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

Re: Beware! Hideous BMI Feed problem...


  • To: omega-list@xxxxxxxxxx
  • Subject: Re: Beware! Hideous BMI Feed problem...
  • From: Neal Hughes <neal@xxxxxxxxxxxxx>
  • Date: Thu, 23 Mar 2000 14:23:52 -0800
  • In-reply-to: <200003232001.NAA47889@xxxxxxxxxxxxxxx>

PureBytes Links

Trading Reference Links

>At 05:07 PM 3/23/00 -0500, Ian MacAuslan wrote:
>Yup, that's what I'm saying.   Yes, my 5-minute charts were crap.  I
>interpreted their large range  as "volatility"--and spent a good bit of
>time tweaking my systems trying to accomodate it.
>
>I have been using TS4.
>
>I have no idea what "caused" it.  I can only say I was watching TS4's
>System Tracking Control Center once day and noticed the prices moving
>erratically--e.g.,
>140^5 200
>140^5 1000
>139^8 500
>140^4 1000
>
>Not surprisingly, TradeStation kept whippsawing itself in and out of
>positions.  I got really suspicious when I was watching AT Financial's
>Time-and-Sales window for the same stock, which uses an entirely
>different datafeed (Comstock)-- and saw no such "139^8" trade occur-- but
>was able to see it on the AT/Comstock tape exactly 20 minutes earlier.
>
>In my case, in August '99 I had added the CME exchange to my BMI account
>(which requires BMI's commodities/futures package--which includes, as it
>turns out, delayed stocks, as well).  But I believe I started noticing
>the problems in December/January.
>
>Ian
>

Wow!

As if traders don't have enough stacked against them..

How do mechanical systems cater to these situations?
Does anyone have good ELA code for these eventualities??

How do traders feel, placing their money at risk based
on signals generated by mechanical systems? Who knows
what the next data error will be, how severe, which
data vendor?

It takes a TON of emotional strength to trade mechanically!
Or you need to ignore the risk? I may learn to admire
mechanical traders, it takes guts! Or ignorance..

This is a good one for Mark Brown.. If you're reading this,
Mark, how does your programming test for validity of data?
What are you testing for? Everyone should incorporate such
tests, because any data vendor could start having problems..

How about back-testing.. It's wise to back-test with some 
faulty data, to test for true robustness... Any system tested on
"clean" data could be said to be curve-fitted for clean
data?

-Neal.

-----------------
Neal on the 'net.
Trade well. Train hard.
http://www.halcyon.com/neal/