Herman, Bill, Marshall, Excellent feedback, perhaps we should continue discussion on the -ts board. As Herman stated, longer term systems would suffer less from overcrowding, although the number of people on this board really is a drop in the ocean to the liquid US markets (unless everyone is running large hedge funds ;-) ).
As you say, finding a tradeable mechanical system is quite tough due to drawdown and runs of losers. To overcome this problem, I have developed a list of objectives required to minimise these problems. For example (objectives and reasoning): 1) US index system (liquidity, available leverage, availability of backtesting data, NO SURVIVORSHIP BIAS, internals - more detail below) 2) Timeframe - daily (day job, less time in front of the pc) 3) Max. drawdown 2-3% (index percentage). For example, on a dow system trading one YM contract per $5K account value, 600 points or 5%DD would lead to a margin call. Objectives 4 and 5 are also critical. 4) Winning percentage > 70% (Small runs of losers and strong bounce-back after drawdown). 5) Small amount of capital allocated initially, say $10K or 2 contracts. For emotional wellbeing, but also ties into point 6. 6) Profit target 30%. In the dow example this allows capital to double several times per year using fixed fractional money management. While indicators do work for some, they are only derivatives of price. I believe that using the power of amibroker to study the index internal components, one can arrive at a robust, high win% system fitting the above. If this type of system interests collaborators, I have some starting points which work quite well. Kevin --- In [email protected], "steesehwy" <[EMAIL PROTECTED]> wrote: > > Marshall, > > I'm in group 1, I suppose. I have a good grasp of programming and > trading rule development, and recently came to AmiBroker (and the > Yahoo groups) to learn how to use this product to validate methods > (back-test, optimize, etc). I'm considering automated trading, to > remove the occasional discipline problems I have with taking all my > setups, and exiting according to all my exit rules. I'd be glad to > join a group 4 (non-freeloading research), though each of us may have > very specific issues. Note that the AmiBroker-ts group discusses > trading system development, but is a much less active group than this > one. If some folks would like to continue this discussion there, or > in a new group, I'd be glad to discuss AmiBroker trading issues and > research areas of mutual interest. > > BTW, my current trading consists of intraday Russell 2000 emini > (Globex, ER2). > > - Bill > > --- In [email protected], "M. Smith" <ink@> wrote: > > > > Herman, I have been trading for over 35 years. I have met many a > trader who says indicators are the only way and I have met as many who > say you are nuts to waste your time on them because of the randomness > of the markets. Now, Those who use indicators and do relatively well > are statistically in the minority. That is usually because even with > a good system, they can not consistently use it. They alter their > trading rules during the day. Personally, I have been with the > Amibroker group for a long time. They have all been as helpful as > possible. What I have noted about the group over the years is that it > can be divided into groups. Group 1 has the programming language > capability and good sense about trading rules. They put together their > systems and really do not share their work. Personal choice. The 2nd > group is a group who has a great mind for sequential formula > development which assimilates the rules on paper,but does not have the > programming skills. They seek advice from simple to complex issues of > language. #rd, is the group who want everyone else to work out the > bugs and then write out the formula and tell them when it is ready to > use. That is strictly the way I read the group. I would love to see > the great minds of the group work as a team and see what is there to > assemble into a winning formula for the group to try and produce data > for testing. It is called research. Then everyone has a stake in the > development process and is not freeloading on the ph.d's and other > brains. I would love to see it run as a separate group and have a > large number of participants. That is only my feelings and would like > as much feed back as humanly possible. Limited feedback means no > takers and an interest level in work is nil. "Sey la vie". Marshall > > ----- Original Message ----- > > From: Herman > > To: [email protected] > > Sent: Monday, January 01, 2007 11:40 AM > > Subject: RE: [amibroker] Very profitable DOW system > > > > > > > > Group efforts have failed consistently. People are ready and > willing to help with programming and technical issues but when it > comes to sharing serious ideas on trading systems it simply doesn't > happen. Dimitri might have been an exception by sharing many excellent > ideas however his systems were always long term, such systems are not > easily jeopardized by over-trading. Many of his systems can be found > in the library, I haven't tried any lately so it is hard to say > whether any would still be working... most systems have a definitive > lifetime. There are many private groups that share internally and > appear (not sure) to do reasonably well. > > > > Perhaps there simply aren't many good mechanical systems around. > Personally I believe that the most successful traders are successful > because they use good judgement and money management, not because they > have a good mechanical system. I also think that a trending market and > lots of good luck are critical ingredients for many traders. Also, > adequate money for a portfolio or a stomach for 13% DDs may be > required to succeed - I wonder how many are in such a position? > > > > best regards, > > > > herman > > > > > > > > --- In [email protected], "kevinoversby" <kevinoversby@> > wrote: > > > > > > Thanks Randy! > > > > > > Your post just set off the lightbulb - that timing issue completely > > > explains the outperformance. > > > > > > Herman - your question is academic now unfortunately but I think the > > > answer would be that the results do not change much. > > > > > > There is some work I have been doing on tradeable instruments > such as > > > DIA using the individual components (based on Dimitri's work). This > > > in turn was inspired by the results obtained by www.dowtrader.net > > > using a similar approach. Do you think there would be any > interest in > > > collaborating on such an EOD system on this board? > > > > > > Kevin > > >
