Beta Symbol Properties Window, Data Tab, Feeds

Sept 21 , 2005

16:37 @Ensign_HA: Welcome to today's training class.

16:37 @Ensign_HA: Over the past month, I have been busy with priority projects for the Ensign Windows program. We showed last week some of the upcoming enhancements, and will have a little review of them again this week. This version will be released as a Beta version for a few to test and provide feedback and then it will be released as the regular version of Ensign. One of the changes involves the Symbol Properties table, which can be accessed from the chart private property form by pressing CTRL-P keys.

16:40 @Ensign_HA: In this example, note that Sunday and Friday hours are available. Most often, these will be set the same as the day session and evening session hours for Mon-Thur. there are a few futures who have different Friday hours, including a Friday evening session. So I will consider this as 'handwritting on the wall' that the industry may be headed towards more futures following suit. Therefore, in advance of it becoming an issue, Ensign will at this time evolve to provide for unique hours on Sunday and Friday to accomodate.

Data Tab

16:42 @Ensign_HA: The next tab for the Data on the chart properties is shown above. On this tab you have the ability to select the source for the data refresh when more than one source is available. For this example, the subscription is for the Interactive Brokers feed, and the charts can be refreshed from either IB or from DTN Market Access. Other sources will be shown enabled depending on the symbol and the time frame, such as daily stocks would have 1st Internet Source enabled as well. Each source may have a particular limitation or advantage, and the notes below the selection are a brief reminder of what the refresh source offers.

16:45 @Ensign_HA: In this example for a 5-min futures chart for ES, the refresh can be from IB, and if IB is used the refresh from IB can update up to 1 week and do so without any delay requirement. Thus the data for the prior 10 minutes is in the refresh. For most users this would be a preferred setting for refreshing intra-day bar charts.

16:47 @Ensign_HA: Now if you happen to need or ask for a refresh that is more than 1 week and the selection is on IB, the program will switch to make the request from DTN because DTN has data for the past 4 months. You do not have to change the bullet selection, the program will just switch to DTN because of the amount of data that is needed.

16:48 @Ensign_HA: Also, the selections on Setup | Manager have been eliminated in the Beta and replaced with a single check box for Auto Refresh Charts. This is a mode that is either On or Off. When it is On, the program will be smarter to see the time stamp of the last bar on the chart when it is opened, and automatically make a refresh request that is big enough to bring the chart current. Thus we do not need Auto refresh selections for 2 days, 1 week, 1 month etc, which were often abused anyway. The program will just be smart and refresh the chart as needed.

16:50 @Ensign_HA: There is still the ability to right mouse click and use the Refresh pop up menu to request a big refresh if you see a chart that needs it, possibly the chart has a gap a month ago, that is not auto refreshed because the refresh brings it current based on the last bar's time stamp. So you can manually refresh a time period of your choosing, such as 2 days, 1 week, 1 month, maximum, etc.

16:52 @Ensign_HA: Another change has been to add Weighted as one of the average formulas to the Stochastic and Williams %R studies. Thus there is a new Weighted selection entry available in the Formula frame on these property forms.

16:53 @Ensign_HA: There have been a half dozen other small improvements hardly worth mentioning, but one is we redesigned the Open dialogs to overcome a windows platform limitation so we could initialize the dialog with the path last visited. This is helpful and convenient, but hardly earth shattering or even worth mentioning. Some may never have even noticed that the Open dialogs used when browsing for a path may have started in an illogical place.

Default Settings

16:55 Josh: What about the ability to reset to the default settings?

16:55 @Ensign_HA: Josh, is your question specific to the check box for the Show 1st and 2nd lines? If so, that has been improved on in this Beta version as per the e-mail from you16:56 @Ensign_HA: Josh, I think your question is more related to what is restored when you click the Cancel button if I remember your e-mail feedback.

16:57 Josh: No. On all studies. If I setup something and change my mind, I no longer have the default settings in my head or otherwise.

16:57 Shar: Good question Josh.

16:58 @Ensign_HA: On a property form, you can click the Default button and it will load a file that was asked to be the default. If you are editing that Default button when selected, then you are defining a new default. If you change your mind, click the Cancel button which aborts your changes and closes the form.

16:59 Shar: When I first set my Gann box to the Alan box setup, I made it a defaut. Now I have another A Box and now don't know what the real Gann box is suppose to look like since I changed the default.

16:59 Josh: Correct and if I edited it I am stuck.

16:59 @Ensign_HA: There is nothing magic or special about the default. It is just whatever happened to be the setting I or Kimball set up to distribute the program with...truly nothing special and if you have thought they were special, you give us more credit than we deserve.

17:00 pw: lol

17:01 @Buffy: also pw others have asked this too and it is easy to google the study and get the "standard" along with lots of other information on it.

17:03 Shar: Thanks Buffy for answer about where to get the default back.

17:01 @Ensign_HA: So go do your own settings, and they will be as worthy as my choices or preferences.

Newsletter

www.ensignsoftware.com/tips/tradingtips68.htm

17:01 @Ensign_HA: Scroll to near the bottom of this newsletter shown above to see a picture of yours truly and Larry Pesavento

17:02 @Buffy: :-)

17:03 @Buffy: I have been asked a couple of times if you have any idea when another seminar might be

17:03 @Ensign_HA: Larry is having a seminar in October, I think in Tucson. I am not attending. It is a seminar on his trading methods. I have not thought of another seminar yet since I have been super busy with the IB refresh project and redesign of chart properties form.

17:01 Islander: Is the weighting on the stochastic properties window a volume weighting?

17:02 @Ensign_HA: No islander, it is time weighted with current bar having most weight - older bars have linearly less weight

17:02 Islander: ty Howard

17:05 pw: I have a question regarding IB refresh, when we are open for questions, thanks

17:05 @Ensign_HA: Now, I would like to have a little training about the IB refresh

17:05 @Ensign_HA: pw, ask now, probably on topic

17:05 pw: ib has come out with some fantastic market stats, can I chart these in ensign? I understand i'm not the first to ask this, but it is very useful

17:06 @Ensign_HA: pw, until I learn otherwise, I have not found a way to get TWS to issue the stats as a data feed. I do not see how to enter them as a symbol on the TWS quote page nor do I know what IB might be expecting when I try to tell TWS I want to watch the symbol and have it send data on the data feed. The stats page seems to be a separate entity in TWS and not available as part of the feed exposed to 3rd party applications. Possibly it is, and I just have more to learn. Until then, I do not think we can chart the stats you are referring to.

17:08 pw: Thanks, perhaps they will change that down the road - ty

17:08 @Ensign_HA: ok

17:09 Josh: You mentioned you could give a brief overview of DYO eg for determining the number of days up vs down etc.

17:09 @Ensign_HA: let me tell more about the IB refresh.

17:10 @Ensign_HA: There is no weekly or monthly refresh from IB so refresh of weekly and monthly charts will continue to use the DTN Market Access resource. There is no daily refresh either from IB, but for daily Ensign's implementation was to request large time frame intra-day bars, such as hourly, or 30-min or 15-min bars and extract the daily range, and sum the tick volumes to create a daily volume.

17:11 @Ensign_HA: Since the intra-day refresh is limited to 1 week, our IB daily refresh implementation will be limited to 5 days. The one advantage is the IB daily refresh can generate a daily bar for today.

17:11 @Buffy: an ib user was told by IB tech that they are going to limit the amount a user can refesh a day is this true?

17:12 @Ensign_HA: possibly, buffy, I have not heard or read about any such limitation. Fortunately, we have the DTN Market Access as a resource too for IB users.

17:12 @Ensign_HA: On the intra-day refresh, the quality and speed of reception seems excellent. The intra-day bars have identical volumes when compared with bars from other data feeds, such as eSignal and IQFeed. So, I feel very good about IB intra-day bar refreshing. Again, IB has the advantage that the refresh is through the current time..... there is no delay requirement.

17:14 @Buffy: Hi Larry

17:14 @Ensign_HA: Thus if you have a 1-min chart, experience down time, get back on line, the refresh brings you current right now.

17:15 @Ensign_HA: Hi Larry, scroll to the bottom of that newsletter to see our pictures.....taking a poll on who is better looking,.

17:15 @Buffy: :-) now those monitors behind you guys........... :-)

17:18 pavoni: I guess size DOES matter

17:15 @Ensign_HA: so far, have had one vote for the guy with all the hair....

17:16 @Ensign_HA: As mentioned, if the refresh manual request or auto sizing is for more data than 1 week, then the DTN source will be used automatically because up to 4 months of intra-day data is available.

17:16 pavoni: How much data is refreshed when we have "auto refresh charts" box checked, i'm refering to intraday tick charts

17:16 cc_k: only issue is with refreshing tick/volume bars, right Howard

17:16 @Ensign_HA: Now for charts based on ticks, which are Constant tick, Volume and Range charts. For these charts, unfortuantely IB does not have a tick by tick refresh like eSignal, IQFeed, and DTN Market Access have available. IB uses time slices, and the time slice that is closest to documenting individual ticks is the 1 second time slice. So our implementation for a tick refresh from IB is to request the 1 second time slice refresh. This time slice is like a 1 second time bar, and it returns 4 prices and a total volume for the time slice. Obviously if the 4 prices are not all the same, there MUST be more than one tick present in the time slice. If all 4 prices are the same, we do not know if this is a single tick or multiple ticks at the same price.

17:20 cc_k: I see

17:20 @Ensign_HA: We to be honest, we never know how many ticks might be present in any 1 second time slice, so we had to make some assumptions or achieve good approximations when compared with other data feeds, and when compared with the original ticks collected by IB. The assumptions that we have settled on as 'our best effort' to reconstitute individual ticks from the 1 second time slice refresh are these: If all 4 prices are the same, we will assume it represents a single tick. This first case of 4 equal prices happens about 70 percent of the time. IF there are two prices present such that the open does not match the close, then we will assume there are 3 ticks present. We know for sure there has to be 2 ticks. This type of tick happens in the feed about 8 percent of the time.

17:24 @Ensign_HA: When there are two prices present, and the open equals the close, will assume there are 4 ticks present. This happens 2 percent of the time, and we know for sure there had to be at least 3 ticks present. Now by starting with the 70% and then adding the 8% and the 2% we can account for sure about 80% of the original ticks. By inflating the last two categories with an added tick, ie 3 instead of 2, and 4 instead of 3, we get another 10% to around 90% of the original. and that is where I have settled in on for my assumptions.

17:27 cc_k: Sounds reasonable

17:27 @Ensign_HA: Yes there are other possibilities than the ones described above, but they happen so very occasionally, like maybe once or twice in a set of 2000 ticks. So they basically are 0% for the other possibilities. I did not dare inflate the tick count for the 70% group and I did not want to have only 80%. So I inflated by one tick for both the 8% and the 2% groups, to pick up another 10%. Now, 90% to me seems better than 80% for tick counting.

17:29 cc_k: yep

17:29 @Ensign_HA: I am sorry there is not perfection here, which would only be possible if IB sent every tick in a refresh request instead of this 1 second time slice thing but we work with what is available. So for the record, if you refresh using IB as the source a tick or volume chart, there WILL BE DIFFERENCES in the resulting chart when compared to the original. The tick chart will have fewer tick bars than the original.

17:30 cc_k: Yeah - their position is that the data is really courtesy to customers primarily to facilitate order related things -

17:30 @Ensign_HA: If you refresh using IQFeed, then you had the opposite problem of showing MORE bars than IB originally collected. The volume charts also have refresh issues that there seems to be less volume than originally sent, and I cannot account for why that is so in the IB refresh. The volumes in the time slicing does not sum back up to the orignal. Have pulled my hair out, and cannot make them match and I have lots of hair to pull....lol

17:32 @Ensign_HA: I think it might be an issue of how the TWS is or is not doing their volume. For example I tracked down a comparison where the original had 2 adjacent ticks at 5000 each. Yet in the time slice refresh, there was only a single 5000 reported. So the original appears to be inflated. This could happen in that the TWS issued the 2nd tick, but did not issue a companion tick volume at the new value. Thus the prior tick volume value of 5000 was to be assumed the tick volume for the new tick. This they do to conserve sending values that do not change. If the tick volume last sent is 100, and they have multiple ticks all at a tick volume of 100, they do not resend the 100. They only send the tick volume when it changes.

17:34 kongfuzi: may I ask a question?

17:35 @Ensign_HA: ask kongfuzi

17:35 kongfuzi: Ignoring IB feed, do all the other feeds agree when compared?

http://www.ensignsupport.com/email/361.mhtml

17:37 @Ensign_HA: Read that post from our Knowledgebase - answer is No. For example IQFeed does not match eSignal, as per the article discussion above

17:37 cc_k: yeah I think a lot of data providers are confounded by the new world of high frequency, computer-driven trades that can go off incredibly fast. Many large firms have found even what were once considered great feeds to be inadequate and so have developed their own in-house ticker plants connected directly to the exchanges

17:37 kongfuzi: OK so the problem is a matter of degree with IB being the oddest (generally)

17:37 @Ensign_HA: The example given was eSignal might have 3 ticks with volumes of 10, 3 and 2 on their feed and the IQFeed and IB feeds might have single tick with volume of 15. I do not know for sure why that is, but have offered my theory in the article.

17:38 kongfuzi: ok so there is "variation" among all feeds

17:38 @Ensign_HA: Correct kongfuze. There are feed differences and the one with the sampling of the feed is IB. However, they do a very good job in my opinion and are a great free feed. I would use it any day to trade from. The issue is for the tick worshipers who compare charts with others and see differences in bars, and studies and think that something must be wrong, broken, inferior, etc.

17:40 kongfuzi: man, not the tick worshipers, again ;-)

17:40 @Ensign_HA: if you would quit worshipping ticks, then you will be just fine. If you use Range charts or intra-day bar charts, then you have a very high likelihood of matching charts with other vendor feeds.

17:41 kongfuzi: great thx for insight and clarification

17:41 @Ensign_HA: welcome, seems to be a point needing discussion or clarification often

17:42 @Buffy: With many using tick charts in bline with the different feeds and comparison charts done and seeing signals still there the worshiping isn't so bad now - IMHO

17:44 @Ensign_HA: Thx for feedback buffy. You for one would be well aware of feed comparison going on and from my point of view, all are great feeds and none of them are the gospel truth.

www.dacharts.com/faq/Comparable-tick-charts-by-feed.htm

17:44 @Buffy: See the chart comparison at bottom of that link. It shows signals still there. It has helped everyone relax over it and we do run a 3m too

17:42 @Ensign_HA: In summary, the IB feed is a great feed, but it will be different than eSignal, just as IQFeed is different than eSignal and different than IB. The only two feeds that seem to be quite identical are the DTN satellite and the IQFeed feeds, and that is probably because they both originate from DTN

17:43 cc_k: yes, everyone should understand that IB feed is not a true 'feed' in the sense of one you're paying for using to power a high-end analytic app without some misgivings since it's desinged to power their own TWS which is an execuition platform

17:42 pavoni: Howard, do you have a mega historical file for a continuous ES futures chart? daily, weekly, etc

17:43 @Ensign_HA: pavoni, yes, about the ES mega file, click the Internet Services button and use the 1st tab, History drop down box to download the ES #F file.

17:43 pavoni: thx Howard

17:45 @Ensign_HA: All one has to do is visit an open outcry pit at the exchange to realize that a feed is a sample of what really goes on. It is not all that goes on, it is not perfect, it does not represent any absolute such as the order in which trades happen. It only represents what the watcher calls out of the pit and is subject to his watching and the order he calls what he sees. Also, why is there a need to reconcile order decks after the session closes? Because it is not a perfect system.

DYO - Count Up and Down Bars

17:47 @Ensign_HA: ok, now lets move on to Josh's DYO question - ask again JOSH what you want help with and I will try to whip out an example to help you

17:49 @Ensign_HA: I think he wanted a DYO that counted Up bars and Down bars. So with that basic idea, let me whip out an example. The Up bar I will assume to be one whose close is above its open and the Down bar one whose close is below its open. This is the basic Close vs Open color bar study used by candlesticks

17:51 @Buffy: or dunnigan up/down bars

17:51 @Ensign_HA: I assume he wants some kind of balance ratio of which number is higher, like lots of Up bars or lots of Down bars. So lets invent two GVs that represent the count for UP and a count for DN bars to make this a plottable number instead of an ever increasing number. I will assume it is more useful to do the counts across a fixed number of bars, such as a period. How about we choose a period of the last 20 bars? Thus the max Up or Down counts would be 20 and the minimum counts would be 0 and then all numbers in between 0 and 20 would be possible. Would plot the two counts as curves in 2 different colors, kind of the like the Directional Movement study's Up Index and Down Index.

17:54 @Ensign_HA: This is the proposal I am making since Josh is not here to guide me on this. Does it sound like a good DYO idea, or are there suggestions for change before I whip it out. If you want a change, say so now please.

17:55 @Buffy: The 20 could be changed?

17:55 @Ensign_HA: Buffy, we can try to implement that - do not know if I have talked myself into a corner or not

17:56 @Buffy: :-)

17:56 @Ensign_HA: By days, I translate that to bars on a daily chart. Would also be useful for counting up and down bars on any time frame chart. Ok, you all chat for a couple minutes which my attention is diverted to this DYO example. And then we will finish the class with the example and call it a day.

17:57 pw: Buffy, was he requesting a simple HH LL count, similar to box counts, I think quartz did those on box charts, it's the 10mAbarCnt.prg which is on dacharts/templates

18:01 @Buffy: yep - not sure that is what he wants but we can find out. This dyo is going to be interesting also

18:02 pw: This is where we all say.... cheeky devil

18:02 @Buffy: look at all the colors on that one of Mike's

18:17 @Ensign_HA: Hi, back..... sorry it took me so long but I couldn't find the Count function because I kept looking in the Value categories instead of in the Flag categories for it.

18:18 @Ensign_HA: In this DYO example, Line A will use a set of 20 GVs to hold the last 20 flags for when Close > Open

18:18 @Ensign_HA: This set of GVs will be 10-29. For each bar we will begin by rotating the 20 set rightward to make room for the next answer to be entered into the set by Line B which puts its new value in the beginning of the set at GV[10]

18:19 @Ensign_HA: Then Line C counts thenumber of True entries in the set of size 20 beginning at GV[10] and plots this True count

18:19 @Ensign_HA: LIne D, E and F are mirrors of A, B and C to control a 2nd set of GVs from 30 to 49 with the new entry for counting the Close < Open cases posted by LIne E done by Line F and plotted in a different color. As expected, typically an increase in the color of one line will be a decrease in the other line value. There are some bars where Open=Close and in those cases neither line counts it.

18:22 @Ensign_HA: To make this example more generic, the GV arrays should be moved to use private GV variables owned by the chart but since the private set is 240 to 255 we have arrays in our 20 example that are too big for the private set. Private set would be OK if the counts were for the last 7 instead of last 20.

18:23 @Ensign_HA: Anyway, Josh can take a look at this suggested implementation and have a headstart when he returns. Any questions about it??

18:23 pw: Thanks Howard

18:24 @Ensign_HA: Maybe we are looking at the holy grail and don't recognize it.....lol

18:24 @Ensign_HA: Thanks for attending class and you all have a good week trading...have a good evening....bye.

18:25 pw: LOL, thanks for the information and have a good evening all

18:25 dblue: Thanks Howard

18:25 @Buffy: Thanks Howard - have a great evening