|
PureBytes Links
Trading Reference Links
|
Pardo has a great system optimization book where he dicusses parameter
robustness and walk forward analysis, and the proper testing proceedure
needed to be statistically valid.
tproeber@xxxxxxxxxx
> -----Original Message-----
> From: Scot Billington [SMTP:scot.billington@xxxxxxxxxxxxx]
> Sent: Tuesday, April 28, 1998 8:43 AM
> To: RealTraders Discussion Group
> Subject: Re: True Range Volatility Breakout
>
> >Date: Tue, 28 Apr 1998 08:40:59 -0500
> >To: tgparker@xxxxxxxxxxxxxxxx
> >From: Scot Billington <scot.billington@xxxxxxxxxxxxx>
> >Subject: Re: True Range Volatility Breakout
> >In-Reply-To: <3545C756.C1C9F1F4@xxxxxxxxxxxxxxxx>
> >References: <3545A112.3BCD4632@xxxxxxxx>
> >
> I would like to make some comments here that are critical, but not
> meant to
> be offensive.
>
>
> >>The trick seems to be picking the markets that seem to model the
> system
> >>best, and stay away from those that don't. Some markets have been a
> dream
> >>to trade, and then a couple of the grain markets almost destroyed
> me.
> >
> This is a prime example of curve fitting. If the technical system has
> a
> market edge it will work across all markets, and it is advisable to
> trade
> as many markets as possible. Certain market modes, i.e. trending or
> non-trending, will result in better returns for a given system.
> However,
> just because a particular market has trended well in the past few
> years
> does not make it more likely to trend in the future. If the system
> only
> 'works' on a few markets, the system doesn't work.
> >
> >>Identifying and getting out of the "false breakouts" fast also seems
> to be
> >>a necessary acquired skill that's cost me a lot of money to learn.
> >>
> >>I'm learning that the problem with "system" trading you have to
> somehow
> >>learn when to pay attention to the system and when to ignore it and
> listen
> >>to common sense. The trouble with being a "newbie" is you don't know
> when
> >>the system is running you over a cliff.
>
>
> I think this comment is a good example of where most 'system' traders
> go
> wrong. Firstly, sometimes taking the signals and sometime listening
> to
> "common sense", is discretionary not systematic trading. Secondly,
> notice
> the undefined term "common sense". I find it impossible to create a
> working trading plan around "common sense".
>
> >>I got the feeling TRABOS is a rather "oldie but goodie" type of
> system..
> >>i.e.
> >>"cave-man" type of trading. Lot's of grunt work, and no heroics.
>
> Assuming heroics means enormous returns, this is another common
> mistake
> made by new traders. They forget the power of compounding and are
> facinated by grandiose claims. At 25% a year $25,000 turns into
> $888,178
> by the time your newborn can drive (You will need it for the
> insurance.).
> That year it will earn you $222,000 almost 10 times your initial
> investment.
>
> My suggestion is to take the TRABOS system and CORRECTLY backtest it
> yourself on a portfolio of markets two from each sector, grains,
> meats,
> energy, currency, interest rates, stock index (one only). Follow the
> rules
> to the letter and use $150/round turn in commission and slippage over
> a 250
> trade sample. Run statistical tests every 50 trades to check the
> robustness of the system. If it is profitable, you should now have
> confidence to take every trade signal generated. If you are not,
> backtest
> another 250 trades or follow it realtime for several months. You now
> need
> to develop a money management strategy that will give you the return
> drawdown ratio that you find comfortable. I would figure out the
> probabilities of various drawdown levels etc. Lastly, you must
> develop the
> discipline to trade your new winning plan. Trade it for a month.
> Keep a
> diary and a log of each trade. If you skipped any signal or fudged
> any
> rule even if it resulted in more profit or less loss, your discipline
> needs
> work.
>
|