After installing R18.1 (I now skip .0 Releases) I experienced the whole kit and caboodle with repeated crashes to boot.
I had achieved a setup that gave me what I needed, to which I cautiously added:
- a 3rd-party add-on to locate the mouse on multi-monitors (now removed, although it caused no problems and removing it didn't improve matters, and find 'active crosshairs' in R18 near useless in practice),
- a Ninja add-on for shading Sessions and Dynamic Range (Repeater), and
- Hurleydood's TickRefresh (set at 10fps, fit for my purpose, stock Ninja downgrade from NT7 to 4fps isn't)
That Workspace worked fine for several months.
Release 18 introduced crashes galore, 'volatility-induced' but not only during extremes, which I sent to Patrick. Remove add-ons etc. which I did, one by one, to little avail - still lagging at the Open etc. albeit a little less so but crashing. Crosshairs, OnRender, SharpDX etc etc.. errors.
My Setup is very 'light' on indicators or anything else (everything stock Ninja - OHL, Vol (not Vol Profile, just plain old vanilla volume, I'm a simple human) & PriceLine - that's it, nothing else, nada, zip, and follows ALL the performance best-practice guidelines. The only way it could possibly be viewed not so is that I have several instruments open concurrently hence charts, Data Series etc - the common Futures, and I must use 3 Data Series per instrument but that is ENTIRELY due to a limitation of the stock platform i.e. inability to segregate RTH and ETH simultaneously. If I could do so with stock platform and one data series I would. The multiple data series solution is Ninja's solution, not my bad. EVERYTHING is stock platform. During this exercise I saved a Workspace with only US Indices for use at the Open. So YM, ES, NQ and RTY. 15min, 1min & fast tick for each displayed (there are others in tabs - the point is 4 instruments and not lots of 1-tick data series set at 'oneachtick' update. As I've already stated, following the Performance Tips guide). Still problems.
PatrickG - "Just to add some context, in all cases we've obtained files for and worked on individually so far, most cases break down to workspace complexity/custom add-ons along with market volatility. For example, any custom add-on that uses tick replay or in general relies on a 1 tick data series will cause heavy loads on your system and this is exacerbated my high market volatility."
PatrickG - "When I use the ChartLagTimeV3 on my end, I never receive a delay of more than 50ms."
Patrick took my log and trace and installation files and had QA look into my situation with Development. Whilst they weren't able to pinpoint the issue, they did offer specific Support/Development interventions to try to identify what was causing 18 build to trash my setup that had worked fine in 17 even with the add-ons, yet lagged and crashed in 18 with or without them under high volatility i.e. completely bare-ass naked stock platform. Ninja told me they have made some changes to some code that make it more efficient and may contribute to correcting my problems and asked me to try a pre-release build.
I've been using it this week and have not been able to break R19 with my setup - and I've tried hard doing things I know are likely to end in tears (mine)
Oh, and now back running at 10fps on my fast tick charts (only), having added TickRefresh again.
Rock solid, no chart/DOM lag perceptible even at the Open and Close. For e.g. NQ has 15 tick, it flies in sync with the DOM and smoothly at 10fps.
For example I placed Orders on the DOM today at the reversal at yesterday's overnight low where volatility was mental - the highest of the day at >4000 on the 1-Min. (I use this example wrt comments that thinner fast(er) and farther-moving may be worse than absolute volume throughput such as ES, for example, and I didn't place a trade on ES at that time!). Simply flawless performance on chart, DOM and execution.
So the end of my tunnel is currently full of light.
It is a positive note re: the Support offered.
It is a positive note re: it looks like R19 may help others with their issues (unless they youknowhat it up by the time it is released, per 18 over 17).
It is negative that a major Release turned a successful Setup into an un-useable one
It is negative, hence my dismay at stumbling upon this thread, that throughout I got the 'no-one else is experiencing the symptoms you are reporting' mantra.
I don't need some technical response about how the error codes weren't the same as reported in this thread or whatever. Given the experiences reported in this thread, it is disingenuous, albeit 'twas ever thus.
It is pretty clear - changes introduced in 18 resulted in performance reduction wrt the platform's ability to cope with high volatility, exacerbated lagging etc. Whether this manifests itself as Chart/DOM discrepancy, overall lack of responsiveness and/or lag, DOMs, charts, Order Entry whatever, obviously further exacerbated in extreme conditions such as current
Whether or not you're on R18, whether it contributes to your specific issues or not, hopefully changes to R19 will alleviate some or all of your issues. Here's hoping...
This is the core platform with very light load setup and zero add-ons that was failing in current market conditions, where R17 didn't.
"If we can make a definitive statement like "NinjaTrader performs normally when I remove indicator X', that allows us to troubleshoot very specifically. For example, contacting that developer to see what, if anything, can be done to optimize their add-on for high volatility scenarios, or perhaps even combinations of settings that were unanticipated."
Look first unto thine selves….
"There are effectively infinite unique ways to configure your NinjaTrader workspace, so we need to gather specific information to learn how to best mitigate the symptoms you're experiencing based on your configuration."
All very well but a platform should be sufficiently robust such that it operates without fail under the majority of configurations for which it is designed to be used. No-one here as far as I can tell, is trying to do anything unusual and without a reasonable expectation it should work without failing and specific tailoring. Certainly I'm not. Quite the opposite.
All this points me to concur with Erwin Brewers -- Releases need more severe and extensive real-world application stress-testing before release. Curiously, or perhaps not, I don't recall ever seeing any insight given into what Ninja's testing entails - just the 'we've tested internally.... we can't replicate your problem' etc. Would be interesting to know what the setups are used. Perhaps users could help with feedback. There's far more experience out here than in there
Chins up fellow Ninjas, hang on in there for R19.(point 1!)
Kind regards,
Bruce
Comment