[ 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: 116 KB, 675x680, 1612891399717.jpg [View same] [iqdb] [saucenao] [google]
28702232 No.28702232 [Reply] [Original]

Do any of you algo trade crypto?

>> No.28702398

I just use an indicator that cost $200. It seems to be accurate like 80% of the time on the 4h time frame. And even if it's wrong you usually sell for breakeven.

>> No.28702594

>>28702398
what's the indicator?

>> No.28702743

>>28702398
I mean like fully automated

>> No.28702779

>>28702232
no. i was going to try making a bot using sentiment analysis from biz and some historical data. but historical data is so rediculously exensive. its 70 dollars a month for the CMC api which basically only gives you OHLC data going back 1 month.

>> No.28702848

>>28702779
*the 70 dollar price point only gives you

>> No.28702933

>>28702232
Cryptos too volatile, doesn’t work

>> No.28703468

>>28702232
No, I just btc and avoid giving my money to exchanges

>> No.28703575

>>28702232
Imagine bringing this to class lol

>> No.28703937

>>28703575
Lmfao

>> No.28704316

>>28702232
source on that BASEDtop?

>> No.28704834

>>28702933
Is there anything that's worth algo trading other than stonks?

>> No.28704981

I wrote a python script that combined the most popular indicators (MACD, EMA, RSI) and gave me buy and sell signals. It worked relatively well (compared against buy and hold, in a week, it did 5-6% more). Currently trying to implement a optimizer to see which indicator parameters maximizes my returns, however I can't seem to make it work.

>> No.28705101
File: 50 KB, 803x759, coolkeyboard.jpg [View same] [iqdb] [saucenao] [google]
28705101

>>28703575
i bring something similar to my classes...with this keyboard

>> No.28705184

>>28704316
>expanscape

>> No.28705442

>>28702398
What's the indicator?

>> No.28705573

>>28704316
It's a prototype, I forget by who. Do the reverse image search.
>>28704981
What platform do you use? I got a free account on quant connect. But I guess it's different with crypto. Does proximity to the exchange even factor?

>> No.28705966

>>28702232

Currently trying with a similar idea to >>28702779

The lack of historical data means you need to have giga balls. There is so much potential, but I don't think it's easy to get something that appropriately captures the bull / bear markets and btc / alt season though.

One idea I was playing around with is copying the trades of known bot wallets on an hourly timeframe, since data and api keys are so expensive.

>> No.28706278
File: 160 KB, 942x708, bot.jpg [View same] [iqdb] [saucenao] [google]
28706278

>>28705573
It's all hard coded. I had some free time during the quarantmeme. Read some python books and started writing the framework.
It uses the Binance API and the usual python libraries (Pandas, Scipy.optimize, mplfinance (for graphing), ABC).
My objective was to write a program that did 1 or 2 trades per day with a reasonable take profit strategy.

Don't have my compute with me at the moment, but the picture is when I had just begun (it clearly was shit back then).

>> No.28706551

>>28702933
imagine being this retarded

>> No.28706667

>>28702232
Why the fuck would you trade that shitcoin? go trade an actual crypto like ether or something pussy.

>> No.28706686

>>28704834
power

>> No.28706872

>>28706667
wat

>> No.28707152

>>28702232
this is something i would like to start trying out when i have some free time.
with stocks at least you can follow forums, DD and particular news sources as stocks are rarely on big news sites or blowing up on twatter. but cryptos for some reason (boomer media, elon, etc) can get so hyped and pumped from just mob mentality.

>> No.28707173

>>28706278

Out of interest, is your strategy for taking profits hard coded, or based on indicators?

>> No.28707317
File: 9 KB, 56x56, pet.gif [View same] [iqdb] [saucenao] [google]
28707317

>>28706667
take you meds bro

>> No.28707555

>>28707317
>take you meds bro
bro, you can't even type a coherent sentence, no wonder you'll believe anything the jews are shilling at MIT

>> No.28707572

Im working on one right now using coinbase API
>pick three coins that aren't ETH
>math
>arbitrage
I'm not planning to use it until crab or bear market.

>> No.28707698

>>28707572
The fuck is a crab market

>> No.28707801
File: 1 KB, 93x125, 1613258710637s.jpg [View same] [iqdb] [saucenao] [google]
28707801

Isnt it just easier to find a whale wallet and stalk that?
I know theres PnD groups out there.

>> No.28707855

>>28702232
I trade Algo the crypto, like a chad.

>> No.28707907

What laptop is that + accessories? I’d love that for a coding setup.

>> No.28707942

>>28707572
crab market isn't real bro

>> No.28707987

>>28706278
is something like this feasible for someone who hasn't had experience writing a large project before?

>> No.28708099

>>28705101
Kinda limits the words you can spell

>> No.28708107

>>28707555
This isn't an Algo thread you schizo

>> No.28708275

>>28702232
I hire a pajeet for 20 rupees a day to make an excel spreadsheet of all the coins shilled here in a 24 hour cycle. I filter based on count of which coin were shilled the most then I just buy and hold them for 3 years. It works pretty well.

>> No.28708281

>>28704981
Cool, how do I start? Can you show me your script and work flow?

Thanks anons.

>> No.28708423

>>28706551
Seriously lmao, like the whole point of algo trading is taking advantage of volatility when normal investors are running away.

>> No.28708533

I don't get the "data is expensive" comments, isn't Binance API free to use?

>> No.28708854

>>28708281
The concepts are the same no matter what market you're automating for. Hit up quant connect or quantopia and make an account. They have a bunch of tutorials and good forum community.

>> No.28709106

>>28708533

In my case I wanna use sentiment analysis of biz, and haven't been recording shitcoin mentions for a long enough period to properly back-test.

>> No.28709544

>>28705101
>>28708099
I can still write, "Y0V FAGG0T."

>> No.28709628

Botting how you make real money in crypto. Suckers and newbs try to day trade. Do you know how much snipe, frontrun, and arbitrage bots make?

>> No.28710006
File: 7 KB, 227x222, ngaplz.jpg [View same] [iqdb] [saucenao] [google]
28710006

I don't think you can actually do anything that great with sentiment analysis, if you're using it in buy/sell it should be lightly weighted amongst other harder insights.

>> No.28710078

>>28709106
This site only allows you to go up to 7 days back, but it's free:
https://www.merv.tech/mentions/biz/all

>> No.28710249

>>28702779
Anyone want to bounce some ideas around? I have writen a backtesting fw and a bot that i can plug any strategy into, but i can not find any good strategies as of now. Ahoot me an email if you want to colab.
Algo_trading_biz at protonmail.com

>> No.28710463

>>28702232
i do fully automated algo off and on. use NN so i have to go offline when there is a paradigm shift. it can be reliably profitable in sideways markets. however it's hard to beat buy low and hodl.

>> No.28710805

>>28706667
kek

>> No.28711903

>>28702232
delet this

>> No.28712352

bump

>> No.28713418

Bump

>> No.28713628

I can't justify actively trading when it is trivially easy to just buy and hold. Generating a million taxable events and owing short term capital gains while assuming all risk is annoying. Makes me feel like I am waging my time away to feed the ZOG 30% of my gains..

>> No.28713856
File: 9 KB, 292x173, back.jpg [View same] [iqdb] [saucenao] [google]
28713856

>>28713628
you have to go back

>> No.28713939

>>28702232
Best way to make money with crypto is holding AT LEAST 10 days. So no, I don’t trade

>> No.28714289
File: 312 KB, 500x282, cringe.gif [View same] [iqdb] [saucenao] [google]
28714289

>>28713939
Algo trading is not day trading, very different. Also holding is not the best way to make money with crypto.

>> No.28714405

>>28705101
Why does this make me want to use your boyhole, anon?

>> No.28714772

>>28713856
niggers don't love you back, pal
80% of bidens cabinent are jews, and the heads of major departments are all jews

>> No.28714952
File: 83 KB, 706x1024, 1522685083147.jpg [View same] [iqdb] [saucenao] [google]
28714952

Can anyone help me get some data on BTC drops/gains?

I'm looking to compile a normal curve for BTC drops/gains by the second. I want to see what the average drop/gain is in terms of percentage as well as the standard deviations.

Is there somewhere I can download the trading data for BTC?

In return. Here is a female

>> No.28715775

>>28710249
perhaps -- sounds like you might have more sophistication than I do though. :)

I was just going to make a (dumb) project for myself that I can run that will run a couple different types of continuous jobs:

1. low-frequency "scraping" jobs (data from APIs preferably, possibly actual scraping if no API is available - the point of this is to attempt to identify candidates for purchase now or some time in the future.

2. high-frequency price monitoring jobs that check the prices of a list of currencies at an interval, and alert if the price is within your buy or sell target.

this will of course be pretty manual and dumb (heuristic / rule-based) to start with but it could become smarter over time.

>> No.28716017

>>28708533
Binance is free, yes.
Another option is writing a scraper for Coingecko like I did.

The subdomains on Coingecko are always like /{language}/coins/{coin} and writing a Selenium or Beautifulsoup script for it is really simple.

>> No.28716569

>>28716017
Coingecko (at least now) exposes the v3 endpoint, publicly, for free. there are also a lot of unofficial sdks available in your favorite runtime / programming language as well.

see: https://www.coingecko.com/en/api

from my personal spot testing, it generally seems OK, but I'd want to heavily test it myself before relying on it.

>> No.28716574
File: 293 KB, 1080x927, 1519946563188.jpg [View same] [iqdb] [saucenao] [google]
28716574

>>28714952
im sure someone else already compiled this info and i dont want to go through the hassle if its just available somewhere

here is another female

>> No.28716581
File: 33 KB, 1021x701, its_literally_this_easy.png [View same] [iqdb] [saucenao] [google]
28716581

>>28714952
https://mega.nz/file/n4hUBaTa#qhWGto5zFL2F90Fo1TQ2FBumTVR_8fkc6Kfn3GGyIMQ

only goes back to 2014

>> No.28716653

>>28716581
thank you anon

>> No.28716699

>>28702933
Lengthen the time frame and smooth it and it works okay. I still don’t trust it

>> No.28716720

>>28714952
sorry re-read that yeah it's not that easy to get that info, it would be surprising if someone was just giving it away.

>> No.28716989

>>28710249
>>28710078
>>28709106
I have /biz/ data back to 2017. It only tracks mentions of coins on Bittrex but I could redo it to track whatever you want desu

>> No.28717136

>>28716989
If someone starts a trading group I'll post my shit in exchange for ideas, maybe a hedge fund like setup. I don't really want to just post it for the public.

>> No.28717154

is there any place you can get past market data for free somewhere? i just want to play around a little with algos and make some test runs

>> No.28717195
File: 65 KB, 1404x898, ark_biz_sentiment.png [View same] [iqdb] [saucenao] [google]
28717195

>> No.28717318

>>28716989
I've been looking for exactly that data, can you please email me propagandabuster@protonmail.com

>> No.28717481

A gift from California.
Tells you what coins are mentioned in Reddit, Twitter, Biz, Telegram..
https://www.merv.tech/mentions/twitter/all

>> No.28717828

I am tempted to write a python bot to trade for me, I learned a short term trading strategy from a forex trader I know IRL that makes an absolute killing in crypto, but I like little projects and I would love to condense the strategy down to an algorithm and have a bot execute it for me 24 hours a day, it could literally make 7 figures a year if it just ran for 6 hours a day. However some of it involves manually reading charts before committing to a trade, so I'm not sure if it's gonna be doable.
If I could just write something that would pop up and show me the chart when certain conditions are met (it's largely based on indicators, although some of these are closed source indicators) and then ask for a proceed/abort decision and then set my buy orders for me that'd be good as well

>> No.28717860

>>28704981
if you are guaranteed 5-6% profit why dont you then use huge leverage and get all the money in existence on the stock market? within a few months?

protip: because your muh deep learning algo is just random chance like every other algo ever

>> No.28718449

>>28713856
holy shit r*ddit will you fuck off

>> No.28719003
File: 135 KB, 1122x552, Screenshot from 2021-02-14 20-44-10.png [View same] [iqdb] [saucenao] [google]
28719003

>>28717828
That's pretty much what i had in mind as well.

>> No.28719518

>>28702398
VPVR?

>> No.28719876

>>28702232
Yes I run the algo in my brain rig, RTX 140 IQ.
if(year<=2019 && isLinkOut==true){
accumulate();
} else {
hold();
}

>> No.28719989

>>28702933
>Cryptos too volatile
Anon you profit via volatility.

>> No.28719993

>>28702232

I am currently trying to use Hummingbot and figure out if I could get it to read off some strategies from Tradingview.

There are a few indicators in tradingview that have been fairly accurate if I can get them to basically IF x and X then do X. I just am not a code expert or anything so I am sort of limited. Do you use any type of cyphers for your trading indicators? Like just now I had an indicator tell me about the GRT 1.9 to 2.00 dip and coulda cleaned up about .10 to .20 per share over a few hours.

Keep missing them from having to catch indicators though or buying a hair late and hurting my swing chances.

>> No.28720264

Just started getting into algo trading. After backtesting what would you consider a good algorithm? I've made a simple range trading algorithm which seems to outperform the market by 3.5-fold. Is backtesting overly optimistic? How do I make sure I don't over-optimize my parameters?

>> No.28720482

>>28706667
he doesn't mean ALGO the ticker
he means an algorithm
you dumbass

>> No.28720801

>>28720482
Kek

>> No.28720841

>>28716581
Does it only provide daily data? Given how volatile cryptos are I don't think that's sufficient time resolution.

>> No.28721038

>>28720264
have you tried live testing yet? that's the true mark really, just run the algo with fake cash.

>> No.28721145

>>28704316
Google Aurora 7

>> No.28721188

>>28720841
yeah yfinance isn't the best, you can only get finer grain data for up to 60 days and if it's 1 minute or less only 7 days.

>> No.28721194
File: 349 KB, 1080x925, etiquette.jpg [View same] [iqdb] [saucenao] [google]
28721194

>>28713856

>> No.28721324
File: 516 KB, 565x486, brap.png [View same] [iqdb] [saucenao] [google]
28721324

>>28721194
thanks for shitting up my thread bro

>> No.28721368
File: 354 KB, 1080x2340, Screenshot_20210213_001915_com.brave.browser.jpg [View same] [iqdb] [saucenao] [google]
28721368

ProfitTrailer results since Oct 2020

>> No.28721754
File: 57 KB, 640x480, XMRUSDT.png [View same] [iqdb] [saucenao] [google]
28721754

>>28721038
Not yet. I assume that even if it does well now in a bull market that isn't very representative of what I should truly expect. I'm using Binance data and that only goes a couple years back, so I can't really capture any big price dumps like 2017.

>> No.28722137

>>28721754
Huh yeah that makes sense. It's pretty unsurprising that there isn't a free source on fine grain data back to the beginning. It would cost a lot to accumulate and manage it. Why would anyone give it away or sell it at a loss?

>> No.28722300

>>28721324
There's something I don't stand for. And that thing is heckin disrespecting Israel.

Also good fucking luck trying to make a bot. I've got intermediate programming experience, and I've been in crypto for years. I never bothered trying to make a bot because every script kiddie and their mates have given it 100 tries. It's just an arms race. I don't think I've ever seen anyone with a bot that is more than 50% accurate either (which is the goal).

>> No.28722354
File: 56 KB, 474x625, 1612528047429.jpg [View same] [iqdb] [saucenao] [google]
28722354

>>28705101
I can't count past 5 either.

>> No.28722445

>>28722137
The public Bittrex API has minute candles going back to when they listed the currency.

>> No.28722657

>>28722300
oh jeez i guess it can't be done, guess i'll just give up.

>> No.28722907

>>28722445
Better than Binance it seems. Thanks anon.

>> No.28722980

>>28707698
>being this new
Strong sell indicator

>> No.28723795

>>28722445
Based

>> No.28723890

>>28702232
That looks pretty cyberpunk, bro.