What language would you use to write a trading bot?

I missed the pump on BTC because I was too busy now it’s dumping and I want to make sure this doesn’t happen again.
If I had, let’s say a year to learn a language and complete the project, what language would you use and how would you begin going about it?
>pic is me thirsty for answers

Attached: 926AEE79-5E8D-407E-BCCE-8DF06D100A9B.png (338x277, 123K)

Lisp

i don't know anything about trading. tell us something about it.

when I wrote one I used Ruby because I already knew it and it had easy tools for dealing with web apis and json. the actual calculations were slow but it worked well enough. it became a pain in the ass when I wanted to make it threaded.

Honestly though, you missed the days to do this by a long shot. But the guys I know that are still into automated crypto trading are using tensorflow and one of erlang/python/crystal. Language doesn't ever matter much and if you don't know which to use the only real way to learn is to use the wrong one and find out what you need from a good language.

Bitcoin was never good for automated trading except for very very early on. You're too late to the game. It's always been about altcoins. Arbitrage is always the low-hanging fruit. I was making a few hundred $ a day by simply closing buy/sell spreads and price differences between exchanges, and I only put a few thousand into the bot.

sorry, when I wrote erlang, i meant elixir.
these languages are mostly chosen because they are already known, not because they are ideal. Better to write the software in a year than learn to write it in a year

Most trading bots are written in python, which happens to be the most noob friendly language

>Most trading bots are written in python
i've heard this one too

with python you can have a trading bot up and running after a month at the most.
you can also give it artificial intelligence later on but that might be more complicated.

Depends heavily on the market I'd assume. High frequency is mostly C++ from what I can tell (based on the people giving tech talks about C++ optimizations at conventions). Where microseconds matter.
For typical trades I can see scripting languages being dominant.

I'd use C#/.NET Core. Anyone who is recommending you trust your money on top of a C/C++ implementation, especially if you don't already have a decade+ of experience in this language, is a completely retarded nigger. Do you like or hate money?
>Like money: Use C#
>Hate money: Use C/C++
One little mistake in C/C++ with an off-by-one or just some memory safety thing you fucked up could cost you all your money. C# takes a lot of those unknowns out of an implementation. Unless your code is executing on-premesis of the clearing exchange, there is literally no reason to worry about sub-millisecond performance. Waiting for a Gen2 garbage collection is not a big deal if you are a NEET trading BTC from your parent's house.

Python is a bad choice too because of lack of strict type safety. Making API calls where your contract is just a big soup of dynamic being passed around is very prone to fuckups. No serious financial transaction system has EVER been implemented in python. Sure, I bet a ton of redditors made python crypto bots but I also bet none of them are wealthy right now.

Common Lisp

if I were to set up a bot id make it do trades in the span of day/s not milliseconds.
otherwise id get raped by fees.
So yeah in that case python>c++,c#,java,whatever else

Lots of
Infuriating
Small
Parentheses

Second. For high frequency stuff, c/c++ is typically used to communicate with an fpga which is where the real speed comes from. In those systems, trades would be made before a full packet reaches the custom Ethernet controller

ADA Engine with a Ruby API to link up to your wallet or other trading sites.

In that case, I'd recommend going with the language that most easily allows you to write correct code, including writing the code that handles trading logic in a clearly readable manner.

Python/R for research and mid and longish term stuff and C++ if you want to market make and/or do HFT.

HFT is very competitive though. Not in crypto, but you hear stories of HFT guys on wall street using FPGA to parse signals because having packets go through the operating system's TCP/IP stack is just too slow. My suggestion is stick to medium-term stuff and really git gud at math and stats, so you can find an edge

You people saying C/C++/C# and mentioning HFT probably have never looked into this seriously. While you aren't necessarily technically wrong, the average person does not have the means to perform HFT even if they already had the software. In cryptoland especially, the high-frequency traders either have an in with the exchanges, or they have enough money to take advantage of things like volume-discounts on trading fees. Even still, any HFT is going to compete with many more and will have to play a game of being the fastest and smartest, which OP would undoubtedly lose. The proper solution for OP will not be HFT, but will be anything that allows him to make intelligent trades without personally micromanaging his portfolio. This is a situation where its unlikely OP has enough time and money to do any sort of high speed trading, otherwise he wouldn't be asking for advice here.

Surprised that only one post vaguely mentions machine learning and that so many seem hung up on high frequency trading.

Ya pretty sure even a very high skill HFT developer will always lose to someone who has paid money (or is paid in rebates like you say) to be colocated with the exchange.

>HFT is very competitive though.
Competitive is not the right term, you should use "impossible" instead. These companies literally have a wired connection direct from the exchange to their data centers close by. There is no way on god's green earth an average joe can compete in the HFT space.

java/scala

Is Python best for machine learning?

yes

most common

don't lewd the kotori

Attached: received_945598879114265.png (1600x900, 960K)

So anons, where should I start if I wanted to learn HFT?

Unironically get a job at a HFT firm.

bretty good.
R is also a good choice but not worth the time to learn all its quirks just to do ml

erlang. this is actually what hyped erlang

based
>inb4 pajeet jokes