CSI Beta 120

Use this forum to discuss data providers like CSI, charting, or other non testing software.
Post Reply
LeapFrog
Roundtable Knight
Roundtable Knight
Posts: 695
Joined: Mon May 17, 2004 4:18 pm
Location: Boston, MA

CSI Beta 120

Post by LeapFrog » Thu Apr 19, 2012 10:31 am

There are a lot of changes in this Beta version, now available for download for anyone who wants to try it. I am using it in production - have done for 3 months with no issues.

http://www.csidata.com/?cat=4

Many of the changes were initiated by some of us on this forum - TB users.

A couple of highlights:-

1) There is a NEW method for splicing roll data together which is now set as the DEFAULT method for NEW instruments added to your portfolio. To match existing instruments up to use this method you have to go through each one and edit it. It is worth doing IMHO as all the existing methods do it in a way that does not match the reality of trading. We roll DURING a session, not after a session has closed. All the existing CSI stitching methods roll BETWEEN sessions. This new method is the ONLY one that rolls DURING a session and is therefore the one to choose as the stitching method. Choose that method and work through a roll between the OHLC from the previous contract to the new contract and you will see how the stitching works.

2) The custom roll editor is a powerful feature which now allows us to set ROLL BY DATE to a high level of accuracy, based on TRADING DAYS and based on any day of the month we choose to roll on. One example being now able to roll 'x' days before FIRST NOTICE DAY. You define when FND is, based on the exchange rules, then set the roll day to be 'x' days before that. For example, I've set my grains to roll 7 business days before the last day of the month preceding the contract month - the last day of the month being FND.

I know many people here do not like to touch these Beta versions of the CSI platform, but for those interested, I believe it is well worth the effort to take a look at the new features and try them out, test them in a sandbox, and get ready for when it will eventually become the new CSI production platform.

As I mentioned, I have been using it for over 3 months now, in production, in live trading, and have no issues with its stability or accuracy.

Roger Rines
Roundtable Knight
Roundtable Knight
Posts: 1946
Joined: Wed Oct 06, 2004 10:52 am
Location: San Jose, CA

Post by Roger Rines » Thu Apr 19, 2012 11:08 am

An important aspect to being successful with UA is to use the Generate Forward option. This option is needed to ensure the contracts are stitched together in the manner in which data was broadcast by the markets.

In this last version is a restriction that will prevent the contracts that were rolled from becoming the current contract again. In testing here I've found no instances where it would fail. However, the discussion about this change hadn't been marked completed at the time of this beta release.

In previous versions not all the calculation based roll options were enabled with the Generate Forward option, but that was changed for this version.

A lot of testing and checking went into weeding out the issues implemented so far, but that only means we didn't find what you might with how you'll use the software. Be sure to check the changes before using them in live trading.

Take some time to click on the new features. There is a lot of capability available behind some of those new features, not just in the displayed edit screen shown below, but also in some improvements made to the chart data display.
Attachments
UA_Beta-120_Edit_File_Screen_20120419.png
Updated CSI UA Edit File Screen
UA_Beta-120_Edit_File_Screen_20120419.png (28.02 KiB) Viewed 6582 times

Mats
Roundtable Knight
Roundtable Knight
Posts: 133
Joined: Thu Feb 28, 2008 2:59 am
Location: Stockholm

Have this affected your simulations..

Post by Mats » Mon Apr 23, 2012 7:36 am

Roger

Thanks for your input, this simulates real world trading more. Had those new options affected your simultation results?

Roger Rines
Roundtable Knight
Roundtable Knight
Posts: 1946
Joined: Wed Oct 06, 2004 10:52 am
Location: San Jose, CA

Post by Roger Rines » Mon Apr 23, 2012 10:29 am

Data changes most often will affect simulation results, and some orders. I build my trading files using my own software, so I never tested the differences in the way you asked your question.

I can say the process CSI's UA offers in this release is very close to what I've been doing by other means. Where it differs is in how I still look at comparative data profiles of price, volume and open interest to see the changes that supply, weather, and world events can cause in a market, and selectively take those decisions to adjust my roll timing. To some extent, CSI has provided a custom timing option that can be applied to a symbol that can override the main screen settings, or act as a fail safe timing option.

My approach is tedious. However now that the UA roll price adjusting math now has a roll option that is working correctly, it will be useful and interesting to understand how much of a difference there might be between the two methods so as to understand if the daily time investment cost of my approach has enough worth to keep spending that time.

An advantage the UA has that makes changed market data series easier is the Symbol Link feature. Leapfrog spent a lot of time with the CSI getting that feature working correctly so that it would build a data series like the EURO Currency series using information form the period when the DMark was dominant.

Jez Liberty
Roundtable Knight
Roundtable Knight
Posts: 123
Joined: Tue Nov 03, 2009 8:49 am
Location: London
Contact:

Re: Have this affected your simulations..

Post by Jez Liberty » Mon Apr 23, 2012 6:46 pm

Mats wrote: Had those new options affected your simultation results?
When reading this thread over the weekend, I was thinking of doing such test to measure the actual difference in TB back-testing results using different methods. I'll try and publish the results here and/or on my blog soon..
Roger Rines wrote:I build my trading files using my own software, so I never tested the differences in the way you asked your question.
I am actually wondering if there is a market for a back-adjusting software (independent from data vendor source). Is it something that you considered at the time of writing/publishing your application Roger?
(there have been a few discussions about such platforms on this forum, but I wonder if you considered it as an option for your software?)

Roger Rines
Roundtable Knight
Roundtable Knight
Posts: 1946
Joined: Wed Oct 06, 2004 10:52 am
Location: San Jose, CA

Re: Have this affected your simulations..

Post by Roger Rines » Tue Apr 24, 2012 12:19 pm

Jez Liberty wrote: ...[SNIP]
Roger Rines wrote:I build my trading files using my own software, so I never tested the differences in the way you asked your question.
I am actually wondering if there is a market for a back-adjusting software (independent from data vendor source). Is it something that you considered at the time of writing/publishing your application Roger?
(there have been a few discussions about such platforms on this forum, but I wonder if you considered it as an option for your software?)
A few people have asked me about the software over the last few years. In the past I've given pieces of it away, but my data handling process isn't user friendly because I didn't invest in the user interface. It would help me save time to make it into a simpler process, but I’ve been spending all my available time with TB programming and methods research that I wasn’t finding time to get the roll handling software interface improved enough for humans to use.

To get the software out of its primitive interface state and into something normal people might find useful, I've asked the programmer who worked for me in the fund to pick up the ball so the project can get back on track. He probably would be open to making it available, but I'll have to ask him if he wants to take on that support load.

If you send me an email I’ll send you a screen capture of the roll-setup screen, which allows the data to be rolled by various means with a date fail-safe. Once the error handling is resolved it won’t be far from being a good back-adjusting tool, but it doesn’t have the charting details, or the grid control I use to examine the profiles, and then make random roll timing changes, yet. This version of the software requires a TB license & CSI subscription because it is using the TB Futures Dictionary as its data dictionary, and UA for its data source. Different data source can be enabled, but I haven’t seen any other data vendors recently that didn’t mess up the volume and open interest fields, so there hasn’t been much motivation to make it work with other vendor ASCII files right now.

Are you interested in being beta tester when he gets to that stage?

jklatt
Roundtable Fellow
Roundtable Fellow
Posts: 79
Joined: Sat Feb 17, 2007 11:51 am

Re: CSI Beta 120

Post by jklatt » Fri Apr 27, 2012 5:49 pm

LeapFrog wrote:2) The custom roll editor is a powerful feature which now allows us to set ROLL BY DATE to a high level of accuracy, based on TRADING DAYS and based on any day of the month we choose to roll on. One example being now able to roll 'x' days before FIRST NOTICE DAY. You define when FND is, based on the exchange rules, then set the roll day to be 'x' days before that. For example, I've set my grains to roll 7 business days before the last day of the month preceding the contract month - the last day of the month being FND.
I spent a fair amount of time with Josh Reed at CSI to get this implemented. I explained the design, he implemented it and I ran quality control on my end to help Josh work through the bugs. Works great for me.

This feature will allow the user to roll on a liquidity event OR X number of user defined weekdays prior to a user defined "drop dead date", _whichever comes first_. You guys might want to try it out. I think UA users will really like it. It essentially replaces fixed date rolls AND liquidity event rolls and merges them both into one elegant solution.

The user interface could probably use some documentation. If anyone decides to try it out and has any questions, leave a comment on this thread and I'll try to help explain.

Tomaso
Roundtable Knight
Roundtable Knight
Posts: 121
Joined: Wed May 19, 2010 7:00 am
Location: London, UK

Post by Tomaso » Mon Apr 30, 2012 6:45 am

@LeapFrog

Could you kindly explain in a more extensive and as simpy as possible (maybe with some examples) how the new stitching procedure works vs the previous close close ?

Many Thanks

Jez Liberty
Roundtable Knight
Roundtable Knight
Posts: 123
Joined: Tue Nov 03, 2009 8:49 am
Location: London
Contact:

Post by Jez Liberty » Mon Apr 30, 2012 10:11 am

Tomaso wrote: Could you kindly explain in a more extensive and as simpy as possible (maybe with some examples) how the new stitching procedure works vs the previous close close ?
I have not downloaded the new beta and/or worked through a real example, but I think the following explains it (from the CSI link posted by LeapFrog):
CSI website wrote: Close new – Close Old Same Day, Roll Day Adjusted (preferred)
-Here the close of the new lead contract is compared with the same-day close of the former lead contract and the price difference is applied to all historical data including the date record of the contract change.New Close to Old Close difference is applied to all price fields, (Open, High, Low and Close) existing at the time of the Roll timing event. This pricing adjustment allows the new contract prices for the next record to be, seamlessly appended to earlier data without any contract pricing gaps. It also preserves the pricing relationships of all prices prior to the Close on the day of the roll by allowing for the execution of the roll during the trading session on the day identified as the roll day.
They also have an example with data in there (http://www.csidata.com/?cat=4) but I feel it's probably best (for me anyway) to try one roll with real data and compare the results with other methods. (I'll just make sure I backup my whole UA folder before upgrading - in case I need to revert to my current version)

RedRock
Roundtable Knight
Roundtable Knight
Posts: 941
Joined: Fri Jan 30, 2004 3:54 pm
Location: Chicago

Post by RedRock » Mon Apr 30, 2012 10:14 am

Had an error saying I didn't have some wrapper layer from MS on my machine. Perhaps too early for me...

Tomaso
Roundtable Knight
Roundtable Knight
Posts: 121
Joined: Wed May 19, 2010 7:00 am
Location: London, UK

Post by Tomaso » Mon Apr 30, 2012 2:03 pm

Thank you Jez.
I read the blog but I am not sure I understand the difference between old close close and new close close by reading the explanation provided.
If someone smarter can translate it for simple minds that would be much appreciated.

Roger Rines
Roundtable Knight
Roundtable Knight
Posts: 1946
Joined: Wed Oct 06, 2004 10:52 am
Location: San Jose, CA

Post by Roger Rines » Fri Jun 15, 2012 6:08 pm

Tomaso wrote:...
If someone smarter can translate it for simple minds that would be much appreciated.
Hello Tomaso,
I edited the CSI detail shown, so I have to take some responsibility for the limited explaination.

Old Close = Closing price of the current contract that will be closed out.
New Close = Closing price of the selected next contract that will be entered.

When the roll adjustment transaction is made, it happens on the same date as the user intended. Some roll timing selection do the adjustment on a day earlier, which isn't how your account will reflect the roll-forward.

Sorry to be late. I got way behind in some of my posting links.

Post Reply