Support Board
Support Board
Search Board
Control Panel
View My Posts / Threads
Direct MessagesDate/Time: Mon, 16 Sep 2024 19:48:07 +0000
Post From: Chart data errors and the message log
| [2015-08-06 07:16:18] |
| i960 - Posts: 360 |
|
Messages like these are triggered by error messages from Interactive Brokers TWS:
HD Request # 2468 | Message from IB: No security definition has been found for the request. IB Error Code: 200. Request ID: 52. | 2015-08-05 22:14:56 HD Request # 2468 | Error downloading historical Intraday data for D-201509-LIFFE_NF. The symbol is unknown to the server. | 2015-08-05 22:14:56 * I know it's from IB, but I'm simply using the definitions that are there. A similar one exists for W-######-LIFFE_NF and C-######-LIFFE_NF-GBP. Perhaps the problem here is that IB no longer uses that as the exchange label. It should be ICEEUSOFT: https://www.interactivebrokers.com/en/index.php?f=2222&ns=T&exch=iceeusoft&showcategories=FUTGRP I have just tested this with D, C, and W and there are no errors after that. The Sierra side data still needs to be fixed so others don't run into this and it also needs the rollover periods fixed because they aren't defined. I don't know what's up with the VIX contract, I believe I'm using the one provided server side. Yes it is reasonable.
If one wanted to chart a contract from the beginning of inception of that contract they'd have to change the days to load every single day! If something came to life 100 days ago, and I set the chart to load exactly 100 days back then I have to reset the chart to 101 days the next day, or 120 days 20 days after the fact. That isn't reasonable. Think about how this would affect somebody trying to trade a new contract introduced - they'd be throwing errors for months on end until there were enough data. You're thinking with a coding mindset here. It shouldn't be throwing errors for something that's *pragmatically* a non-error situation to the overwhelming majority of users. A reasonable approach from a coding standpoint is to simply only throw errors on missing data files **only if a previous datafile in the continuous series has already been loaded**. It's still free to report it to the log, but it shouldn't be a hard error. If the number of successfully loaded contracts in a continuous contract being assembled is >0, *then* make it a hard error. Date Time Of Last Edit: 2015-08-06 07:26:52
|