|
PureBytes Links
Trading Reference Links
|
----- Original Message -----
From: "Jimmy Snowden" <jhsnowden@xxxxxxx>
To: "Ivan Figueredo" <ivanf1@xxxxxxxxxxxxx>; <Omega-list@xxxxxxxxxx>
Sent: Saturday, February 28, 2004 2:39 PM
Subject: Re[2]: QuantStudio
> Ivan,
>
> You may need to work with TS2ki a bit more. If I understand your list
> of errors you believe TS2ki makes it seems that you are bending or
> breaking the rules of EL or the program in general thus causing what
> you call errors. It takes a lot of knowledge about what you can do in
> TS2ki to get honest results. That is a big part of what the Omega
> list is for. List your problem and see how people on the list are
> getting the program to do exactly what you want.
Well, that may be. "It takes a lot of knowledge about what you can do in
TS2ki to get honest results." How can I reply to that? LOL, are you saying
that if I knew all of the ways that TS has bugs, I would then know whether I
had "honest" results or not? And non-programmers get this program sold to
them?
> Below is the best I can do with out more info. on what you are
> experiencing. Perhaps others might be more helpful or maybe we will
> need more information.
>
> 1. Changing the days to look back will throw your values off if they
> have not had time to calculate properly. My guess is you are causing
> this problem and it takes some real effort to check the values out.
I have already stated several times that the effect has nothing to do with
the added data for calculation. The system is mysteriously affected by the
previous day of added data.
> 2. Here again with proper setup this should never happen to you.
That could be. I am not ruling out the possibilty that it could be me.
However, I REALLY doubt it.
> 3. Here you seem to not know how to check the results of your code.
> I use the expert commentary a lot but others use different ways to be
> sure the values are what you think they are. Some export values to a
> file others use the debug in the Power Editor.
Hmmm, well, I have tried both to no avail.
>Many times your system
> will not give the results you expect because of the difference between
> an indicator on the chart making it look like you should have a
> signal, but in fact your system shows different values then the
> indicator when you look at them. Indicators will, if setup to,
> automatically compute the number of bars to calculate on.
AHA! This may be it, since I am looking at an indicator that is set up
EXACTLY the way the code expects it, but perhaps the iternal logic for
indicators is different for systems. In either case though, that would be a
bug. However, I will look at this more closely - this is the first
suggestion that actually makes sense.
>A system
> must be setup to do this in max bars back. The results should be
> exactly the same if you have it setup correctly.
Like I said above, I do not believe this is the problem, but I will verify
it.
> 4. I don't have this problem, but if it does happen then why not keep
> the scroll forced to the bottom. Actually why do you want to look at
> performance during the day. This is when most of us trade.
I have no idea why this happens. It does not happen all the time, but it
does spuriously.
> 5. Here you have a serious problem. TS2ki loads data into a temp file
> and does not need to reload old data to the chart. To fix this we
> would need to know a lot about your computer because you have a
> serious problem. Is the machine capable of keeping up. What OS?
> What version of TS2ki do you have? When did this start? Have you
> done a clean install for TS2ki? Are you downloading data all the time
> or is your data feed working fine all the time? Get TS2ki and the
> computer to work properly before you attack any other problems. Big
> time red alert.
I guarantee you, the machines are more than ample for the task. The OS is
windows 2000 Server. I have SP5 2000i Platinum. It started as I added more
and more systems, and more and more charts.
> I WOULD BE VERY CAREFUL TRADING UNTIL I GOT EVERYTHING FIXED AND LEARN
> HOW TO CODE TS2ki's WAY SO THINGS ARE REPEATABLE AND ACCURATE.
Emphasis mine. I could not agree with you more.
> That
> stuff takes a long time to get right, but then I'm pretty slow maybe
> it will go faster for you.
I have been trading for years. I would never use TS to base anything
on, except to display values and chart them for me - At that, it does a
marvelous job. However, I think your warning is a very good and honest
response to people first getting started with TradeStation.
Ivan
>
> Best regards,
> Jimmy Snowden
> mailto:jhsnowden@xxxxxxx
>
>
> Saturday, February 28, 2004, 1:12:39 PM, you wrote:
>
>
> IF> ----- Original Message -----
> IF> From: "DH" <catapult@xxxxxxxxxxxxxxxxxx>
> IF> To: "Omega List" <omega-list@xxxxxxxxxx>
> IF> Sent: Saturday, February 28, 2004 10:01 AM
> IF> Subject: Re: QuantStudio
>
>
> >> > I
> >> > can give you five different situations where TS is simply doing buggy
> IF> things
> >> > (as of TS2000i.)
> >>
> >> Please do. You have stated that the calculations of TS are inaccurate.
> >> I'd like to see examples of that, other than the obvious things
everyone
> >> knows about - single precision, bouncing ticks, etc.
>
> IF> 1) Take any system that exits at the end of day so that P/L for a
given day
> IF> is not affected by overnight trades. Go back say, one day. Go to
> View->>performance Report->daily Tab. Note the P/L. Now, go and bring up
the
> IF> symbol again. Change number of days to look back to 2. Go back and
look at
> IF> the daily report. Look at the P/L for the day that you looked at -
they are
> IF> different.
>
> IF> 2) Take any system that gives signals in realtime. Watch it during the
day
> IF> without ever closing TS down. When the markets close, shutdown TS and
come
> IF> back in, do not save the workspace. Now, run the system again.
Different
> IF> values.
>
> IF> 3) This is is MUCH harder because it would require me to give you code
that
> IF> I cannot. But here is the gist of it. Make a system. Apply it to say
ES and
> IF> YM. Look at the signals it generates. Take note that sometimes, where
a
> IF> signal UNQUETIONABLY should have been generated, one was not. You sit
there
> IF> and stare at the code, and unfortunately, you do not have a debugger.
Printf
> IF> debugging does not help here.
>
> IF> 4) I have about 12 charts open, each with 1 of two systems applied to
the
> IF> symbol. Try to look at the performance report during trading hours.
Evertime
> IF> I scroll to the bottom of the trades tab, it keeps going off, doing
> IF> something, and taking me back to the top of the page.
>
> IF> 5) A variant of the above is, often a symbol, for whatever reason,
keeps
> IF> needing to "get data." I have no idea why this happens, and it does
not
> IF> happen all the time. The effect is that the entire history for the
symbol
> IF> has to be reloaded every minute or so. This makes following a system
under
> IF> these conditions intolerable.
>
> IF> I have about 20 others that I have accumulated over the years. I do
not
> IF> remeber them all, but I will start to write them down as I see them
again.
>
> >> BTW, TS7 fixed the single precision problem for those who insist on
> >> working with the difference between two large numbers, e.g. bad code.
> >>
> >> > That is why
> >> > we spent close to a million dollars getting building a tool that
tests
> >> > hundreds of systems in realtime over hundreds of millions of ticks
over
> IF> a
> >> > cluster.
> >>
> >> Like I said, if you need more power than TS, it's time to shop
> >> elsewhere. Certainly it was never intended for that application.
>
> IF> Yes, well, I am saying less than that. I am saying that EVEN in the
case of
> IF> "simple" systems, TS is at best a huge time drain because you can
never be
> IF> sure that what you are getting back is repeatable. That is the tennet
of
> IF> scientific research, repeatability, and TS fails badly when thousands
of
> IF> ticks are involved and what I consider relatively simple systems.
>
> IF> But look, this is not a debate between you and me. I know where TS
sits on
> IF> my usabilty scale, and you seem to know as well. this "warngin" is
for
> IF> those that blindly come into this business, buy TS, and think that
they are
> IF> performing real world tests. Nothing can be further from the truth.
Buyer
> IF> beware.
>
> IF> Ivan
> >> --
> >> Dennis
> >>
>
>
>
>
> Outgoing mail scanned by Norton
>
>
|