It is a really cool package—LightweightCharts.jl. Thanks, @g-gundam, for providing all this info, and many thanks to the author of the package. As an alternative, my sincere advice would be to set up a database and use Grafana or Pulse (TimeStored). I know it’s not pure Julia, but especially with respect to the speed of operations, it’s a different story. And you can visualize a lot. Seriously. :- )
May I ask if you would consider adding a custom online indicator to your package?
Would such an idea be feasible? To be honest, I’m not entirely sure what this might involve. For example, just off the top of my head, some basic ideas could include: smoothing with LowLevelParticleFilters
; instrument selection with CepstralCoefficients
or CausalityTools
or OnlinePortfolioSelection
; dynamic data assimilation with DataAssim
and with the Julia port of FourCastNet
; prediction with SymbolicRegression
, AnalyticComb
, and HiddenMarkovModels
; state vector correction with least squares
; execution with SPSCQueue
and Lucky
…
What do you think, does it make sense?
Thanks @j_u about Pulse (it seems to be an interesting dashboard app alternative to Grafana) but I think it’s a quite different story than what I’m saying about a pure Julia app using an immediate mode plotting (ie not a web app).
I don’t consider adding stuff outside of “classical” technical indicators in this project which is essentially a port of talipp. But you can create new Julia projects for each ideas you mentioned (if they don’t ever exist). I will look at them and maybe, if I can help, contribute to your projects.
The problem of execution of trading activities should also be considered in an other forum post.
Yes, I fully agree; it is an entirelly different experience with Pulse, and particularly with Grafana. As I mentioned, this is especially true regarding the speed of implementation and, frankly, in many other areas related to capabilities. I brought this up to ensure other users consider it. To be clear, I appreciate all Julia plotting packages, including LightweightCharts.jl. However, I find the Pulse/Grafana setup significantly more convenient.
I will certainly consider your suggestion about creating a new project. I find this idea appealing, which is why I made my previous post. However, I’m uncertain if your proposed approach is feasible at this stage, especially considering the unique characteristics of the financial markets. Nevertheless, despite this being a spur-of-the-moment post, I still believe that instrument selection, prediction, data assimilation, forecast correction, and execution form a compelling proposition. Nonetheless, as indicated, at this stage, a discussion might be more appropriate. What do you think, @g-gundam? Do you have any thoughts, if I may ask? In case you do, I suggest a direct message (DM) for the reasons expressed earlier by @FemtoTrader, or a brief discussion after the JuliaHPC Community Meetup (held every second week currently, assuming the hosts don’t mind). Additionally, I’d like to mention that it would be prudent to contact the authors of the mentioned packages directly to ensure they have no objections (I admit I haven’t conducted any legal due diligence on this matter yet). I’d like to add that in case anybody else might be interested, I would be happy to get in touch.
I understand your position regarding OnlineTechnicalIndicators.jl, @FemtoTrader. I still believe that adding a custom indicator would be the perfect finishing touch, like a cherry on top of a delicious cake, if you know what I mean. :- )