|
PureBytes Links
Trading Reference Links
|
George Pruitt of Futures Truth fame and I both were trying in earnest to get his latest breakout system (NR4 days) published in Active trader or Futures mag to work properly about 2 or 3 months ago ---- we finally discovered TS7 data was screwed up --- that's why the system was not even profitable compared to his system run on his clean data.
John
On Wed, 19 Nov 2003 09:23:41 -0700 , "Chris Evans" <evanscje@xxxxxxxxxxxxx> wrote:
>Some examples . (for the S&P active contract)
>2/8/1999 TS7.1 high=open+9 pts ....actual high=open+3 pts.
>4/5/1999 TS7.1 high=open +14 pts .. actual High=open +17 pts
>5/21/1999 TS7.1 low=open-2 pts .. actual Low=open-13 pts
>
>there a more but I stopped writing them all down ...these variances caused meaningful discrepancies in the performance of one of my systems -
>
>----- Original Message -----
>From: "carrslem" <carrslem@xxxxxxx>
>To: "Chris Evans" <evanscje@xxxxxxxxxxxxx>; "Omega List" <omega-list@xxxxxxxxxx>
>Sent: Tuesday, November 18, 2003 12:46 PM
>Subject: Re: beware bad data
>
>
>> Could you please provide a couple of specific examples - dates, times, and
>> prices. I have occasionally spot-checked TS7 S&P500 front-contract prices
>> against the CME's official T&S files and have, so far, never found any
>> significant discrepancies.
>>
>> Regards,
>> Carroll Slemaker
>>
>>
>> ----- Original Message -----
>> From: "Chris Evans" <evanscje@xxxxxxxxxxxxx>
>> To: "Omega List" <omega-list@xxxxxxxxxx>
>> Sent: Monday, November 17, 2003 4:00 PM
>> Subject: beware bad data
>>
>>
>> > for the record .. I design and trade systems using TS2K and my data vendor
>> has been Genesis Data. .. so I finally succumbed and got TS7.1 so some of
>> my signals would be entered automatically , so before setting them up I
>> checked the systems and results to be sure I got the same history as I got
>> on TS2K ..
>> > I found significant discrepancies .. which I then had to reconcile..
>> in the end all the differences came down to different data histories so I
>> used Bloomberg to check and see which data was correct .. in every case
>> Genesis was right and TS7.1 had errors that created incorrect historical
>> trades and performance. I'm talking about major errors in daily prices for
>> the S&P 500 lead contract - not some obscure palladium contract!
>> > After seeing what 'I've seen I could never use TS7.1 to design a new
>> system...
>> >
>>
>>
>
>
|