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

System performance

Post by LeviF » Wed Sep 22, 2004 9:31 am

What do you consider to be good performance? I have been working on a long/short stock trading system for a while and i'm running out of ideas. My system scans the universe of liquid stocks (about 5,000) and takes a position size of 1% of equity. I have tested the system on the last six years of data. The results are approximately an 18% annual return, with a 27% drawdown. These are my best results so far. My question is, how do they rank for a stock system? Should it be returning 30% per year, or is 18% good?

daveineagan
Full Member
Full Member
Posts: 14
Joined: Mon Aug 30, 2004 11:44 am
Location: Minneapolis

is 18% good?

Post by daveineagan » Wed Sep 22, 2004 10:51 am

Well, if you start with 50K, 18% in 10 years becomes $221K. Seems pretty good to me. In 15 years the 50k becomes 1/2million. Sounds like there is potential with your system.

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

Post by Forum Mgmnt » Wed Sep 22, 2004 8:18 pm

levijean,

The returns aren't bad for an unleveraged stock system.

However, the risk/reward is not good enough to trade IMHO. I'd look for the same level of returns with a 12% to 15% drawdown at the very minimum for an equities system.

In order to get this, you will probably need to diverisfy across sectors and take trades that are consistent with the general market direction. These are two of the most important stock trading system requirements if you want to get decent risk/reward ratios.

I also think it is harder to do long/short equity systems. You might try dropping the short side or trading something completely different on the short side. I haven't found that symetric long/short stock trading systems work well.

If you can give us some idea what your system does, we can probably help with more specific suggestions for improvements.

- Forum Mgmnt

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

Post by sluggo » Wed Sep 22, 2004 10:22 pm

Someone named "levi" has posted the same query to Van Tharp's board and has developed a fast friendship / greenhorn-to-expert-worshipful-relationship with Paul "PMK". A couple weeks ago "levi" requested similar assistance and received pointers to free, public-domain systems on the Wealth-Lab website. Those systems are, in a way, opposite to "trend following": they buy weakness (rather than strength) and they hold positions for exactly and precisely one day, no matter what. They don't "let your profits run" while "cutting your losses short." Taking careful note that it's two levels of hearsay, some guy says, that the Wealth Lab website test result pages say, that these systems achieve an MAR of 3.0 on a basket of 100 stocks for the past 6 years. This includes both the Clinton-Nasdaq boomlet and the subsequent misery of malaise.

http://mastermindforum.com/phorum/read. ... 209&t=9209
http://mastermindforum.com/phorum/read. ... 033&t=9033

Reggie Johnson
Full Member
Full Member
Posts: 16
Joined: Thu Apr 17, 2003 12:27 pm

Post by Reggie Johnson » Thu Sep 23, 2004 12:14 am

Perhaps I'm missing something...

I traced the code to http://www.wealth-lab.com/cgi-bin/Wealt ... m?id=12868

This code seems really incomplete, where are the money management rules? How can you have test results without bet sizing? What does Wealthlab do when you run out of cash for buying stock?

Ted Annemann
Roundtable Knight
Roundtable Knight
Posts: 118
Joined: Tue Apr 15, 2003 7:44 pm
Location: Arizona

Post by Ted Annemann » Thu Sep 23, 2004 8:18 am

Reggie, the way a Wealth-Lab user specifies money management parameters is to click and pull-down various buttons and menus and dialogue boxes within the WL graphical user interface. You don't specify MM by writing source code in the WL script. note: Veritrader seems to work the same way: you enter the start account size and the slippage/commissions values and the equity algorithm to be used for calculating position sizes ("Trading Equity Base = Total Equity / Core Equity / Closed Equity") into Veritrader's GUI, on the "Global Parameters" tab.

These two pages from the WL website explain the standardized conditions under which all tests are performed, for uniform reporting of results. What's the portfolio? What's the timeslice? What's the starting account? What's the position sizing algorithm? What commissions and slippage are used in the simulations? and other questions are answered.

http://www.wealth-lab.com/cgi-bin/Wealt ... p25APR.htm
http://www.wealth-lab.com/cgi-bin/Wealt ... base?id=55

Bernd
Roundtable Knight
Roundtable Knight
Posts: 126
Joined: Wed Apr 30, 2003 6:39 am

Post by Bernd » Thu Sep 23, 2004 2:47 pm

:wink:
Last edited by Bernd on Fri Apr 18, 2008 6:28 am, edited 1 time in total.

CRM114
Senior Member
Senior Member
Posts: 35
Joined: Tue May 06, 2003 7:51 pm
Location: Florida

Post by CRM114 » Thu Sep 23, 2004 5:35 pm

sluggo wrote:Clinton-Nasdaq boomlet
Was that followed by the Bush bustlet? But then the market improved after everyone saw Bush's bottom.

Asterix
Senior Member
Senior Member
Posts: 44
Joined: Mon Apr 05, 2004 11:16 pm
Location: San Diego

Post by Asterix » Thu Sep 23, 2004 6:34 pm

This may be more relevant under the topic of trading psychology, but I think "good performance" also depends on what YOU want from your system.

For example, if getting a good night's sleep is important, maybe a little lower rate of return but a lot less variability in your equity is a better trade-off.

Or, if your account size is $5000 but your total net worth is $25 million, you might not care as much about variability and be looking for maximum possible returns since you aren't really risking much of your total assets.

But if you take that whole $25 million and put it into your account, you'd probably want a lot less variability and might find an average return rate of 5% would be OK since another 10% wouldn't really make much difference in your lifestyle.

In other words, when you are talking about trading your own money, I think it's really important to judge your system according to what YOU want and what is important to YOU, rather than trying to compare your results to what others have done.

MCT
Roundtable Knight
Roundtable Knight
Posts: 102
Joined: Fri May 16, 2003 7:27 pm

Post by MCT » Mon Sep 27, 2004 5:01 pm

"good performance" also depends on what YOU want from your system.
Couldn't agree with you more ...

To me, ultimately, trading means honestly expressing oneself. It is very difficult to do. One can create some really fancy back-tests and optimizations and be blinded by it. But, to experience & express oneself honestly, not lying to oneself, now that is very hard to do.

Mechanical systems, though they play an important role, should not be too restrictive, complex or overly mechanical. If we cling to them, we will become bound by their limitation. It’s important to remember we are expressing the strategy, and not doing or performing the strategy.

The action of a master trader is like the immediacy of a shadow adapting to its moving object. His task is simply to complete the other half of the oneness, spontaneously. Spontaneity and calm nerves rule; rote performance of technique regardless of changing market conditions and jittery nerves perish. In short, enter a mold without being caged in it, obey the principle without being bound by it; formless, unlimited, alive and free self expression. The highest method is to have no method.

This self expression is highly dependent on process. Now it doesn’t mean that I’m ignorant of my systems optimal parameter values; far from it. The range of risk management parameters ought to be calculated with reference to the characteristics of the underlying process rather than purely from historical testing or data. I find, for practical optimization of risk management methods, real time records are of paramount importance. It captures ones abilities in timing the markets. Timing does not mean predicting the future. It simply means finding the spot, with the best risk/reward ratio, in the flow. I want to optimize over my ability to find this spot, not historical drawdown. Historical drawdown and all the statistics derived from it are unreliable. If one has better timing, ones system is naturally drawdown resistant and all other stats will be just lovely. Why do most good discretionary technical traders invariably have tighter risk/reward ratios than mechanical trendfollowers? Why do trendfollowers have a tougher time with stocks? The answer lies in ignoring timing as just described. Its analogous to playing every hand, which isn't wise at all. A good trader, in my opinion, is concerend with the mathematical expectation of every trade thats put on. Without timing one is no longer the shadow. A simple 20 or 55 day breakout does not in any way capture long term structural relationships. Benoit Mandelbrot has done some great work in the areas of long term dependence(markets have long memory) & market timing(not prediction). He puts MPT, EMH, etc. in their right place. In short, trendfollowing systems retain sensitive dependency on initiation of intial risk and eventual exists; and in most systems this fact is grossly under weighed. This is not to say results matter more than prcoess or decisions. Rather, what Im saying is the current principles over emphesis on money management, although better than anything out there currently, isn't quite robust.

Most trendfollowing methods retain a lag factor and are neither spontaneous nor robust enough to trade all markets. These methods are not like a shadow adapting to its moving object. Instead, they’re like echoes – always responding after the fact. As we all know, over time, the rules of trendfollowing tend to drift, or become dated. Now, if principles can become dated, they’re not principles. And no amount data will change that fact.

The above opinions are simply meant for the purpose of contemplation; nothing more.

Independent types, for further inquiry, explore “Against Methodâ€

Post Reply