What is the lookback period on your charts? And other than the DOM indicators, do you have any other ones? I'm assuming you are only using time-based charts, since those are the fastest implentation by far on NT - range and volume charts are slower. Anyways, 5 charts is not a lot - we should be able to run with 10-15 charts for a properly designed piece of software.
I seem to recall, search around if you want, that it was recommended to disable DB and record if you can afford to, as these eat CPU too. I only have one chart for now, but would like to have 3. I know conventional wisdom says multi core shouldn't matter for single thread app, but everyone I know with a quad and max RAM has reallly no issues. One of them has I7 OC'd to 3.3 with 12 gigs or RAM running W7, previously on Vista.
I actually use 4 volume charts and one time based chart. I do have other studies loaded as well. The charts look back 3 days.
I'm going to add my two cents here, because I also use NT/ZF and it may help others make informed decisions. I currently use a Dell Dimension 8400, 3.4g, 2GRAM that is a few years old at least. Connected to Road Runner in the NYC region and running the following while monitoring both the ES and YM feeds: 5 - 6 tick and volume based charts with varying lookback periods from between 5 and 60 days. DOM T&S 5 relatively simple strategies. A number of canned indicators. My PC is relatively clean and I run CA Antivrus (no firewall) in the background. I monitor the feed daily and I can count on one hand the number of times where the data appeared to have frozen in the past two years. I can say that it performs on a tick by tick basis for me 99.8% of the time. Occasionaly, I can see my CPU usage jump up towards 100% during extremely heavy market volume, only to settle back down in the 5 to 15% range right away. The app. itself will average about 275mb when running all loaded up. I am not trying to start a war here but it may be possible that ISP and ZF server location and/or compatibility may be an issue for some, in addition to having some heavy computations going on in the background. I sympathize with those having problems with the platform/feed, but has anyone else ( aside from those who have said so) tried running alongside a trusted application or data feed for comparison purposes? I would be interested in hearing more about that. Ninja has it's quirks as I am sure most other front ends do, but in the end it's all about the platform that works best for the individual trader.
Unless you try to trade as a short-term scalper, the only way you are going to notice the lag that NT is suffering is if you run another copy of it side-by-side with nothing but a time and sales ticker, or maybe a simple chart with no indicators. It's possible that you are actually suffering lag, but just don't notice because you aren't trying to enter/exit trades at those crazy times. And when things quiet down, it catches up. Are you actually trading during the craziness with that setup?
Good points. Actually, I have run a basic chart alongside one of my own templates and the truth is they both appear to populate at the same speed. So, I guess the question is can it be the data feed itself that is lagging, assuming that NT is building the chart at a satisfactory speed? Yes I have been unfilled on limits in fast markets, only to enter at market price, close enough to the DOM listed price not to suffer, usually on point. During "craziness" as you put it, well, I am either already in that, or not entering until my next opportunity. Normal market price action always fills limits or better and market orders get bought at ask. The data seems smooth and charts build nicely. And, yes, I suppose you could be correct in saying that it is lagging and I don't even know it but that is why I asked for others' experiences with running alongside another feed.
NT updates everything in a single thread. So if you have 10 charts open full of indicators and one simple chart open with nothing on it, the simple chart will be limited by the processing on all the other charts, and will update at exactly the same speed. You can set up another computer and install NT on it, or run a second copy in a virtual machine with only a time and sales or only a single chart in it. I think if you do that, you are going to be unpleasantly surprised at how much lag you actually do have during the heavy times on your normal setup, and then you will see how much money it may be costing you. For scalpers that lag is a killer, and it means that you can't run with more than a few very short-range charts without too many indicators. NT 7 is supposed to fix this to some degree.
I have a bright idea...how about just switch to a multi-threaded application! Why hack around Ninja's poor design?
Because those total jerks promised us a beta of their brand-new much-improved version 7 at the end of June, and then royally screwed us about 1 week from that date, saying "guess what, now you have to wait until September!". If I had known 6 months ago I would have switched, but they kept stringing us along. This time it really is coming out in a month though, so I'm going to give them this one last shot. If NT 7 beta doesn't solve the issues, I'm gone. It's not worth switching and setting up a new platform at this point with such a short time to go until all the supposed improvements.
For clarification⦠- NinjaTrader is a multi-threaded application however NT7 will support multiple-cores - NinjaTrader DOES NOT process everything in a single thread although it is accurate that all UI components are updated on the application main thread which is a hard requirement that canât be changed - The majority of our market data UIâs are not timer driven (which smoothâs out and reduces CPU performance at the expense of purity) but tick driven meaning, each incoming tick will update a UI. The benefit of this is that you will see the market data event as soon as NinjaTrader receives it from the market data provider. The downside is that it costs you more in CPU usage. On a snap shot or throttled at the server data feed, the CPU performance can be seriously lower than on a feed that is unfiltered. The latter will impose higher CPU requirements since there is just more data to process. - We have made various performance improvements across many areas of NT7 and CPU load is one area that will be positively impacted. Those running unfiltered feeds will benefit the most. - Yes we are late on NT7 and yes I received some black eyes for it and they have since turned purple. I even hear some people have called me a jerk For those interested in what we are doing with NT7 you can read more at the link below http://www.ninjatrader.com/webnew/NT7/NinjaTrader7.html Ray