|
PureBytes Links
Trading Reference Links
|
"Are they the same? Are they ever going to be the same? If not, what does
this mean
about your system that calculates entry points to the tick?"
This means you need a system that's substantially profitable to remain
profitable after taking into account commish, slippage and margin of error
between EOD and real time implementation. The less it trades and the more
it makes, the less these items will be factors. Highs and lows are usually
the same between EOD and RT. Opens and closes can be off by a couple ticks.
The assumption I make is that over a substantial period, the differences in
open/close between RT and EOD (+/-) will average out to be insignficant (i.e
2 ticks). Therefor I expect results to be similar in RT implemenation as
backtestig results are. Of course, the less my system trades and the more
it makes, the less these items are factors in it's overall success. If
you're testing to the tick level then it's probably not best to use the open
and close any way because, as was pointed out, can be different. Better to
use say the bar 1 or 2 minutes after the open and 1 or 2 minutes before the
close. I find this not only gives me more resliable backtesting results but
more closely simulates the way I trade. I feel a lot better getting in 1-2
minutes after the market opens and getting out 3 minutes or so before the
market cloes than I do placing MOO and MOC orders. This applies mainly to
CBOT and CME not NY where getting in during the preopen is difficult unless
you place your order before the open. Regardless, I use my backtesting
results to look for profitable trends in the data and techniques that will
signficantly improve results, not exact p/l figures. Implementation can and
will vary results.
> -----Original Message-----
> From: The Omega Man [mailto:editorial@xxxxxxxxxxxxx]
> Sent: Friday, July 09, 1999 2:22 PM
> To: omega-list@xxxxxxxxxx
> Subject: PS2Ki WORKS
>
>
>
> I've said it before and I'll say it again: You guys are living in a dream
> world if you think that *any* real-time feed is even *close* to
> what you'll
> get if you download or purchase historical data.
>
> I challenge those who think that real-time and historical tick data are
> similar to run some experiments and investigate the situation. Don't
> believe me! Check it for yourself. Save your tick data and then go get
> some "refresh" or historical data somewhere. Check the tick counts, check
> the opening price, see what you find.
>
> You think your broker's data is more accurate than yours? Fine.
> Check your
> real time feed (any feed) against the quotes your broker shows. Are they
> the same? Are they ever going to be the same? If not, what does
> this mean
> about your system that calculates entry points to the tick?
>
> Look: the question here is one of *fact*, not opinion. Experiment for
> yourself and see what you discover. Then, think about your technical
> analysis in light of what you learn about the quality of your data.
>
>
> Good trading,
>
> The Omega Man
>
>
>
>
> ----- Original Message -----
> From: William R Wood <wr_wood@xxxxxxxxxxxxx>
> To: <omega-list@xxxxxxxxxx>
> Sent: Friday, July 09, 1999 5:04 PM
> Subject: RE: Prosuite 2000i problems
>
>
> > To Folks who have no problems:
> >
> > Im curious about whether the satisfied customers have actually
> checked the
> > accuracy of thier data. Do you call your broker and your data
> vendor and
> > get them to read off price quotes, bid/ask changes and intraday volume
> > numbers at random times during the day on several active stocks
> like INTC,
> > DELL. And do you then call at the end of the day and get total tick
> counts
> > which you check against GS/Edit Symbol/Edit Data/Load Data -
> Trade Record
> > 1Tick. Im not sure what you check for futures since I am a
> stock trader.
> > But unless your intraday prices match your data vendor and your broker
> with
> > no delay you are not getting real time data. And unless your end of day
> > tick counts are accurate within 1% you are not getting accurate data and
> you
> > do in fact have problems
>
|