I have been noticing that from time to time the bars on my 233 tick chart will change between times I open ninjatrader. I am using the CQG demo data at the moment. I saw the post on the MDP 3.0 migration but this seems to be something different and more subtle but maybe not.
For example I took a trade this morning at a bar that closed at 7:00:07. When I opened ninjatrader this evening to look at the charts omy two 233 tick charts were different from each other. (I have one set up with Heikin Ashi candles and one with normal candles) And neither of them were what was shown this morning. I disconnected from CQG and "repaired the database" after reconnecting and restarting the two charts matched each other but still neither were the same thing I saw this morning. I can tell because I took a screen shot this morning and because I recorded the time stamp for the bar close on my trade. The closest thing I could find was a bar that closed at 7:00:42. The disconcerting thing is that the chart I see now does not show a trade entry.
Ninjatrader replied with the following:
"Thank you for your post.
While using a tick interval chart, the chart's bars are going to refresh, which will cause new and different sized bars to appear at different times. (As ticks come into NinjaTrader in real-time, they are time stamped based on your local PC time if they do not already have an associated time stamp that is provided from the real-time data source. The majority of our supported brokerage feeds DO NOT time stamp ticks, where most of our supported market data vendor feeds do provide time stamped ticks. NinjaTrader then builds bars based on the time stamp of the incoming tick and displays these bars in your chart in real-time.)
Please see the link below for more information on "How Bars are Built''; see the section on 'Understanding why a chart can look different after reloading historical data from the server.':
So their reply was that bars are going to look different when refreshed with historical data when the program is reopened. (if I am reading that right) That seems pretty unacceptable to me. There is no way that a professional could deal with two totally different charts between live and historical data. I could accept a slight difference in time stamp but bars with OLHC totally different from one another is simply broken charting.
Does anyone have any experience with this? Am I crazy to think that's unacceptable? If I do accept it, that means a trade set up I see live I may never be able to revisit. It would be a one and done deal. It also means any form of back testing would be totally invalid. no?
I find that hard to swallow. There is no way a professional trader would be ok seeing a chart only once. Meaning the way it looks when he trades it is potentially the only time it will look that way. This would negate any data you get from back testing. In addition any trades you took because of the price action live would disappear never to be reexamined again. (like they did for me yesterday)
Well it changes a bit the chart but not THAT much. Professional traders doesn't pay much attention to how bars are, but to what they're telling you (HH and HL or LL and LH, which doesn't change). In the case of algorithmic trading it's common some kind of noise filter.
Maybe I am not explaining myself properly. For a set of candles yesterday the OLHC were all different. I dont mean a tick here or a tick there. I mean the relationship from one candle to its neighboring candle changed. This HAS to matter. People base entries and exits on candlestick relationships. For example buy with a close above the previous candles high. People also make decisions based on indicators which are based on the OLHC of the candles. For example, yesterday I had a turn on the stochastic from an over sold position. After the change on the chart that turn didnt happen for another couple of bars.
So I am trying to verify something as is relates to this issue. As far as I can tell this seems to be a problem specific to ninjatrader and the way it handles the incoming data and time stamps. I have been searching other software forums for the same issue and haven't come up with anything. Can anyone confirm this for me or if you happen to be a user of another piece of software and you know this is an issue that would be helpful to know.
I just came across this thread, so this is a late comment, and pretty much in line with what @Malthus wrote.
I have seen the same thing with Ninja for both tick and volume bars. The other day I had a real-time chart that had some noticeable differences from one that I had posted to futures.io (formerly BMT) just an hour before; I had forgotten the issue with historical data.
Basically, what's going on, as I understand it, is that NT supplies historical data from their data servers when you first connect, and then during the day the chart is updated by whatever feed you are getting from your data provider or trading broker. So you have two sources of data on the chart. Aside from whether there may be data differences between the two, when a tick or volume chart is built, the bars depend on when the accumulation of ticks/volume starts -- unlike time bars, which always match up based on time, tick and volume bars are somewhat artificial in that they are built from transactions, and when you start counting matters. So it is not surprising if the bars that are built from the two sources do not coincide. You don't see that during the day, because you just see the chart as it is presented. But when you log off and then log back in, NT will repaint the chart entirely from its historical data, and if that is not in sync with what came from your real-time provider (which is likely), then the chart will be different.
As to whether this happens to any other platform, I can't be sure of all others.... It does make sense that if you have the exact same situation -- real-time providers not providing historical data also -- then you would likely have the same issue.
It is likely that this is more a data provider issue than a platform issue. You may want to check around for what the different data providers do. Bear in mind that a dedicated feed comes at a cost, and is not just what comes with the broker or NT feed.
You can be sure of never having that problem if the historical and real-time data are always from the same source. I know, for instance, that you won't have that issue with the ThinkOrSwim integrated platform -- but the TOS data is pretty crummy, bundled and dropping ticks, because they aren't in the data business -- or at least it was when I used them a few years ago.
Basically, this may be something to live with. The more each individual bar matters to you, the more this issue may matter. The more you are looking at larger movements, the less it will matter. But remember that you can't even expect two different traders' tick charts to always match up anyway. When you look at tick charts for the same period that are posted by different traders, there often are small differences in the bars, because the bars don't actually independently exist, the way time intervals do; they are summations from a certain beginning point, which may be different. That's a different issue from the one you have, but it's closely related, because if the bars different traders see are different, the price action on their charts is different, just as the price action on your chart may be different when the bars are refreshed from the historical data. But if you're not looking at very fine detail, the difference is not large. There were long discussions in the old PATs thread that @Malthus linked to on the subject; this is pretty much in the nature of tick and volume charts.
Everyone who uses tick or volume charts eventually has to understand and accept that any two tick or volume charts of the same span of time may not be the same. That's why, when your chart is redrawn from your historical data source when you log back in, it may not be exactly the same as it was in real time. Usually the same basic trend and chart features will still apply, however.
Last edited by bobwest; August 1st, 2015 at 08:55 PM.
The following 3 users say Thank You to bobwest for this post: