WARNING: CSI Dollar Index price bug

Use this forum to discuss data providers like CSI, charting, or other non testing software.
Post Reply
longmemory
Roundtable Knight
Roundtable Knight
Posts: 128
Joined: Tue Apr 12, 2011 11:46 am

WARNING: CSI Dollar Index price bug

Post by longmemory » Fri May 13, 2011 9:36 am

CSI has a bug in Dollar Index unadjusted close.
The decimal place is shifted.

Applicability: UNADJUSTED Close.

Coverage: Bug appears to apply to be independent of back adjuster setting and can be replicated even in Raw or IFG files!

Time frame: price data prior to 2007-April-4
An order of magnitude price jump occurs on 2--7-04-05.

Impact: any computation using unadjusted close will be off by an order of magnitude.

Example: Dollar Index RAW, not adjusted:

20070404,DX,"FINEX",200706,82.92000000,82.94000000,82.63000000,82.71000000, 631,200706, 632, 29989,82.96000000, 8.27100000
20070405,DX,"FINEX",200706,82.73000000,82.78000000,82.37000000,82.45000000, 2488,200706, 2489, 31159,82.67500000,82.45000000

On on April 4th, last column shows Unadjusted close of 8.271.
Next day, unadjusted close is 82.45000000 which is in line with cash price of 82.67500000.

Check the back adjusted prices, same problem.

L

Eventhorizon
Roundtable Knight
Roundtable Knight
Posts: 229
Joined: Thu Jul 08, 2010 2:36 pm
Location: Boulder, CO
Contact:

Post by Eventhorizon » Fri May 13, 2011 9:55 am

You will find a similar issue in the unadjusted closes for the Euro and ND.

The Euro series is in Dmark 1998-12-31 and prior. ND is out by a factor of 10 for the entire series.

On the Euro / Dmark issue, I believe CSI's position is "We report the unadjusted data as it was - if you want to convert it to something else it is up to you".

longmemory
Roundtable Knight
Roundtable Knight
Posts: 128
Joined: Tue Apr 12, 2011 11:46 am

Post by longmemory » Fri May 13, 2011 9:57 am

Start of error, Nov 20, 1985 (start of contract):

19851120,DX,"FINEX",198603,156.16000000,157.36000000,156.11000000,157.17000000,12.88600000,128.54000000

Compare the last 3 values, C, U, $:

157.17000000,12.88600000,128.54000000
Not so good.

End of error, April 5th, 2007

20070404,DX,"FINEX",200706,87.90000000,87.92000000,87.61000000,87.69000000, 8.27100000,82.96000000
20070405,DX,"FINEX",200706,87.71000000,87.76000000,87.35000000,87.43000000,82.45000000,82.67700000

The 4th is the last day with shifted decimal place.

L

longmemory
Roundtable Knight
Roundtable Knight
Posts: 128
Joined: Tue Apr 12, 2011 11:46 am

Post by longmemory » Fri May 13, 2011 10:10 am

Hi Eventhorizon:

The CSI position you mention is very sad.

"On the Euro / Dmark issue, I believe CSI's position is "We report the unadjusted data as it was - if you want to convert it to something else it is up to you".

Screws up the data set.
CSI obviously does not care.

Well, can someone please report the bug AGAIN. (I've reported it.)

LM

Josh Reed @ CSI
Full Member
Full Member
Posts: 13
Joined: Mon Mar 07, 2011 1:03 pm

Unadjusted Pricing

Post by Josh Reed @ CSI » Mon May 16, 2011 8:08 am

The Dollar Index prior to July 4th, 2007 was quoted in lesser precision than it is today, by a factor of 10, so the unadjusted close has been reflecting this unadjusted price, many other markets have been represented this way, as if Historical Adjustments were not enabled. I'll be releasing a new version upgrade today that performs the Historical Adjustments on all markets today as to provide continuity. Build 118.

My apologies for this oversight.

Josh

Josh Reed @ CSI
Full Member
Full Member
Posts: 13
Joined: Mon Mar 07, 2011 1:03 pm

Dollar Index change was May 5, 2007

Post by Josh Reed @ CSI » Mon May 16, 2011 8:10 am

To amend my above comment, the date was May 5, 2007, not July 4th.

longmemory
Roundtable Knight
Roundtable Knight
Posts: 128
Joined: Tue Apr 12, 2011 11:46 am

Post by longmemory » Mon May 16, 2011 11:23 am

Josh Reed of CSI corrected the bug really FAST.
He sent me a beta release on Friday night (yes, Friday!) and it seems to work well.

Thank you very much.

After downloading the new version, you might like to review your settings. I've tested the new version with "apply historic adjustments" and seems to do the trick.

L
Attachments
apply_historic_adjustment.jpg
apply_historic_adjustment.jpg (241.57 KiB) Viewed 3208 times

sluggo
Roundtable Knight
Roundtable Knight
Posts: 2986
Joined: Fri Jun 11, 2004 2:50 pm

Post by sluggo » Wed May 18, 2011 12:33 pm

One thing you could do is write a computer program to test for this kind of discrepancy. Over in the Blox Customer Support area of this website you can find (using "Search") a message from 4.5 years ago, showing that somebody did it once, and found 3 futures markets (out of 130 that were tested) which exhibited this particular anomaly. For those who don't have access to the Customer Support area, here's a screengrab image
Attachments
unadjusted.png
screengrab
unadjusted.png (35.31 KiB) Viewed 3170 times

longmemory
Roundtable Knight
Roundtable Knight
Posts: 128
Joined: Tue Apr 12, 2011 11:46 am

Post by longmemory » Wed May 18, 2011 3:42 pm

Sluggo,

Thank you very much.
Excellent suggestion.

I do have a Python program to manipulate back adjusted contract by splicing the Nth data Close into place of Total OI field. (That is the only way of inserting the real close into TradeStation compatible data file.) Since I am just learning TBB, I've yet to update my utilities.

While doing a few searches as per your advice, I've noticed that
--comewish
and
--EventHorizon
have noticed the same bug and reported it.

In fact, "comeWish" spotted identical contract and date!

I guess it took an act of kindness on part of Josh Reed to fix it (plus coming up with the right spin on why this is a bug and not a feature).

Thanks again for pointing me old messages.
This board has a LOT of thoughtful material.

L

Post Reply