[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/biz/ - Business & Finance

View post   
File: 102 KB, 657x539, 1514431188367.png [View same] [iqdb] [saucenao] [google]
5473724 No.5473724 [Reply] [Original]

Has anyone here made their own trading bot? How's it worked out for you? How complex is it (pattern matching or just simply buying low selling high)?

>> No.5474127

>>5473724

Bump senpai

>> No.5474209

You didn't buy BZC yet anon? Kek

>> No.5474311

>>5473724

I'm learning python to make a bot right now

>> No.5474427

>>5474311
>he wasnt born speaking python

>brainlets

>> No.5474484

I have and have been running algotrading models for around 5 years now, crypto for the past 6 months. I focus on trend and momentum, generally exploiting known market inefficiencies, however I have also experimented with all kinds of machine learning models and non-linear setups. Is there anything specific you'd like to know?

>> No.5474726

>>5474484
>>5473724
What programming languages are best for a trading bot? Books for learning about this topic? And how good are these things?

>> No.5474878

>>5474484
yes how much have made roughly?

>> No.5475121

>>5474726
> What programming languages are best for a trading bot?
Unless you want to do HFT or in general something where execution speed is the alpha generating factor, programming language choice is relatively unimportant. I have personally written systems in Python, R, various ml implementations (ocaml, F#). C++. As long as the language is stable, has good support/libraries for stats and math youll be good. I'd personally recommend pythong to most beginners.

> Books for learning about this topic?
Depends on your background really. Goes without saying that a good grasp of entry level math (at least) is a pre-requirement. Dismissing that
"Trading Systems: A new approach to system development and portfolio optimisation" is pretty good. Anything by Ernie Chan is also great for beginners.

> And how good are these things
Well theres no magic there, these things are as good as the underlying model. If the model is sound, you just have to worry about automatic execution and thats all there is to it.

>> No.5475274

>>5474484
I already know programming, I'm curious where you get your algorithms for trading from. Whats a good place for beginners?

>> No.5475298

>>5473724
auton.io

ai trading bots that you can program yourself or buy algos from tohers with the coin that burns itself all way till theres 20% left. you can also backtest. first 1k users are paid for.

>> No.5475307

>>5474484
hey anon, do you have a basic algorithm that would be best for implementing first?

>> No.5475344

>>5475274
>I'm curious where you get your algorithms for trading from

That's why you study computer science. You learn that + runtime estimation for algorithm there. Otherwise get a book about datastructures + algorithm & learn a bit analysis.

>> No.5475368

>>5475307
check it out. make and backtest your own

https://auton.io/

>> No.5475904

>>5475274
Thats a hard one to give a non contrived answer to anon. Its pretty obvious that nobody will share a winning model, depending on the market the advantage will be arbitraged away almost instantly. Anything you read in a book should only serve as inspiration to a real production models. That being said, you have a number of options.

If you are starting from scratch, I think it would be a good idea to pick up a solid quant book like "Quantitative equity portfolio management" by Qian and familiarize yourself with the general terminology and concepts from a quant perspective, specifically risk management, market inefficiencies, portfolio construction, anomalies and why the EMH and random walks and such are not universal assumptions, etc. Also, anything from behavioral finance, especially books and papers from the late 90s are very good.

After you've established some familiarity with above concepts, get your hands dirty right away. Don't wait for some kind of absolute clarity to hit you one sunday morning. Its important you start building dumb models and build up an intuition about these things. You can start by taking a known anomaly, for example "momentum" and build models to try and exploit that. There's cross-sectional momentum, there's autocorrelative (time-series) momentum. Try your hand at those.

Gonna mention him again, but Ernie Chan is very beginner friendly and he generally focuses on the modelling rather than the programming. Have a look at his blog, check out his books.

A lot of interesting ideas and concepts can be found in academic papers. https://arxiv.org or any other similar site is great for this. Obviously, with academia, all should be taken with a grain of salt, as their setting is overly romantic.

In general, start following quant resources, reading and building models. The rest will follow naturally.

>>5475307
see above

>> No.5476183

>>5475904
Yeah yeah yeah you fucking virgin the question is does your model actually work?

>> No.5476287
File: 56 KB, 645x729, d27.png [View same] [iqdb] [saucenao] [google]
5476287

>>5476183
you think I've been doing this shit for free for the past 5 years or what?

>> No.5476371
File: 621 KB, 649x706, 1481306963875.png [View same] [iqdb] [saucenao] [google]
5476371

>>5473724