|
PureBytes Links
Trading Reference Links
|
Bob - re: "adapting the parameters dynamically over time"
It's important to know whether or not you are employing Average profit per
Trade or simply the equity curve or some other non-equity related measure.
As you have heard on this list, many feel that trading the equity curve is
just more curve-fitting.
> -----Original Message-----
> From: Bob Fulks [mailto:bfulks@xxxxxxxxxxxx]
> Sent: Sunday, July 14, 2002 7:22 PM
> To: Bilo Selhi
> Cc: omega-list@xxxxxxxxxx
> Subject: Re: Limited life span of mechanical systems?
>
>
> At 9:06 PM -0400 7/13/02, Bilo Selhi wrote:
>
> >well, that one of the best equity curves i have ever seen,
> >but not the ideal, since it is near linear and not exponential,
> >although the latter might be due to no pyramiding in which case
> >it is near ideal. congrats.
>
> My test was for trading the equivalent of a single SP contract so did
> not include scaling the trade size with profits. As you said, trading
> a fixed size tends to produce a linear, not exponential, equity curve.
>
>
> >was wondering if you could post the stats for the system?
>
> People posting results without telling how they got them always bugs
> me so I will give some of the numbers and then tell you a little about
> the system:
>
> Net Profit: $643,000 over 51 months = $12,600 per month
> Trades: 376 over 51 months
> % Profitable: 52%
> Win/Loss: 2.0
> Bar compression: 60 minute natural hour bars (sound familiar?)
> Always in the market long or short (including overnight)
> Profit Factor: 2.2
> Max Drawdown: $22,000 = 2 month's profit
> Average Drawdown: $2,600
> Drawdowns over $10,000: 3 over 51 months
> Average Trade: $1,700
> Average bars in winners: 28
> Return on account: 2900%
> Sharpe Ratio: 3.0
> No commissions - but their effect would be small with the
> large average trade size.
> No slippage - actual slippage would depend upon the exact
> order entry method
>
> (With $150 slippage/commission, profit would drop to
> $643,000 - $150 * 376 = $587,000)
>
> What is it? It is one of many variations of Mark Brown's OddBall idea
> I have tested (with some proprietary "tweaks"). The code contains
> only 13 EasyLanguage statements and no non-standard functions. So I
> will leave it as an "exercise for the reader" to duplicate that
> equity curve with 13 EasyLanguage statements based upon the original
> OddBall idea. :)
>
> The code does not include any money-management stops, bad-tick
> filters, or position sizing, etc., so is not a "finished" trading
> system. I do not trade it but am pretty sure it would make a tradable
> system.
>
>
> >and is this really an adaptive system?
>
> The term "adaptive" means different things to different people. I use
> it to mean adapting the parameters dynamically over time. This is one
> of the reasons this system is a bit more robust than the original
> simple OddBall system.
>
>
> Several people have asked privately if the system is for sale but
> sorry, I do not sell trading systems... I posted the equity curve
> because the original question was about the consistency of trading
> systems over time. This linear equity curve indicates that an
> enhanced version of Mark Brown's original idea behind the system can
> be pretty robust over a period of several years.
>
> As we have discussed several times, the key to a good trading system
> is finding some phenomenon that has a consistent "edge" over time
> (like the original OddBall idea). Once you have that, it is pretty
> easy to improve on the basic idea to make it more robust, as many
> people have done with OddBall.
>
> And since I often help others turn their "tradable phenomenon"
> discoveries into workable trading systems, I am not free to disclose
> much about how the systems work, but, as with OddBall, you would be
> surprised at how simple the ideas often are:
>
> > Something happens in the morning that indicates what is likely
> to happen in the afternoon
>
> > When you look at the price data a certain way, a tradable cycle
> becomes apparent
>
> People who can spot these are pretty rare, but often need help
> converting their "discoveries" into tradable systems. A
> "diamond-in-the-rough" has a lot more value after it is cut
> and polished.
>
> Some people may be able to build systems by optimizing parameters on
> common functions such as RSI MACD, Stochastics, etc., but these
> always seem to me to be "curve fitting" that may or may not hold up
> over time - most do not in my experience. (If it were that easy, why
> would anyone work for a living?)
>
> A good "tradable phenomenon" tends to hold up well over time (or
> until too many others start trading it and neutralize it's "edge").
>
> Bob Fulks
>
|