundtable Knight
Roundtable Knight
Posts: 335
Joined: Fri May 02, 2003 6:10 am
Location: Florence - Italy

On Aberration and other trendfollowing systems

Post by kianti »

Hi Forum Mgmnt,

the importance of Sortino and Calmar Ratio is also explained on

Trend Following: Performance,Risk and Correlation Characteristics
http://www.mfainfo.org/images/pdf/newsl ... b03rep.pdf

I noticed you mentioned Aberration; not a very good year.

I've been looking at the system and I believe that the way the portfolios are are constructed is too much optimized (not to mention the 1 lot money management rule).

Let's say you trade the medium-size portfolio, would you include the Natural Gas (BigPointValue) or the Lumber in you portfolio?

The question is more general: what's the best way to select contracts to include in a portfolio?

I noticed you considered contract correlation in your Veritrader demo, I had a look on the CSI web site for contract correlation and I think the best way as far as I know is still what you did on Veritrader.
I was wondering if there are others manageable ways to include contracts or sectors correlations into a trading system.

best regards, as ever




[/quote]
Erwin Dicker
Senior Member
Senior Member
Posts: 25
Joined: Mon Dec 22, 2003 3:11 pm
Location: Moelingen - Belgium

It is logical that indicators don't work.....

Post by Erwin Dicker »

Hello, very nice this forum, i just joined it.

I want to put some hypotheses here and i'm interested in reactions.
It is my goal to develop a system (together?) from these hypotheses.

Here they are:

1. no one can predict
2. indicators don't have predicting value
3. entries are unimportant
4. trends have the attribute that signals in the direction of the trend work better.
5. to be succesfull you have to act (very) against human nature / habits

Who reacts?

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

Post by Forum Mgmnt »

1. no one can predict
2. indicators don't have predicting value
3. entries are unimportant
4. trends have the attribute that signals in the direction of the trend work better.
5. to be succesfull you have to act (very) against human nature / habits


Welcome to the forum!

I agree with everything except 3. You can gain an edge using a better entry in most circumstances.

You don't need a good entry to make a profitable system, and they aren't as important as most people think, but that doesn't mean they are unimportant.

- Forum Mgmnt
Erwin Dicker
Senior Member
Senior Member
Posts: 25
Joined: Mon Dec 22, 2003 3:11 pm
Location: Moelingen - Belgium

Post by Erwin Dicker »

Oké, it is not impossible, but you are right.

So we will leave nr 3 out / or better we change it in

3. Every price-action consists of "noise", or random movement, agree?

So the trading hypotheses are:

1. no one can predict
2. indicators don't have predicting value
3. every price-action consists of "noise", or random movement
4. trends have the attribute that signals in the direction of the trend work better.
5. to be succesfull you have to act (very) against human nature / habits



How to make a system now?

1. Choose a time-frame and define the "noise".
2. If price moves MORE than the "noise", within the timeframe, in a certain direction a trend starts. (entry is not unimportant....you are right)
3. Start a position in the direction of the trend.
4. Keep risk constant (this means, adding as long as trends develops), always put a fixed part of equity at risk.
5. Get out if a new trend starts.

So crucial is the definition of "noise".

Good idea? This is "conceptual turtle-trading", isn't it?
Very interested in reactions!!!!

Erwin Dicker
kianti
Roundtable Knight
Roundtable Knight
Posts: 335
Joined: Fri May 02, 2003 6:10 am
Location: Florence - Italy

Post by kianti »

I would give more weight in my central defuzzoider to money management and psicology.

In money management I think there is room for a certain grade of control and optimization; think about risk of ruin or max drawdown.

But it all comes back to psicology, do you wanna become the next Soros or Ninja Turtle overnight? how you gonna cope with drawdowns? you cannot have 80% of winnings ( as we say in kianti, the wife drunk and the bottle full) if you are trend follower; a friend of mine started trading a trend-following system and because he had a few losing trades in the Live Cattle early this year, he 'took profit' and let go one of the best trends of thise year. You call it 'noise' only after. Entry is important, because you're looking for a long option profile for your trades, you're hunting black swans not white. Identifying low risk/high reward points is essential, the noise is the price you pay to be early on the bandwagon.

Merry Christmas and Happy New Year
Erwin Dicker
Senior Member
Senior Member
Posts: 25
Joined: Mon Dec 22, 2003 3:11 pm
Location: Moelingen - Belgium

Post by Erwin Dicker »

Number 4: Keep risk constant in the trade is all about money management. E.g. start with 3% risk, when a trade is good for you, you get a profit in the open position. Add again a position and keep total risk (of 2 trades) at 3% risk. There are a different ways to do it.

Noise: you know the signal/noise ratio? You can compare the random movement with noise, it is the price moves which don't tell you anything about the trend. If your stops stay out of the noise-area the chance is good that you won't get stopped out and stay long in a trend.

Number 5: Do all what is against human nature..... is all about psycho. It is e.g. buying high, not taking profits etc......

So the answers were in the text. Thanks for your reaction!

Merry Christmas and Happy New Year.
kianti
Roundtable Knight
Roundtable Knight
Posts: 335
Joined: Fri May 02, 2003 6:10 am
Location: Florence - Italy

Post by kianti »

Fixed fractional money management seems to be a good way to start and you could also have a look at Kelly formulas to see if 2% or 3% is a reasonable percent rate considering the system payoff ratio and to see where is your mathematical risk of ruin.
Noise also could is linked to the money management, too much noise reduces the number of contracts traded with fixed fractional; and last but not the least what's the difference between noise and volatility? :lol:
Erwin Dicker
Senior Member
Senior Member
Posts: 25
Joined: Mon Dec 22, 2003 3:11 pm
Location: Moelingen - Belgium

Post by Erwin Dicker »

In my view you can better take the largest historical loser as indication for the optimal bet. The new largest loser occurs with the biggest position, isn't it. And this can be far out of the statistical region of Kelly, see my point?

And YES, volatility tells something about noise.

So now is the question what can we take as entry?

Merry Christmas to all readers!
kianti
Roundtable Knight
Roundtable Knight
Posts: 335
Joined: Fri May 02, 2003 6:10 am
Location: Florence - Italy

Post by kianti »

I see your point. I would consider more than just the largest loss on a single position but also other portfolio indicators that consider drawdowns and downside volatility ( you could find the main indicators in the Veritrader documentation: MAR, Calmar, Sortino etc..). In practice a good look at the equity curve will tell you the same things.

If you trade on a fixed fractional % in theory all the positions have the same risk: stop loss + slippage and all the. In practice things are different, extreme events again.

By the way I'm just listening to news, foot and mouth disease, Live Cattle futures.... Entry? still the same, try to enter a trade before it's news; how do I know? I'don't know; the only reliable indicator seems to be price.

Entry price seems to me something fuzzy;

I like Buddha A AND not-A
instead of Aristotle A OR not-A

Merry Christmas again
Erwin Dicker
Senior Member
Senior Member
Posts: 25
Joined: Mon Dec 22, 2003 3:11 pm
Location: Moelingen - Belgium

Adding positions

Post by Erwin Dicker »

----Entry price seems to me something fuzzy---

Imagination is the solution......

Imagine that a price movement is significant. Imagine that this will become a huge trend, what will you do? But you have to obey the rule to keep the risk constant. :?: The risk is the gas pedal of your portfolio.

Conclusion: as long as no stops are hit you HAVE to add as long as possible. Or isn't it?

Keeping the risk constant, or accelerate..... press the gas pedal deeper.

How important is it to add risk when your position has profit? So how much of your profit do you have to risk, and how much of your initial equity. Does it make difference......?

In my view there is no difference between initial equity or open positions + initial equity. It is only psychologically more easy to risk your profit than to risk your initial equity.

Instead of building up a position (in case of the mega trend) you could take more risk per trade and not build up?

Merry Christmas.
Post Reply