The bad point with Tradestation : They are very US centric, no access to European or Asian markets data and it seems they are not considering it a priority(?) - maybe they don't know EUREX is the biggest future market in the world in term of volume....
I agree with you George. They should include european futures plus intraday historical data, not 20 years like american, but 3 would be very good. Is it really true? Can I backtest 20 years of historical intraday data on major american futures using TS7?
In my opinion TS7 just offers more data. There are hardly any new features and it still lags basic back testing features. For 500$ I can get years of historical intra day data on the web, I can use it on my computer and I can do as many tests as I wish. So where are the great enhancements?
Shankar has a good point. Key issues like: 1. datafeed reliability (quotes/charts keep freezing intermittently) 2. availability of DELAYED intraday quotes if you choose not to subscribe to real-time quotes (what is the use of 20+yrs of intraday data if you still have to subscribe to every exchange to get it; you will end up paying $300/mo for the delayed data!!) 3. ability to construct continuous contracts (for futures) the way each person would like to all these issues have not been addressed by TS 7.
I find this very important, too, especially since I found out that @NQ, @ES, @NQ.D, @ES.D contain some serious data errors. I notified TS about that recently. I was wondering if nobody else checks the continuous futures data for accuracy??? Erroneous data makes strategy testing almost worthless. For now, I will be backtesting on the equivalent ETFs and cash indices.
PXG: Could you please elaborate on the errors that you found on the E-mini continuous contracts with TS 6? Thanks. -kk70
I'm pretty excited about the list of improvements and features. The TradeStation discussion forums are pretty responsive to requests for feature improvements and additions. The features chosen for this upgrade must represent what TradeStation believes are most important to and most requested by their customers so in a sense the features both included and left out tell us a lot about their customer base. I like the improvements to the user interface. Finally, window linking! Adding macros and hotkeys is great, and I like how macros can be assigned to buttons and can be called from within EasyLanguage. And having a prompt line "Command Bar" that accepts typed commands should be powerful. Finally there will be a debugger for EasyLanguage, and access to the bid/ask spread from within strategies could be useful. Export to Quicken, Excel et. al. is long overdue and frankly is more of a catch-up feature but nevertheless very important. Support for multi-processors is pretty forward-thinking - how people have an MP machine now? But in a few years, they should be cheap. The Real-Time scanning features in the Radar Screen option sounds very good. I've been waiting for scanning. It sounds like this will cost extra beyond the basic platform but TradeStation's fees are competitive so maybe the increase won't be too bad. The big disappointment is in the lack of improvements to strategy/portfolio testing. Compared to the portfolio testing abilities of Wealth-Lab and MetaStock, TradeStation is way behind their competitors in this very important area. Hopefully this is coming soon. Leaving it out of this release must mean the other features are more important to their customers, which surprises me - hasn't TradeStation's strength always been in back-testing? Dale
sure, kk70, but itâs rather long: 1.) the daily closes of @NQ and @NQ.D should match, but they don't My e-mail to TS DataIntegrity: "Jawad, thanks for your prompt response. One more question: Is @NQ.D backadjusted? It has different closes than @NQ. If @NQ.D is backadjusted, than why the difference between daily closes on @NQ.D and @NQ? Thanks again, Patricia" From TS DataIntegrity: "Patricia, Yes. The @NQ.D is back adjusted as well. The difference is this (and it applies to all .D symbols): The .D symbol is synthetic, as are all the @, .D, and .C symbols. The @NQ.D daily bar is comprised of activity between 9:30AM-4:15PM and the @NQ (same as @NQ.E) daily bar is comprised of activity between 4:45PM-4:15PM. Basically, different session times. Best regards, Jawad Shah Data Integrity Representativeâ âJawad, thanks again for your explanation. However, it still doesn't explain why the daily closes between @NQ and @NQ.D are different. I understand why the daily opens are different, but why the daily closes? Since both sessions (@NQ and @NQ.D end at 4:15 pm), it would seem to me that the daily closes should be identical. e.g. 12/4/02, daily charts: @NQ: 1076.00 @NQ.D: 1073.5 Thanks, Patriciaâ âYou're right about the closes. It may be bad data. Send me the server details (file-preferences-tradestation network-click details). Best regards, Jawad Shah Data Integrity Representativeâ The bad data is not only on one server, I logged to a couple of them, refreshed my data, even deleted the Cache folder; check out for yourself. 2.) A more serious problem: rollover adjustments not calculated correctly: â Hi Patricia, TradeStation6 Continuous Contracts are back-adjusted to eliminate the gap that may occur as the front contract rolls to the next active contract. An adjustment factor is calculated by subtracting the close of the "old" active contract from the close of the "new" active contract on the specified roll date. The result is applied to all data prior to the new roll date. Over time, the cumulative effect of the applied adjustment factors can significantly alter the data to the point of negative "prices" in some cases. The value of the resulting data series is that trends are preserved at the expense of price. Each contract has its own rollover rule which specifies a different rollover date. Below is the formula using a mock symbol (XX with H M U Z as the trading months) @XX.P Del months: H M U Z 1st del month traded: XXM9, 1st day traded 990425 Roll over rule: 6th business day of delivery month Adjustment factor ( Distribution ) = close new del month - close old del month on the roll date 1st segment of data del month: XXM9 roll date: 990608 time period: 990425 - 990607 ( roll date minus one day ) adjustment factor = close XXU9(new del month) on 990608 - close XXM9(old del month) on 990608 2nd segment of data del month: XXU9 roll date: 990909 time period: 990608 - 990908( roll date minus one day ) adjustment factor = close XXZ9(new del month) on 990909 - close XXU9(old del month) on 990909 3rd segment of data del month: XXZ9 roll date: 991208 time period: 990909 - 991207( roll date minus one day ) adjustment factor = close XXH0(new del month) on 991208 - close XXZ9(old del month) on 991208 and so on up to current date. This calculation process and the roll-over date criteria that are used were based on methods used by other sources. Continuous contracts can be calculated in numerous ways. TradeStation plans to provide users the ability to create their own methods of calculation in the future. Your input is appreciated and should be directed to https://www.tradestationworld.com/discussions/default.asp in the Enhancement Suggestion section. Customized continuous contracts will be available in the future but at this time there isn't a set time frame. Best regards, Jawad Shah Data Integrity Representativeâ Here is only one example of what I believe to be an incorrect rollover calculation, or am I wrong? I am still waiting for Jawadâs answer: âJawad, here is another problem I have mentioned I will report to you: I believe there is a couple of inconsistencies in your continuous data series (@NQ and @ES). Here is one: Last rollover date for NQ was 12/12/02 According to your calculation explanation below the spread should be calculated as follows: On daily chart, 12/12/02: NQH03 - NQZ02 = 1044.5 â 1041.00 = 3.5 => then the difference on 12/11/02 between @NQ â NQZ02 should be 3.5, since from all the prior data before 12/12/02 you need to substract 3.5, right? However, on 12/11/02 @NQ = 1040.5 and @NQZ02 = 1035.5. The difference is 5! WHY??? If I am correct in my calculations and assumptions, then probably you should have a close look at the whole continuous data series for @NQ and @ES and also the .D contracts, since this is just one example that I found the data to be calculated incorrectly. Thanks, Patriciaâ Kk70, you may also check out: http://www.tradestationsupport.com/discussions/topic.asp?TOPIC_ID=825&pollresults=1 I posted a Tradestation data question here on ET a couple of weeks ago, but it didnât get much interest. This was after I got a couple of very confusing answers from TS DataIntegrity people, but before I got the answers above from Jawad: http://www.elitetrader.com/vb/showthread.php?s=&threadid=12165
PXG: Below is my interaction with TS6 Support and the explanation I got for non-matching data. -kk70 ******************************* Greg: For 11/20 (the day of the data disruption) I have the daily bar of ESZ02.D having the high as 918.75 and close as 918.75. However, the 60 min chart of ESZ02.D has the high as 919.25 and close as 919.00. I am not sure which is the correct one. Could you please see that this is corrected? Thanks, kk70 PS: I have already refreshed both charts many times ********************************* The Data Integrity department has given me the following reply: "The High has been corrected. However in reference to the close, there may be discrepancies between the intraday charts and quote windows and/or daily bar charts. This is due to charts displaying the ticks transmitted during the day (including erroneous ticks), while the daily bars (after running the daily data collection) and/or quote windows will reflect data refreshes. The difference results from the settle being sent later in the evening. The last tick (intraday) and the daily close will generally match until this is sent. The daily bar is updated while the tick (intraday) remains the same. The intraday usually matches the daily until the refresh is sent. After that, they may not match. A method is being researched where the user is given the option to accept or decline the refresh. We actually include 2 additional minutes of dribble into the last 1 minute bar of the day. But because regular trades come through sometimes several minutes after 4 p.m., we may in some cases not match the intraday and daily closes. Also, note that the last minute bar of any given day may not reflect the same closing value as the daily bar. The reason for this is that the minute bar reflects the last traded price and the daily bar's close reflects the settlement price. For an example, on 8/21/02 ESU02 per the CME, the last was 952.00 and the settle was 952.25. So in TradeStation, the last minute bar will show 952, but the daily bar shows 952.25." Greg TradeStation Technologies, Inc. ************************************