Same code, Same Preferences, Same conservative pos-sizing, 3 different PCs, data errors on 2 PCs
Author: tedclimo
Creation Date: 11/17/2009 7:09 PM
profile picture

tedclimo

#1
Here's the situation: I have an EOD script created on PC #1, that when copied to my backup PCs (#2 & #3) via Wordpad shows nearly identical results in backtesting on all 3 PCs(using conservative RP mode Pos-Sizing). The script is in WL 5.5 (not WSTL). BTW, this EOD script uses TC2000 data not Fidelity data.

Here come the problem: My intraday(15min script), which is WSTL(not real WL 5.5) has nearly identical results in PC #1 as it does with WL 4.5... so far so good. But when I copy script to the 2 backup PCs (via WordPad), I end up with bizarrely different results when backtesting with the most common error message of "Error Processing symbol XYZ. Specified argument was out of the range of Valid values. Parameter name: position < 0"
My Watchlist only has 3 symbols, EDC,FAS,TNA.
Here's another clue: When backtesting over same time frame, PC#1 says it covered 5615 bars of data, while the two backup PCs say they only covered 3640 bars of data.... yet the 15min Fidelity datasources are identical & the timeframes requested are identical & the backtesting pos-sizing is identical & the preferences are also identical.

My best assumption is that PC#1 has a complete set of Fidelity 15min intraday data & for some reason, PC's 2&3 are missing nearly 2000 15min bars(since 1/1/09). I tried deleting the datasources & recreating them, but that did not seem to work.

Is there is special technique to remove flawed data without it remaining resident in my PCs... or am I barking up the wrong tree? I am only using Fidelity intraday data.

Your wisdom would be greatly appreciated.
Thanks in advance,
Ted!

profile picture

Eugene

#2
Recently, there was a known problem accessing the entire set of Fidelity intraday ETF data. Fresh/new downloads (only) are missing mostly all of the Summer data. I'm not sure if Fidelity has fixed it or not. Sounds like it applies to you, because each of the 3 symbols (EDC,FAS,TNA) is a leveraged ETF.

Please call Fidelity when you have a problem with Fidelity data. If they fixed it, right-click on the instrument's chart and select 'Reload Chart History'.
profile picture

tedclimo

#3
Thanks for your wisdom & taking the time to point me in the right direction!
Ted
profile picture

Cone

#4
Warning! If the problem still exists, and you have symbols for which the problem is not present, you'll overwrite the good data with bad.
profile picture

tedclimo

#5
Eugene & Cone,
FYI, the tech support people at Ftdelity's Wealth-Lab help line said they knew nothing of an Intraday ETF data problem. They quickly found out that the problem does indeed exist(still) when I had them scroll thru the ETFs I mentioned to you above.
That's not good news for some, but turned out not to be a problem for me.
We simply copied the data files from my PC that does have a full history of data into the 2 PCs that were missing data & BINGO... No more data errors on the 2 backup PCs.
Thanks again for your tireless support.
Ted
P.S. Thankfully, files are updating good data without replacing it with bad data.
profile picture

Cone

#6
QUOTE:
said they knew nothing of an Intraday ETF data problem
This is why Fidelity customers need to report Fidelity data problems. We actually reported this one through our channels, but maybe the ones who need to know didn't get the memo!
profile picture

tedclimo

#7
No surprise you spotted it first.
You rule.
They drool.
:)
profile picture

Cone

#8
I'm told the gaps in Fidelity intraday historical data will be fixed after the market close. If you've recently created intraday DataSets or downloaded intraday for the first time (after Oct 1st), we recommend that you physically delete the intraday data from your user Data folder (User Guide: Data > Where Data Are Stored) and recreate the histories with an update from the Data Manager.

To check if your intraday data have gaps...
1. Turn off File > On Demand Updates
2. Set the Data Range to "All Data"
3. Run this Utility Script on a DataSet (should work for a MSB).

CODE:
Please log in to see this code.
profile picture

Cone

#9
Seems to still be a problem at this time. Maybe it's after tomorrow's close?
This website uses cookies to improve your experience. We'll assume you're ok with that, but you can opt-out if you wish (Read more).