[ 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: 37 KB, 1012x685, 1614634293508.png [View same] [iqdb] [saucenao] [google]
30036243 No.30036243 [Reply] [Original]

good morning beautiful big brain aryan eurochads.

>brag about your bots, discuss strategy, etc.

last night an anon posted pic related and discussed his trading bot. i spent the evening looking at every single trading bot UI out there but could not find one that matched. does anyone recognise this software?

i have recently begun working on my bot which will use crossing EMAs as buy triggers with other indicators to help strength of signal. if anyone has any experience with this strategy, and can recommend candle periods to work on, which indicators to use etc, do share.

>> No.30036568

Python. Sell when coin hits high for the day. Buy back in when 10% lower than that.
Rinse and repeat.


(I'm a shit programmer)

>> No.30036812

>>30036243
Crossing EMAs will never work as a strat, way too oversaturated. It's used in every babys first trading bot tutorial unfortunately. Good luck finding a profitable strat, it's not that easy

But I can recommend some python libraries for you to use. Ccxt is useful for exchange API wrappers. Freqtrade is good for writing strategies and shit, and backtrader is a library for testing your strategy on downloaded data, which can give you some info about how your strat performed.

Join here on telegram if you want to chat with other biz algo fags
t(dot)me/biz_crypto_bots

>> No.30036824
File: 27 KB, 450x429, toydarians-watto02.jpg [View same] [iqdb] [saucenao] [google]
30036824

>Exchange changes API
>Your orders are now fucked up
>Go broke in approximately 5 seconds

No thank you.

>> No.30036877

>>30036243

I haven't programmed a trading bot per se but I have a simulation I can run with historical data, But I just can't get it to make profits, no matter what ta indicators I plug in. The exchange cost of 0.001- 0.003% really fucks everything up when you want a bot that does hundreds of trades per day.

Anyone have any ideas? Thinking about moving away from ta and using Twitter trends instead.

>> No.30036907

>>30036877
go back

>> No.30036950

>>30036824
>Exchange changes API
Do you think that happens without notice?

>> No.30037069
File: 236 KB, 960x1440, 1473010029726.jpg [View same] [iqdb] [saucenao] [google]
30037069

>>30036950
>Devs read documentation

>> No.30037269

>>30036907

So ta is the way to go? I spent endless nights on it just combining tas weighting them differently, getting correlations between them, using different window sizes. I'm at the point where I'm thinking of automating this aspect with self reinforcement learning. But do far the number always goes down compared to hodl strategy. I will jizz my pants instantly if I get the numbers to go up exponentially.

>> No.30037278

>>30036812
very much appreciated anon. i will definitely check out and use those libraries. i hear your words of wisdom RE: EMAs and i'm sure you're right, but i have to try it anyway of course and see it with my own eyes.
>>30036877
last night's big brain told me that the way to go is limit orders when selling to reduce fees. even with low profit margins per trade, it seems possible.

>> No.30037326
File: 37 KB, 500x467, 1270931589930.jpg [View same] [iqdb] [saucenao] [google]
30037326

>>30036243
>does anyone recognise this software?

Not 100% sure but those charts look like they're from 'ApexCharts.js' which is just a .js charting library. The anon you're talking about probably used that to display the results of his bot.

I'm also working on a trading bot. Specifically one for binance, though I'm having a bit of trouble deciding which strategy I should use. Is trend analysis by itself enough? Currently reading "Technical Analysis Explained" by Martin J. Pring to get some more insight but I feel like some of it is going over my head.

>> No.30037489

>>30037278

I think on binance the fee is still 0.1% for spot.

>> No.30037494

>>30036243
>select 10-20 wallets on bsc that are swing profitably and often
>make a bit that copies their trades
>profit

>> No.30037532

>>30036243
>aryan
Yes how did you guess we were Indians?

>> No.30037570

>>30037326
you might be right anon, they look similar. even if its not apexcharts it could be another library unrelated to trading bots. the anon said that he did not program the bot himself though, only configured it. but it's possible he was given it by a friend.

>> No.30037987

I sit in front of screen all day so i dont need bots. I just look at MACD, RSI and DMI. Easy as that. Fuck the EMA crossover, by the time it crossed over, knowing this market it will do a reversal. EMA is only good for macro trends

>> No.30037999

>>30036243
imagine a world without trading bots

>> No.30038111

>>30037999
Necessary evil, bid/ask spread would be too large without them.

>> No.30038509

eurochad here
just got done with my coffee after filming another porno for your mom.
now that I'm ready to get the day started, I looked at my imaginary ALEPH bot and thought "i wish i had a yacht"

>> No.30038570
File: 204 KB, 686x522, EvMCalAWgAILzGu.png [View same] [iqdb] [saucenao] [google]
30038570

>>30038509
kek, good morning chad.

>> No.30040066

>>30037269
>the numbers always go down
just do the opposite of what you're doing? buy when your indicators say sell and sell when they say buy?

>> No.30040245

>>30037987
>I just look at MACD, RSI and DMI

Are those the only indicators you use to make a purchasing decision anon? I would like to know what else I can include in my strat

>> No.30040310

>>30040066

numbers go down because of exchange fees. Otherwise the indicators work to a decent extent. If I do the opposite I go broke pretty quickly to an extant. Problem is that I'm not making more then the 0.x% profit per trade. I could zoom out and stop working with minute candles and only do like 1 trade every few days. But this is not really my idea of an ideal bot. An ideal bot should try to exploit micro fluctuations in the market.

>> No.30040456

>>30037987
>MACD, RSI and DMI
I use MACD, StochRSI, TRIX and BBpercentage.
I've not used DMI much, but I've just added it to my view

I have a Gunbot licence (from years ago) but not used it since then. Might fire it up, but am also a PHP dev so thinking of building something with AJAX etc

>> No.30040689

My bot looks at the order book changes to determine what and when to trade.
Order book one can see whales flashing their dongs and normal people placing orders. My bot can detect other bots too, by looking at past trades and order book.

There is no need to look at charts of any kinds.

All you need is look at the order book, changes to order book over second intervals without regard for price or where price has moved. Look at past trades actually executed. Now you can see how much fakeness there is in the books beyond the first 3% depth.

Now you can profit.
My bot is faster than your, 10ms, so dont bother.

>> No.30040771

>>30040689
>order book
what is a stop-limit

>> No.30040801

>>30036568
Does it make a profit?

>> No.30040830

>>30040689
high iq
>>30040771
very low iq, use your protein computer faggot

>> No.30041321

>>30040689
well done anon. i'm afraid i'm too uneducated at this point to understand how you go from
>Now you can see how much fakeness there is in the books beyond the first 3% depth.
to
>Now you can profit.

are you saying that your bot can detect large orders that will affect market price, so it can ride off them to make quick buys and sells?

>> No.30041537

>>30036568
Python is for pajeets, use Lua.

>> No.30041615

>>30040830
well clearly actual whales don't show their hand until their stop limit is triggered, so how he can possibly see something only an exchange can see and not publish is clearly LARPing faggot

>> No.30041819

>>30041615
would you want to have a obfuscated view or a obfuscated^2 view on the market?

>> No.30041862

>>30041321
When humans place orders and they go unfilled after 3-5s (the bigger size the more they wait) the human always cancels the order then places another one just higher/lower to get it filled.
These are the orders you want to look for, disregard all the bot orders and all the fills.
Another filter against bot orders is remove the non-whole number orders. Bots place buy at 0.15354 humans almost always work with nice whole and usually even numbers.

When you see humans placing orders and waiting less time for them to get filled, price will move in that direction it is 110% certain, then you see market fills happening in that direction.

>> No.30041928 [DELETED] 

Oh, [s that real|are you serious|are you joking}? Why do you buy this shit?
Want to save your portfolio just but your brains back to your head and listen to me lil boy
>YVS Finance yield farming - that’s what youneed since you know about crypto!
So be smart and get cash here, no deposit or extra fees, just a cool staking model!

>> No.30042444

>>30041862
wow, very interesting anon thank you for sharing.

>> No.30042545

>>30041862
What are you using to build your bot anon? 10ms is pretty nice, I'm using C# and it feels a bit sluggish

>> No.30043151 [DELETED] 
File: 88 KB, 1305x827, guess.png [View same] [iqdb] [saucenao] [google]
30043151

tell me biz, where do you it went?
would you enter a long or short position here?

>> No.30043192

>>30043151
neither

>> No.30043641

>>30040801
probably not. google bitcoin backtesting. there are a few tutorials that show how to do backtesting. there's at least one exchange that offers daily historical bitcoin price data, for free.

>> No.30043825

>>30037987
how is this profitable? i'm just learning this shit. i look at RSI, MACD and it seems like a losing proposition. obviously every trade won't be profitable but it seems like more lose than win.

>> No.30044579
File: 852 KB, 2048x1475, IMG_20210217_231430_884.jpg [View same] [iqdb] [saucenao] [google]
30044579

I was about to pay 200usd an hour for a momentum bot to be created for me, based off of the warrior trading guys strategy on YouTube. Seems one could program a long on strong upwards momentum and pull out once momentum is lost with a market SL very close to buy so can escape sudden sell offs.

Any one with exp on this sort of thing?

>> No.30044871

>>30038570
>Where americans use phones
>Where americans use automatic wheelchair devices

Its 2021, i don't understand what your graph is trying to get at. Well done, fast food was invented.

I'm a brit and i'm STILL even defending the yankees over this one

>> No.30045256

>>30036243
>retard wants everything handed to him for free
The stuff you ask for costs big money so fuck off.

>> No.30045408

>>30036243
Don't use EMA crossovers. The crossover happens after the move has been made. Use standard deviations above and below the moving average. I''ll leave it to you to decide what moving averages to look at.

>> No.30045638

>>30037269
>I will jizz my pants instantly if I get the numbers to go up exponentially.
That's a good feeling indeed. I spent literally 14 months working on my strategy and testing various combinations of indicators and variables and when finally I've stumbled on consistent high monthly profit during backtesting it was like finding a holy grail.

>> No.30045739

>>30045638
why didnt you automate combination and testing?
Feel like the holy grail from the beginning

>> No.30046074

>>30045739
I did, but this is just the beginning. There is a lot of combinations and variables and backtesting on literally terabytes of historical data takes time.

>> No.30046138

>>30046074
>backtesting
do you do the backtesting on just BTC or do you do it on multiple coins?

>> No.30046241

>>30046138
Not just BTC.

>> No.30046386

>>30045638

okay I suppose this is some hopium in order to continue. Did you use some kind of Hyperparameter optimization or reinforcement learning in the end to find a strat or did you just turn the dials manually?

>> No.30046442

>>30041537
lmao

>> No.30046639

>>30046386
I don't know shit about machine learning so no idea about hyperparameter optimization or reinforcement learning, but I didn't turn the dials manually. I wrote algorithm to adjust the variables after every backtesting round and run another one with new values then record all results of it, then produce report with the best results and what variables were used. I guess there is some ai magic that could maybe do it better, but I was learning as I went.

>> No.30046637

>>30040310
You need to up your profit margin and reduce the number of trades

>> No.30047017

>>30041862
>price will move in that direction
When you see humans placing buy limit orders the price will go up? Or the price will go down?

>> No.30047173

>>30044871
it's a pepe, anon.

>> No.30047323

>>30046639

this is basically a form rudimentary hyperparameter optimization, only ideally you'd want to narrow in on parameters that work and search further in and around them, and also find best ratios between the parameters. Guess I'll look into this direction.

Also how did you backtest, I have a metric encoded into my dataframes for entry and exit points calculated by the pos or neg divergence from centred moving averages. This metric basically gives each candle a score denoting how good of an entry or exit point it is. Idea is to speed up back-testing through simple correlations instead of running a loop through the whole chart.

>> No.30048925

>>30040689
very interesting, any resources to learn up more on this?

>> No.30049153
File: 722 KB, 3840x2160, ee.png [View same] [iqdb] [saucenao] [google]
30049153

my bot, started it yesterday with 10k balance

>> No.30049354 [DELETED] 
File: 515 KB, 973x775, 1614460635609.png [View same] [iqdb] [saucenao] [google]
30049354

>no mention of colocation, optimal execution or kelly Criterion ITT

>> No.30049672

>>30049153
nice - whats 'DropsDetection'?

>> No.30049692

>>30036243
are you guys programming your own bots?

started 2 weeks ago from scratch (python)

>> No.30049878

>>30049672
it buys on drop
its a strategy
the bot is called moon bot

>> No.30050462

>>30049153
Do the profits also take into account transaction costs?

>> No.30051438

Good thread

>> No.30051722

>>30049878
is it a custom strategy that you came up with or is it included in the bot?

>> No.30052229

>>30041537
>shits on pajeets
>recommends literal hue language
Yes.

>> No.30052416

where can i find historical data for backtesting a bot? Or do I need to start collecting this myself for a period first?

>> No.30052555

>>30046386
What language would you use for a machine learning bot?

>> No.30052920

>>30036568
This won't make more profit than holding a coin (like ADA).

>> No.30052998

>>30036812
This actually works. You just haven't learned how to do it properly.

>> No.30053029

>>30052416
Binance offers public data via their API. I can access kline data within a date-range of my choosing without an so much as an API key.

>> No.30053202

Have any of you tried automated futures bots?

>> No.30054556

is alertatron good for a complete CS noob?

>> No.30055091

>>30049153
Iam just curious if you would not made more money by holding.

>> No.30056648

anyone have a copy pasta like for getting started? or current github project share here plz

>> No.30056833

>>30036877
increase the difference in value from one buy to another and same with sells.

>> No.30057973

>>30049153
Nigga you mde 2.5% in 1 day. You could have hodled.

>> No.30057977

>>30056648
if ur not able to do the work to find out how to get started ur probably not cut out to make a trading bot
its not just
(if (price high)
sell buy))

>> No.30058170

>>30037494
How do I see BSC wallets

>> No.30058258

how do you guys feel about stoic? the demo account seems to return acceptable returns.

>> No.30059378

I wouldn't recommend using a trading bot in a bull run unless you don't care about the price fluctuating. I used a trading bot on cake/bnb and i did accumlate more BNB however the price dropped on me so ultimately i did achieve more coins but when it came to profit I was down. Its better to trade when the market isnt fluctating so much if you care about profits. If you don't and rather accumlate more coin for the next bull run. I would recommend using a trade bot.

>> No.30060357

>>30036243
can you link to that thread you are talking about?

>> No.30061380

>>30040689
how much money would you sell this bot for?

>> No.30062069

I have a kill switch that sells everything if the average color of op thread images is pink.

>> No.30062310

Any Pinescript chads in this thread?

>> No.30063132

How does one even get a bot if I can't Programm it myself.
Seems like noone is sharing them for verry good reason.

>> No.30063275

>>30063132

if you have an army of bots with the same algorithm each individual bot will end up competing with the others to get their asks and bids in. This is why nobody will share code for working bots.

>> No.30063381

Are there some open source backtesting / trading bots? I'm pretty good with code, I just don't want to reinvent the wheel.

>> No.30063442

>>30061380
not him but buying a bot isnt a good idea. If he sells it to you, he will most likely sell it to others, and the more people using the same algorithm , the less effective it is.

>> No.30063519

>>30063275
Yeah seem reasonable. So where would I start if I want to get into this? I'm not a retard, but not a genius either and come from an entirely different field than coding. Any starting points for midwits?

>> No.30063672

>>30063519
if your entirely different field is not one of statistics or math, you are fucked. Your starting point will be hiring someone.

>> No.30063777

>>30063132
You can buy into bot pools/personal use bots. There's a ton of them. Bitbot, bot ocean, blackfisk, etc.

>> No.30063813

>>30063672
What about a quantitative field like physics or chem?

>> No.30063891

>>30063672
Forest management. How fucked am I? I do some kind of statistics and math, like growth estimates and predictions, but that's like high school level math.
So hiring someone doesn't seem like a good idea at all in this field to me...
Are there honest coders offering such service?

>> No.30064011

>>30063813
>>30063891
just start learning and stop questioning, dont let the education system brrrap in your brains

>> No.30064032

>>30040310
This seems like a very simple problem to solve

>> No.30064595

Its profitable to build a good bot. But I will warn you guys its an all in time investment. If you aren't committed it probably won't work out.

Keep in mind your bot is only worth it if it outperforms buy and hold. Just making profit isn't enough.

>> No.30064864

>>30036568
Wouldn’t you get fucked by fees? Would you convert coin to USD or ETH and then back again?

>> No.30065159

>>30036243
How do people actually make money over the fees?
Just small compounding percentages eventually equal real money?

>> No.30065164

>>30064595
is it easier to beat buy and hodl in a bear market? meaning losing way less or even winning.

>> No.30065282

>>30063891

Just start by learning python. If you get the hang of the essentials of the language, learn how to work handle api calls in python and then how to use some technical analysis library with python. also panda dataframes is kind of a must and baisic understanding of csv files.

>> No.30065307

>>30065164
I can't answer that with a simple yes or no. It depends entirely on what kind of strategy, what kind of assets etc. But if you imagine the price completely crabbing out for a long time, yea a bot could scrape more profit there. But its not just about making more money, you have to weight this by the risk you are taking. Bots are generally riskier than buy and hold, so you have to measure that to see if it is worth it.

>> No.30065697

>>30063813
physics might be useful if you work in a quant firm. alone its pretty useless. chem is useless.

>> No.30065837

>>30065697
Is chem at least useful as a background credential to prove you're not a complete tard?
I'm also considering doing an MBA and my level 1 CFA before I look at the job market.

>> No.30065952

>>30036824
You pull out your profit every day idiot... you don't just keep going all in every trade till something fucks up.

>> No.30066622

>>30065837
algorithm/bot trading is not the market for a chem degree.
MBA and CFA are not as wanted as math/stats/coding in quant firms. They would rather you have a coding degree than a CFA.

>> No.30066754

>>30066622
Fair enough. I'll probably focus on building credentials to become a sell-side analyst or something.

>> No.30067107

>>30057973
You going to hold during the next 2 year bear market? Bots will make money regardless while you keep wage slaving.

>> No.30068200

>>30063442
i wondered about this. so say for example there's a project out there that offers automated trading, would it REALLY be that effective if it's a service that many people use? it would seem not, but idk

>> No.30068295

>>30065164
Correct way to compare is with the pairs.
If
(End amount of coin X / start amount of coin X) * (End amount of coin Y / start amount of coin Y) > 1
Print "congrats, you giant faggot"
else
Print "you're just a giant faggot"

>> No.30069286

has anyone tried numerai?

>> No.30069338

>>30068295
>(End amount of coin X / start amount of coin X) * (End amount of coin Y / start amount of coin Y) > 1
nice way of thinking about it
>>30069286
with algos its equal to your holdings. Not your private code, not your algo

>> No.30069656

>>30036243
that job will soon be automated

>> No.30069968

>>30069338
have you tried coding one?
how hard is it to make a functioning one?

>> No.30070087
File: 278 KB, 632x971, lorena.6aaaeb79.png [View same] [iqdb] [saucenao] [google]
30070087

>>30063777
>bot ocean
CTO of Bot Ocean

I'm in love

>> No.30070308

>>30063777
blackfisk is decently small, managed to make a little bit and they only have one bot running atm