>be me, building a trading bot for ETH/USD on gdax
>run historical data sim
>+72% from this past week
>+450% from this past 30 days
>44% win rate
>2:1 win:loss size
Im pretty sure ive calculated this correctly. I heard that i should only expect 30-40% of what the historical sim returns, but im still nervous as fuck
A-am i going to make it?
Also, what is the lowest sleep time i can use to call histdata from gdax before it warns me about spamming? Im using sleep(0.5)
Other urls found in this thread:
kaggle.com
stackoverflow.com
api.bitcoincharts.com
kaggle.com
cryptodatadownload.com
github.com
oakton.edu
twitter.com
fuck i need to (re)learn python. learnt it in school over 10 years ago and never bothered with it since :(
Sleep 0.5 should be fine. I recommend to use the feed though.
What is your strategy?
resources to learn this shit op? i am a developer and have known python for years but know shit about trading with code
Been thinking about making AI for crypto, whoever does it, good job!
You have to actually get filled when trading live. This could potentially wreck your returns, depending on your strategy.
Would love any guidance you have about how you put this together. I'm a data scientist and I'm quite familiar with building time series models. I've also done some work with Python so I don't have a problem with picking it up. What I'm not familiar with is how I go about actually getting the data to train the model, how to get live data fed into the model to produce my outputs, and how to have the model talk to the exchange and make trades.
What do you want to trade and what kind of data do you need (as in OHLC, tick data, order books, ...)?
googling seems to point to tons of big datasets, such as kaggle.com
i imagine you could fill in gaps with other datasets for your model (after formatting and stuff and assuming theyre consistent or from the same source)
not to mention the api for coinbase lets you pull some historical data too, though i don't know how much
stackoverflow.com
Thanks for this! Something to start with for sure.
I guess I figured I'd start with something like BTC or Eth, but really I figured I'd build models for various coins and see what performs best during validation. In terms of actual data, I mean really just everything I can get get my hands on that's relevant to the coin and I'll figure out from there what has statistical significance.
The complete tick data and order books are quite expensive (at least the sources I know). You can find free tick data here api.bitcoincharts.com
>developing on windows
You are not going to make it brah.
you could use DOS and be successful. fucking brainlet.
>setup a bot an hour ago
>already made $0.15
Check out numeraire, might be something you're interested in.
What said
Trading strategies work until liquidity issues arise, and you become a buy/sell WALL and your orders don't get filled fully.
you do realise if this worked to the extent you could make significant reliable gains, people with the same or better knowledge than you would be doing it
it's always surprising to me to see the number of anons with obviously decent programming knowledge fall into logic traps middle school mathematics should save you from
>people with the same or better knowledge than you would be doing it
no shit retard, what do you think hft trading and algorithmic trading firms are? people ARE already doing this
>you notice $100 on the ground and you tell your economist friend, ‘hey, there’s $100 on the ground”
>your friends says “no that’s not possible because markets are efficient and someone would have already picked it up”
OP here
Yeah that was my thought. I know the larger my orders, the more i have to cut profits
GDAX offers an API that lets your program hook into its system. I did some digging and found a python wrapper to make that connection easier on github. Reading the wrapper code, its actually simple to do yourself
>whats your strategy
Buy high, sell low with negative leverage
let's rephrase it another way: what makes you think your homegrown bedroom project will beat hypothetical teams of experts?
yes, a location-specific temporary event is exactly the same as a global market accessible 24/7 from anywhere in the world
If you use a Mac you won't either. Devs used to use them but Apple quality has gone downhill and Linux Desktop quality has gotten better so we all jumped ship.
obviously not, i never claimed that and never claimed op would have a lot of success. if anything, this is a great opportunity to learn more about ML and its still worth pursuing
Jesus, you don't have to tell the complete strategy. The general concept would be interesting though. For example the average duration of your trades? Is it built on regular indicators like the RSI in your picture? etc
Also, what is you sharpe ratio if you know it?
Last month was bull, test it on February data
That’s the beauty of it. I don’t need to beat them. I don’t even need to trade the same timeframes or markets as them. Not all markets are equal in profit potential.
I remember I was reading the reddit post about people earning from making arbitrage bots on crypto exchanges, the most common comment was:
`I wanted to do something similar, but thought "many other people must already be doing this, opportunities in that method should already be scarce and not worth it`
And so those that actually went through with it and did it last year made HUGE bank.
Don't forget, crypto is a babby market, a LOT of big players aren't in yet, so things that might not work in traditional markets MIGHT still have a chance of working for a while here.
I've been reading some penny stock traders posts (mostly Tim Sykes students), and they all also call crypto something similar (easy mode market, babby's first market, etc.)
DELET THIS
>You muse to your economist friend. "Hey there could be a 100 dollar bills on the ground all over the place. I only have to look for them and pick them up"
>Your friend says "the odds of being the first one to spot and pick up the money are very low, and there are hundreds of thousands of professional street urchins already running there own loose money programs"
>oakton.edu
holy shit an actual baby boomer browses Jow Forums
how do you feel about all the "30-year-old boomer" posts OP
Nigga gdax only offers 3x leverage. Howd you get these returns?
Fuck off with your retarded EMH theories poorfag
Lol.
t. started with $2500 and made this during the bear market with a bot
no u didnt
>seething
trading eth and btc?
Were it that easy. I trade everything that moves on over a dozen exchanges. The profits cycle though. Some days, liqui.io is popping, other days, it might be tradesatoshi, then bittrex will bless me, then binance, and on and on. You gotta have a pattern with a good probability for success then you beat it like it stole something. Easy as that. I'm on track for 100 btc by this time next year. Hopefully more
savage. what's your strat? market making, (statistical) arbitrage, mean reversion?
machine learning or your own strategy?
My own homegrown strategy. Can't give it away, at least not until I've made it but I stumbled on a pattern that pops up a couple of times a day at random on any given exchange. The trick is.. well, I'll post that part in a year or two. Sorry. Just know that it's possible.
Did you write this bot yourself?
This I can answer. Yes I did. Python, node, ccxt, lomond, websockets, matplotlib for testing, and cryptocompare for historical data. That's all you need to get rich.
Whats the downside of just giving us the strat? 99% of the these pajeets will write it off as faux anyway.
I sure will appreciate it however.
What degree do you have?
signed message from btc/eth wallet?
>inb4 "i'm worried about my privacy from one address i can tumble easily"/"i keep 200k on centralized exchanges all the time without ever hedging my risk"
>Went from 2500 to 25.
Holy fuck
How did you mess up so bad
>what is overfitting historical data
Nice try pajeet. Run the trading bot live and see what happens
The downside is that other people start taking his trades and his strategy stops working.
How do I follow you so that I can keep track when you release your strategy after a year or two?
It's a strategy I've never seen anyone use with a bot before. No TA, no FA, nothing. Just a pattern that seems to work. It can be wore out though cuz it only shows up a couple of times a day when combing through hundreds of coins. I know that eventually somebody just as capable as myself will stumble on it so I'm really keen on maintaining my edge until I make it. I want everybody to make it but I have to be stingey for now.
BS in compsci
I keep 200k on centralized exchanges all the time but I at least hedge that by having it more or less equally spread out in dozens of spots.
This is biz. Obviously I buy high and sell low
Give me some way I can contact you. No, seriously. I won't forget.
>same pattern
>couple of time a days
>at random
>on any exchange
I know you won't reveal it (of course), but at least can you be a little bit more specific.
Different shitcoins or the BTC/USDT pair?
what is lomond?
All shitcoins all the time. I'll trade anything with my only caveat being there be some kind of volume. Speaking of volume, here's a little tip. The bot used to buy whatever's available but after bagholding a few too many times, now it never goes deeper than 1/7 the daily volume of the coin or 1 btc whichever is lower.
A nice websocket library for Python
Post wallet
Damn, I'll hit you up with my contacts if this is possibu too.
it's striking how every single successful arbitrage bot computer scientist on Jow Forums has limitless amounts of time to spend alluding to their strategies and posting screenshots from an app where the data entered is arbitrary, but there's always a good excuse as to why they couldn't take two minutes to provide unfalsifiable proof through the very technology they allegedly spent months investing in
gullible bystanders would do well to research "signalling" in economics
>using windows 8 on a thinkpad
kek, this
Jow Forums loves larping
Its just an endless cycle of some pajeet larping and other pajeets begging against his lies
bro hes not LARPing he posted his blockfolio. its not like people can just enter random trades on blockfolio or anything
Do it. I'm not kidding, I won't forget.
Your cynicism is understandable. I don't believe much of what I see on here. But, I mean, shit man, I trade constantly and I can't trade coins if they're in cold storage. It don't work that way. I'd screenshot my exchange holdings but I mean how hard is inspect element? I usually judge the veracity of what I read here on the surrounding details. Does the poster sound legit? Is he a native English speaker? Does he give any supporting details that check out? I mentioned the technologies involved, I mentioned a volume strategy as an aside. You be the judge. Here's another old blockfolio.
Hope you dont forget about me
Feel free to tell me whenever, I'm not gonna tell anyone either, and I can't write/use bots or anything so I'm just gonna be doing it as manual trades anyway.
I gotcha bro. I screenshotted your post cuz I'm too lazy to cut and paste. I never delete anything though so I'll get back. Probably just after you forget all about it lol.
Get me in that screenshot too bro.
[email protected] (not my real name btw)
I was on your boat, and then I realized two things: one, unless I'm placing orders, I'm going to get eaten by fees. And if I'm placing orders and no one eats them, I'm fucked. And two, the real reason the strategy was so good was because the stop loss was so low (0.5%). In fact, buying and selling at random in 15 minute intervals with the same stop loss generated returns that were still very good, which means my strategy was shit. No offense, but if you're asking biz, you're probably in that boat. Keep iterating.
How old are you btw? From the states?
You're in.
40 year old boomer. Sunny south Florida
I love that this is even possible. Gives me hope.
I won't forget, I REALLY want to have a working trading strategy.
I'm a NEET that has nothing to do but trade all day, and I can't seem to find a working strategy despite trading 16 hours a day, ending up barely above break even after a few months, so it has been REALLY frustrating.
Wyckoff does another boomer well.
Damn nigga. Are you the same nigga that posted a thread about buying and selling randomly in 15 minute intervals maybe a week back? (the thread pic was a complacent pepe with coffee, or something to that effect).
Or is this strat quite popular on Jow Forums ?
No, but I did make a post to similar effect stating what I said in the post. I don't save pepe images. And it's not a real strat. It's just to prove that any strategy that relies on small stop losses is inherently trash.
God bless your soul.
Well, if it's not a problem add me to the list.
[email protected]
(my Jow Forums shitty email where OPs and people never actually deliver).
Still worth a try, thanks bro.
let's judge you based on the surrounding details, then
do coins need to be on cold storage to post a signed message? no, any btc/eth address will do, so a transfer in and out only takes as long as the exchange's withdrawal times. mentioning cold storage at all could be a slip of the tongue but it could also make you sound either unfamiliar with the technology or trying to mislead those who are unfamiliar with the technology
do coins need to be on your address now for you to sign a message? no, past history would do. while possible, it's difficult to believe your coins would have never gone through a btc/eth address you control at some point
what was your entry post? a simple "lol", and the second post followed with "seething". while we all like shitposting, it still opens the possibility your presence is ego driven above anything else
it comes down to this:
1) the pool of english-speaking people who can namedrop programming concepts vastly outnumbers the pool of people who can build successful automated strategies
2) an anonymous larp has multiple upsides (ego boost, trolling, possibly get contact info from gullible people you can sell useless software or even malicious keyloggers) and no downsides
cynicism doesn't come into play. it's simply not rational to believe you without tangible proof
Please let me in if possible. Thank you.
Created a bot that traded off the 15 min macd (and rsi to some effect) with a small af stop loss. Used it during exams as i didnt have the time to trade. Was up about 5%. Though sample space too small to say if the bot works well.
I understand. I couldn't make shit happen for the first few months if you notice my blockfolio above. I discovered the pattern and made the bot immediately which is when my success jumped through the roof. Like a light switch. One day, nothing, very next day, everything. Keep striving in the meantime though.
I gotcha. I'll get off my ass and actually make a text file. I'll put it in my dropbox so I won't lose it.
Homey, I just got into this shit back in July and before that, I actually believed in a weak form of the EMH so that should tell you something. I'm a trader, not a blockchain expert. That said, one of my exchange wallets has like 9 and a half btc in it. How do I sign a message or whatever?
Thank you
Small stop losses are great in backtesting and fall apart in practice. I've seen BTC lose 1% in an instant before order books could catch up. Either you fill orders and get eaten by fees or place orders too late. Either way, in practice, any strategy with small stop loss (under 2%) is garbage.
Bruv it's all about those NodeJS wrappers nowadays. Shit's asynchronous as fuck.
Fuck, man. I want to believe.
Can you give some basic advice to someone who could only 5x over the past year? I want to /make it/ so bad and just be done with this shit.
Are you the same guy claiming he was up 60x thru the pink wojak on an earlier arbitrage thread?
well i had the idea back when LTC was 5% off of gdax.
i figured it would take me two weeks to do it, and someone else would already do it (and the exchanges would have an unbeatable latency advantage)
sure enough a few weeks later i never saw arbitrage opportunities again
I've got a memory like a speeding truck man never forgetti. [email protected]
Pretty much. Except I'm still hard at work so now I'm up 84x. Would be a bit more but, you know, taxes. I don't arbitrage as a specific strategy though. I'm on so many exchanges that every so often, I'll have a bag that happens to be going for more somewhere else so I'll send it over but I don't do it systematically. I should though. It's not hard and there's money to be made for sure.
Basic advice is buy lower. If you look at something and think, this is it, I'm buying low, I'm about to get PAID, set your buy 20% lower than that.
Node is awesome with the whole nonblocking io but don't underestimate what can be done without that. I thought node or nothing then I ported one of my scriots to Python and it worked just as well and I depend on executing trades asap.
you can't sign a message from an exchange wallet, you need to own the private key. not sure how much blockchain expertise is required to understand the basic concepts of it. this is all one google search away
programmer with the technical chops to build a successful arbitrage bot in a competitive market needs to be told where to find basic info on computer technology, and had no interest in said technology during all these months... idk man. probabilities are still firmly biaised towards trolling, with a newest touch of "look at me, i don't even know how your shitcoins work but i make 6 figures off it". very effective in the same way people get angsty at trust fund kids
your story is possible but it isn't probable
u realize he is larping right?
there's no one out there with some super duper secret strategy that makes 100x the market average.
but there are a lot of people that will claim to but never reveal their wallet.... lol
Point blank. If by any stretch of the imagination what you're saying is true I'm beyond jealous, so my mind has to deal with it in some way.
Burner email. I like talking to other bot devs anyways might as well see if I can be entertained. Two way street as well I'll share my super secret strat soon lol.
wow, if this is real you'll be very wealthy within a few years. gj
>there's no one out there with some super duper secret strategy that makes 100x the market average.
There are, I doubt they brag about it on Jow Forums though.
I can code a trading bot but the hard part is devising a strategy... That's the real beans. Any tips to a fellow programmer?
Why should he stop the bot (stop making maximum profit) to move coins off the exchange to a wallet to prove you wrong
>Oakton ANTI - RACISM Team
niggers
Yeah, I didn't think you could. Just wasn't 100% sure. I'll tell you a story. I have a degree in this shit but for various reasons, I'm not employed. I worked for a few years on server side web stuff, then did mobile apps when that got hot but I got out of the rat race a few years ago and got into making affiliate sites. Built an automated splogging setup and have been cashing checkes for a while. I only publish for one company though and back in July of 2017, I checked my earnings and saw zero. Thought they were having issues so I waited the next day and the same thing. Zero. I emailed and got nothing back and came to the slow realization that I'd been booted. At one point I couldn't log i to the dashboard. I was like fuuuuuuck. Started brainstorming alternatives and settled on day trading stocks. I figured if others can do it, I can too. Saw /rhg/ when that was still a thing here and downloaded Robinhood. Fucked around with it for a few days and wasn't impressed. That's when I created my Coinbase account and the rest is history.
I read Saroshi's writings, read some stuff by Antonopolus or whatever his name is but, I want to make dollars. I could become a blockchain expert and beef up my technical knowledge or I could focus on what is working and trade this shit. I made my choice. I might not know everything about blockchain or Solidity, or sha256 and on and on but I know how to make money. If that seems like LARP then oh well. PS, my affiliate account wasn't banned, they just had some issues so now I make money from both.
It's real. Thanks.
He seems legit. Honestly it seems plausible I don’t know why you guys think he’s larping. Botbro, just come back and when you’re ready to share strategy plz when you made it. We’ll be here and listening. If they think you’re larping then forget them