prev >>61115210>quant newshttps://www.ft.com/quant-investing>quantopian lectureshttps://www.youtube.com/playlist?list=PLRFLF1OxMm_UL7WUWM31iynp0jMVf_vLW>event driven backtester designhttps://www.youtube.com/watch?v=OhGdVdolqkU [Embed] [Embed] [Embed]>bookshttps://github.com/zslucky/algorithmic_trading_book>recommended brokersstocks and options - https://alpaca.marketsfutes - https://www.tradovate.comcrypto - https://www.kraken.comforex - idek>wiphttps://alphapulse.org
Nothing much interesting for me today again. Still just working on the defi arb rewrite, might have end to end data flow working by end of day, but not execution.
>>61131030Still got one question for you remaining at the end of the last thread if you want to go take a quick gander
>>61131515If you're the book guy, just throw the book in the trash lol. Who's the author? That's a completely ridiculous and factually incorrect statement to make, the market runs on algos - of course they turn a profit.If you're the local LLM guy, I've been anxiously waiting for them to get good enough to run on consumer hardware. Thought about buying one of these at one point:https://tinygrad.org/#tinygradBeing able to run something as good as Claude Code locally would be so cool.
>>61131515Oh, if you're the discord guy, yeah sure? I'll start including the server link in the OP. Ideally you can filter users by geo location too or something.
>>61131515/biz/raeli mafia server
>>61131550Uhhh.. well I wrote a reply, but it didnt actually get posted? No I am the rasp pi, C#, autoregression guy that also mentioned the local LLM thing and the new Nvidia spark box that costs $4k>Thought about buying one of these at one pointYeah same specs as this one, but Spark has better hardware afaik, and also runs cooler, and also costs way less. So definitly pretty cool to see>>61131674>Oh, if you're the discord guy, yeah sure? Yupp thats also me, cant believe the reply I wrote didnt get posted by the time you realizedheres the permanent code to the server: apXbwjHxK4, cant just post a link because it gets caught as spam
>>61131709damn that's nuts if the spark is that good for that price. wasn't sure if it was just NVDA hype train shilling or notOK I just popped in the discord, I don't get on there much though desu.
>>61131816Yeah same, I am also using a throwaway account, so I will be on even less, but its nice being able to just collect everything in one place without having to take snapshots of archived threads
suggestions for integrating a screener into alpaca?
>>61131829>rasp pi, C#, autoregression what's the autogression stuff you're doing?
>>61131912Just running a basic momentum automated strategy right now, but I am almost finished making an automated setup where a bunch of strategies like MA, AR, ARMA, ARIMA are all simulated and then it attempts to run the best performing symbol to strategy pairs. Shouldnt take long to finish a prototype once I have some time on my hands
>>61131959nice I think I recall you saying performance was actually decent given how simple it was or something and wanting to upgrade. what assets/instruments, crypto? are you DIY'ing everything in C# or using a platform/framework/library?
>>61131979>performance was actually decent given how simple it was or somethingWell more like I was expecting it to crash and make nothing at all, but it is actually in the green, although its under-performing the stock itself. My main goal is to see if it performs in-line with the backtest simulation, which for now seems to be the case>what assets/instruments, crypto? Just wrote a quick JSON list of S&P top 100 or something, gonna keep it simple like that for now>are you DIY'ing everything in C# or using a platform/framework/library?Its DIY C# stuff yeah, it uses ImGui for the analytics stuff I have been making, and for the terminal program I use a bunch of beautification libraries. Unfortunately unlike python which has a lot of pre-existing libraries and open source projects, C# is lacking. Although python is genuinely garbage performance vice, so it is what it isGotta say though I do all this for fun, I dont expect to make any money, and the money I have made was from something far removed from algorithms and even indicators
>>61132062>Well more like I was expecting it to crash and make nothing at all, but it is actually in the green, although its under-performing the stock itself.This has kinda been my experience with this stuff so far as a whole. I didn't think it'd be possible to compete with the "army of Wall Street quant PhDs", so I never bothered until recently. But the more I've learned, the more I've come to realize that mindset isn't reflective of reality, and there will be inefficiencies and such too small for them to pursue that retail can game (eg low capacity strategies or areas where they aren't allowed to participate eg defi). >Just wrote a quick JSON list of S&P top 100 or something, gonna keep it simple like that for nownice>python is slowYeah, I'm trying to build my system using Rust for performance where it matters, then a Python API on top of it to leverage all the ML stuff optionally. Long ways out from actually getting there though. Maybe you can do something similar with C# if it's ever suitable
>>61132225with the new .NET stuff C# has the leverage to be on par with Rust, or C or C++ without having to deal with all the bullshit you get from them. But I am not the person to do it. To me its just an excuse to write poorly composed code and get away with it still being 10000x more performant than anything well written in python. I am doing a quick write-up on what id like to do with ML on the forum at the server so feel free to comment about my schizophrenic ideas there
>>61132280>with the new .NET stuff C# has the leverage to be on par with Rust, or C or C++ without having to deal with all the bullshit you get from them.what "bullshit"? This has been my first project using Rust (normally C++, but I'm not a pro dev or anything, but familiar enough C++) and I gotta say I'm hooked. Cargo is great, I like structs and composable traits over classes (composition over inheritance etc). And obviously avoiding memory issues is nice. One of the reasons I decided to go for it was that it's sort of a compiler driven development process, and with AI assisted development it gives the LLM rich debug output to code for me, and I have some level of assurance that if it compiles it will run / work.C# always struck me as a Microsoft pajeet lang like Java, bt idk I've never used it or looked at it>But I am not the person to do it. To me its just an excuse to write poorly composed code and get away with it still being 10000x more performant than anything well written in pythonfair enough lol>I am doing a quick write-up on what id like to do with ML on the forum at the server so feel free to comment about my schizophrenic ideas therenice I'll check it out and maybe do the same, put some static content there or whatever for reference
>>61132349>what "bullshit"?issue with Rust is that its limiting and forces you to use unsafe code, which defeats the purpose of using very safe codeWhen it comes to C++ and C I just find it to be a pain, but hey I am a scrub>C# always struck me as a Microsoft pajeet langIt is definitely THE Microsoft pajeet lang. However the paradigm is no longer really object oriented slow programming like Java, and it also beats Java out of the parkThat being said, there are some other reasons I am using C# I cant get into, so just take my word for it, for my situation its the best fit
>>61132405>issue with Rust is that its limiting and forces you to use unsafe codebut C and C++ have *no* safety guarantees - at least Rust offers this optional protection to begin with, right? I know it often ends up having unsafe code, but this way you can narrow down where it is at least rather than it being system wide.>so just take my word for it, for my situation its the best fityeah I believe you, was just curious
https://www.youtube.com/watch?v=s0AOTdzyiLg&list=RDEMH_SvLoTCTjHBc9IhSajIwg&index=3
this is a worthless update, but rust has a nice terminal UI library
i feel like im way too behind to try to get in on the machine learning trying at this point since i have no idea how ML works
>>61134924ML is a lot simpler than you'd think, it's basically just a self-weighting graph
>>61131909i am once again asking for any suggestions on integrating a scanner/alerts into alpaca :(should i just try the tradingview integration?
>>61135537somehow I missed your message. TV setup would probably be easiest, since they have the logic and UI already for the scanner/alters and web hooks for triggering orders with an API
>>61135537https://alpaca.markets/learn/low-code-algorithimic-trading-on-alpaca-using-tradingviewyou can look on GitHub too for a good start then probably just ask ChatGPT for help
>>61135731>>61135743ty anons this is very helpful
>>61135814yeah np I can help you with it intermittently if you have any issues
>>61135835ty man, i checked out that article and am realizing TV needs a payed sub to do this. if you have any ideas for a free way to get a scanner/alert API lemme know :DI might have to rethink my idea for something that doesnt rely on that anways, going to check out the OP links
>>61135856do you pay for data from Alpaca? or are you OK with a 15m latency?
>>61135856>>61135900or do you pay for any data really? you either need to pay for data or be OK with 15m latency, but yeah I can help you out if
>>61135900>>61135905just using a free paper account with alpaca that i set up last night, a delay will be fine. a quick google search( which i shouldve done before asking, my bad) shows some options that look promising. I basically just want to use the screener to narrow in on a few tickers, then ill focus in on one or more of those with the realtime data from alpaca to then execute a few simple trades. does that seem doable with the free options ?
>>61135962>realtime data from alpaca>does that seem doable with the free optionsmost free data is lagged 15m, I think.. IDK. ill look when the market is open tomororw
>>61131132>might have end to end data flow working by end of dayok, it is now past 1am and still nothing works
>>61136635>>61131132it is now 3am and still nothing works
>>61131550https://www.nvidia.com/en-us/products/workstations/dgx-spark/
>>61137300Yeah I recall some long all nighters after trying to get ... just .. one .. more.. feature. To work, and it always ends up taking me another day regardless
>>61137575I JUST finally got where I wanted to get today, been working on it since 10am. I estimated it would only take 3hrs kek see yalls tomorrow
>>61137663Cya tomorrow, or later tonight for me i suppose
