Query to the System Design Community

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.
Dallas
Senior Member
Senior Member
Posts: 36
Joined: Wed Dec 31, 1969 7:00 pm

Query to the System Design Community

Post by Dallas » Fri Feb 20, 2009 9:18 am

The System: Enters on Breakout of "X" Day H/L, 5 ATR Stop,Trailing Stop of "Y" days, risks 1%?Trade, uses no leverage.

The Results: Wins on 30% and Loses on 70% of trades. Winning Trades make 6X dollars of losing trades. Expectation 1.15. Winning years 17%, 38%, 36%, 20%, 15% (Ave 27%), One flat year, 2 losing years of 37% and 4%.

The problems: CAGR is only 11%, RAR is 17%. DD are too deep and too long 44% and 45 months.

The Question: What other performance stats would you look at to diagnose a "fix" or improve this system?

I would be willing to pay someone who knows how to address this.

Thanks
Dallas Rooney

rhc
Roundtable Knight
Roundtable Knight
Posts: 464
Joined: Tue Nov 18, 2008 8:46 pm
Location: Oz

Post by rhc » Fri Feb 20, 2009 9:43 am

How about introducing a longer term trend filter to this system?
You could take long signals only if the the trend filter is in 'Long Mode'
You could take short signals only if the the trend filter is in 'Short Mode'

One of the pre-packed Blox systems does just that.
From http://www.tradingblox.com/tradingblox/ ... dition.htm
Donchian Breakout System -- A single unit Turtle type breakout system with a MACD filter. Only trades long when the MACD is positive and short when the MACD is negative.

Trend filter could be your choice of MACD (as per above), Moving averages, Wilders DM+/DM-, PGO filter or whatever else you think would be appropriate for your needs

It may improve the drawdown figures, most probably at the cost of a lower CAGR though, but you could possibly increase your risk percentage (i.e. bet size) to compensate.
Why not try this first and see what you can see.

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

Post by sluggo » Fri Feb 20, 2009 10:38 am

"Uses no leverage" makes me think the instruments traded are stocks, mutual funds, ETFs.

Rather than trying to decide "is the system good?", how about trying to determine "how good is this system compared to other ways of trading stocks, mutual funds, ETFs?"

How well did "Buy And Hold" perform, over the same instruments, over the same period of time? Does the system beat the pants off "Buy And Hold"?

Among those mutual funds with objectives closest to your own, how well did the best of the best perform, over the same period of time? Does your system outperform the best performing comparable mutual fund?

Finally, the system appears to be a pure trendfollower, yet stocks are notorious for their crummy performance on trend following (compared to futures, FX, etc.) Maybe a totally different system philosophy would be more suited to stock trading? Maybe something like the Stochastic system built into Blox? Or a roll-your-own retracement system?
  • If the long term trend is up, AND
  • If the short term trend has been down for the recent past, AND
  • If the short term trend just switched from down to up, today, THEN
  • Enter a Long position on tomorrow's open
The fashionable description of ETF trading systems these days, is "sector rotation," and AFJ Garner has kindly provided an example system here in the Blox Marketplace (link) Maybe it would be a convenient benchmark, for performance comparison.

EDIT: I see that AFJG has donated another system to the Blox Marketplace (see next article below), this time a Momentum System For Stocks. Many grateful thank-yous!
Last edited by sluggo on Sat Feb 21, 2009 7:59 am, edited 1 time in total.

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

Post by AFJ Garner » Fri Feb 20, 2009 2:52 pm


alp
Roundtable Knight
Roundtable Knight
Posts: 309
Joined: Mon Aug 27, 2007 8:09 pm

Post by alp » Fri Feb 20, 2009 10:34 pm

sluggo wrote:
  • If the long term trend is up, AND
  • If the short term trend has been down for the recent past, AND
  • If the short term trend just switched from down to up, today...
This looks like something interesting to play and experiment with.

P.S.: Is Dallas the oldest member in the community (joined in 31 Dec 1969)?

rhc
Roundtable Knight
Roundtable Knight
Posts: 464
Joined: Tue Nov 18, 2008 8:46 pm
Location: Oz

Post by rhc » Fri Feb 20, 2009 10:52 pm

Dallas,

Do you still have your old computer from ’69 . . . and Blox on punch-cards?? :wink:
http://computer.mng-mn.com/images/histo ... cpdp12.jpg

alp
Roundtable Knight
Roundtable Knight
Posts: 309
Joined: Mon Aug 27, 2007 8:09 pm

Re: Query to the System Design Community

Post by alp » Sat Feb 21, 2009 7:27 am

Dallas wrote:The System: Enters on Breakout of "X" Day H/L, 5 ATR Stop,Trailing Stop of "Y" days, risks 1%?Trade, uses no leverage.
What do you mean by <Trailing Stop of "Y" days>? Apart from other suggestions given above, have you also tried adding an initial stop? With regard to sluggo's suggestion to "roll-your-own retracement system" I am not sure what he means by that, but I wonder if an additional rule is appropriate, something like this:
  • If the long term trend is up, AND
  • If the intermediate term (entry) trend is still up (didn't reverse) AND
  • If the short term trend has been down for the recent past, AND
  • If the short term trend just switched from down to up, today...

Roscoe
Roundtable Knight
Roundtable Knight
Posts: 250
Joined: Sat Jan 24, 2004 2:06 am
Location: Houston TX

Post by Roscoe » Sun Feb 22, 2009 7:19 pm

Dallas wrote:The Question: What other performance stats would you look at to diagnose a "fix" or improve this system?

I would be willing to pay someone who knows how to address this.
My $0.02 worth: separate the system from the portfolio/money management - let's call this the system's raw state - then apply a few of Ralph Vince's stats, my personal preference being the Weighted Geometric Mean as well as normal stats: StdDev, Skewness, Kurtosis, etc. and Conservative Expectancy.

Then see what your system actually does and what markets it works well with and only then apply money management to a portfolio of MarketSytems.

What is the system coded in at the moment?
What markets did you want to trade with it?

Dallas
Senior Member
Senior Member
Posts: 36
Joined: Wed Dec 31, 1969 7:00 pm

Post by Dallas » Sun Feb 22, 2009 9:29 pm

Hello Roscoe,

Thanks for your ideas.

In my search for utter simplicity, the system enters on a 500 day breakout and exits on a 250 day high/low, once it gets beyond the initial 5 atr stop.

The instruments are highly liquid common stocks which trade more than 250K shares/day and are at least $5/ share.

Dallas

Roscoe
Roundtable Knight
Roundtable Knight
Posts: 250
Joined: Sat Jan 24, 2004 2:06 am
Location: Houston TX

Post by Roscoe » Sun Feb 22, 2009 10:01 pm

Common stocks, so can I assume long only? And what length is the ATR entry stop?

alp
Roundtable Knight
Roundtable Knight
Posts: 309
Joined: Mon Aug 27, 2007 8:09 pm

Post by alp » Mon Feb 23, 2009 12:10 pm

I think a "long only" strategy, by necessity, is only profitable during uptrends. For instance, Garner's system has other instruments in the portfolio:
The 400 securities I used included not only many world stock markets and sectoral stock indices but also bonds and bond funds, currencies and commodities.

...

Using such a system on stock ETFs alone is sub-optimal. The inclusion of bonds and currency funds means that at times such as these you are out of the stock markets and into cash and bonds.

P.S.: one interesting aspect of this forum is that there is always a high probability that someone has already asked the questions you wish to make. Take a look at:

viewtopic.php?t=5195
viewtopic.php?t=5374
viewtopic.php?t=4647
viewtopic.php?t=4696
viewtopic.php?t=3797

Dallas
Senior Member
Senior Member
Posts: 36
Joined: Wed Dec 31, 1969 7:00 pm

Clarifications

Post by Dallas » Mon Feb 23, 2009 1:24 pm

Roscoe, "my" elementary sytem had a 5 ATR initial stop and it DID go long & short. When I tried either long or short only the results were totally crappy.

Thank you alp for the threads.

alp
Roundtable Knight
Roundtable Knight
Posts: 309
Joined: Mon Aug 27, 2007 8:09 pm

Post by alp » Tue Feb 24, 2009 10:33 am

Well, I am not a big fan of stock trading, but I was intrigued by your performance numbers. Our friend Eric Crittenden, from Blackstar Funds, seems to have found the right answers. Check their two papers "Does Trendfollowing Work on Stocks" and "The Capitalism Distribution" downloadable at: http://www.blackstarfunds.com

Also note that Garner's and lemahsin's systems, featured in the referred threads, are scanning hundreds or thousands of securities.

Finally, on par with Seykota's typical mottos, I think you might check your feelings about commitment to a system whose results your tests indicate are crummy.

adamant
Roundtable Knight
Roundtable Knight
Posts: 118
Joined: Mon Aug 25, 2008 1:20 am
Location: Norway

Post by adamant » Wed Feb 25, 2009 7:05 am

Alp: Regarding Crittendens work, which is very interesting: It is important to note that the test period includes one of the greatest secular bull markets of all time. Crittenden states clearly that the system performance on data from the seventies is substantially weaker.

adamant
Roundtable Knight
Roundtable Knight
Posts: 118
Joined: Mon Aug 25, 2008 1:20 am
Location: Norway

Post by adamant » Thu Feb 26, 2009 6:54 am

Also, there appears to be some inherent problems with leveraged etfs:

http://www.chartingstocks.net/2009/02/l ... #more-1118

DeanoT
Roundtable Fellow
Roundtable Fellow
Posts: 52
Joined: Sat Apr 15, 2006 7:37 am
Location: Melbourne, Australia

Post by DeanoT » Thu Feb 26, 2009 7:24 am

Leveraged ETFs/ETNs which compound daily certainly underperform over the longer term against the index they are intended to track, and as such are generally only geared towards short term traders. There are however, a number of ETFs/ETNs, including quite a few issued by Deutsche Bank, which compound monthly, which can remove some of the punishing effects that daily compounding can present.

alp
Roundtable Knight
Roundtable Knight
Posts: 309
Joined: Mon Aug 27, 2007 8:09 pm

Post by alp » Thu Feb 26, 2009 6:53 pm

adamant wrote:Alp: Regarding Crittendens work, which is very interesting: It is important to note that the test period includes one of the greatest secular bull markets of all time. Crittenden states clearly that the system performance on data from the seventies is substantially weaker.
Well, apart from some beautifully constructed equity curves from someone's black box system, which we usually cannot find in the real world, I see a lot of talented people having a hard time with stock trading. Does it have something to do with this:
Eric Winchell wrote:Trend following in stocks often puzzles those who wonder why it doesn't work like it does in futures markets, but maybe it's not such a big mystery. Richard Dennis reduced this to stocks being "more random," but there is a simpler explanation: stocks, especially stocks of a single country, are correlated. They represent their parent asset class too much to provide the type of diversification that trend following requires. Soybeans and T-bonds are separated by more degrees of freedom than IBM and GE.

A market in a futures portfolio can go 20 years without a big trend and then "wake up" to provide a winner that makes up for that entire period. The overall stock index is susceptible to this kind of cycle and so are the individual stocks making up the index. There's no reason to expect a trend following system trading a basket of stocks to perform much better than one that trades a few stock index futures contracts.

The correct approach then is exactly what Blackstar has done: represent a trend following method within the world of stocks and then compare the risk-adjusted returns with the S&P rather than to a traditional futures portfolio. Their study shows the way to be a 100% mechanical trend follower in the stock market (and is a great piece of work).
With regard to Jim Simons, is their track record publicly available or is he becoming one of those legendary Market Wizards of whom we hear legends of their achievements?

And, finally, do you know of any magical filters I can use to select the best performing instruments and trades?
Last edited by alp on Thu Feb 26, 2009 9:07 pm, edited 1 time in total.

alp
Roundtable Knight
Roundtable Knight
Posts: 309
Joined: Mon Aug 27, 2007 8:09 pm

Post by alp » Thu Feb 26, 2009 7:29 pm

alp wrote:...I think you might check your feelings about commitment to a system whose results your tests indicate are crummy.
Look for further evidence of an emotional knot if substituting <system> for <relationship>, <job>, etc. rings a bell.

adamant
Roundtable Knight
Roundtable Knight
Posts: 118
Joined: Mon Aug 25, 2008 1:20 am
Location: Norway

Post by adamant » Fri Feb 27, 2009 5:51 am

And, finally, do you know of any magical filters I can use to select the best performing instruments and trades?
Not sure why you appear to be asking me this question.

Paul King
Roundtable Knight
Roundtable Knight
Posts: 207
Joined: Mon Feb 23, 2004 9:13 am
Location: Vermont, USA
Contact:

Post by Paul King » Fri Feb 27, 2009 7:36 am

"And, finally, do you know of any magical filters I can use to select the best performing instruments and trades?"

Yes, only trade instruments that are liquid ;-)

Post Reply