|
PureBytes Links
Trading Reference Links
|
Yuki,
Thanks for the insightful post on how to enhance trading with the
StochRSI. I hope to try out some of your ideas.
b
--- In amibroker@xxxx, Yuki Taga <yukitaga@xxxx> wrote:
> Hi Steve,
>
> Just to add a couple of things to your fine comments:
>
> Tuesday, July 23, 2002, 12:09:31 AM, you wrote:
>
> SK> A couple general comments on the StoRSI/QQQ approach:
>
> SK> Filter suggestions: Try only taking an "opening" position in
the
> SK> QQQ when the 13 (21, or your number of choice) SMA is pointing
in
> SK> the direction of your trade. Sounds too simple, right? You can
> SK> apply the same approach by using any number of linear regression
> SK> tools. Pick a "trend identifier" (and believe me, trend is the
> SK> hardest thing to wrap your arms around), and only trade in the
> SK> direction of the trend. Yes, it's that easy. Eliminate the
> SK> "stinkin" trades that were initiated against the trend. A 13
day
> SK> SMA is a starting point and with the ability to "optimize" using
> SK> AB, you can identify many averages that should improve the
> SK> overall performance and keep you (most times) on the right side
> SK> of the market.
>
> Another simple idea for improving percentage winners here is to NOT
> take signals where price doesn't react to the signal itself. You
got
> a buy signal, but prices opened sharply lower the next morning, and
> you took the signal anyway? Okay. It will work sometimes, but when
> one does that one is saying that one's market genie is now smarter
> than the market as a whole. And when one is trading a very short
> term system (as I like to do), one needs to be very sensitive to
> short term price, IMO. Maybe one of you coders out there could run
> some tests and figure out the results for the following:
>
> 1) what percentage of trades went south (or had uncomfortably --
> yeah, hard to agree on what that means for everyone -- large draw
> downs before coming back) when price failed to exceed the high of
the
> signal day on the day you were supposed to pull the trigger
>
> 2) what is the optimal waiting period for price to exceed the high
of
> the signal day (next bar only? two bars? four bars?) IOW, when
> does the signal become null and void when price fails to follow
> through, based on a back test
>
> Finally, Steve and I both yearn for complete hands-off automation.
> But there's clearly a trade off. On the plus side, if you really
> have a good system, keeping hands off keeps you from mucking it up.
> On the down side, you have to live with situations from time to time
> that are probably pretty easy to second guess. Specifically, taking
> counter trend signals on a short term oscillator when there is no
> nearby reasonable price support (the place where you cut and run if
> it fails) can lead to some pain. And the obvious exit is, to me,
> rather obvious. :)
>
> A couple of comments on my attached cut of Steve's gif.
>
> Best,
>
> Yuki
>
> mailto:yukitaga@x...
|