[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Fut: T-Bonds Data accuracy



PureBytes Links

Trading Reference Links

At 6:49 PM -0500 2/20/99, BrentinUtahsDixie wrote:

>I thought this might be of interest so some. I got some data for the Bonds
>from BMI which I put in Data1 and the data in Data 2 is from a friend that
>uses DBC. I used the Median-Price to show the differences in the data's
>ranges as opposed to using the close only. Would the true data please
>stand up. Makes one wonder about certain signals or something like
>candlesticks doesn't it.


Makes you wonder why anybody could ever want to design a system that
depends upon having "perfect" data doesn't it?

The fundamental price at which you could buy/sell is what you want but is a
fictitious variable that you cannot observe.

They report trades as they occur but the time between the trade and when
they are reported is a variable over which you have no control so it is
late by some unknown and variable amount.

Some of the trades in some markets are reported by a person typing into a
computer what he sees happening in the pit. If the action is too fast, he
has to skip a lot of trades.

Then your cable data source inserts a commercial or switches the
connections around to further mess things up.

Then you slice up this tick data into 5 minute bars, or whatever. Minor
deviations of what is in which bar make very different looking bars.

Then you get occasional bad ticks based upon a myriad of different causes.

>From this you see people trying to match precise patterns to signal a
trade. What a joke! GIGO: garbage-in = garbage out.

So in summary, the stuff reported by BMI, etc., is a noisy sampling of an
imperfect reporting process with communications errors. That is the way it
is. So your system had better be tolerant of it.

If you look at the chart you posted, the overall short-term trend was about
the same in both charts so this is about all you can count on for a trading
system. Expecting anything better is fooling yourself.

Bob Fulks