Cryptocurrencies, stocks, algorithm trading

[Garquillex]

Originally Posted by xXOpkillerXx

Why not do machine learning with "predictions" as output and the trades you do manually as dataset. That way you get to learn even more about the params needed for a good trade, and you might even see surprising results in the weights in the end. Just an idea...

I want to work with genetic algorithms eventually, but I think for now I'm a lot smarter than whatever I can teach a bot to learn (and it could take forever and be difficult to control).

I have like 4 months to make this profitable. I think I can do it by September if I work hard enough.

Originally Posted by 25thhour

I got the money if you want to split profits once your trading bots are tested more. Lets get rich. I do some Forex trading, and love investing.

Maybe if you're serious. I only have 1k to work with while having a safe bank account, and a terrible job I may or may not return to on August 26thish.

I'm splitting off into stocks and forex later on to keep up the diversity and reduce slippage.

Originally Posted by xXOpkillerXx

Well that's up to you. If the AI was trained for long enough, there shouldnt be a problem. In any case, it's easy to add a manual confirmation feature to it.

I spent a few weeks manually trading and I want to get as far away from it as I can, at least when dealing in a 24/7 market. You have to close out trades or never sleep, or risk losing money while you sleep. I just never slept. Except for sleeping through every single bitcoin price drop so far. Now I have an auto-sell to USD system in place that also plays loud music to wake me up.

Originally Posted by Soundwave-

I mean this is basically the point of backtesting, although I'm not sure if traditional machine learning has been applied to it before.

I've wanted to develop some open source procedural trading software, but the problem is to test it well you either need money or market data, neither of which are amazingly easy to obtain.

Bitcoin market data is easily available. It's the natural place to start considering you apparently legally need $25k in an exchange account to be a day trader.

Only problem is they only have it available in 1 minute chunks, which is usually fine, but a really good bot reads live order books too, which makes backtesting that logic essentially impossible without an enormous effort.

Originally Posted by igotrhythm

But programming bots and getting involved in Bitcoin, especially after the last Halvening? I'd like you to be in my Maki Dev Slack with a bunch of guys that are looking to re-make the Internet and "decentralize ALL THE THINGS." PM me if you want a link.

What's this about?

[xXOpkillerXx]

Originally Posted by Garquillex

I want to work with genetic algorithms eventually, but I think for now I'm a lot smarter than whatever I can teach a bot to learn (and it could take forever and be difficult to control).

You do realize that all you can make your program do, you can put it in a deep learning algo right ? I mean, if done right, the learning AI should Not yield lower results than the base settings you give it. But yeah, you gotta know that stuff fairly well ofc

[Garquillex]

I went to bed literally the minute that started ^_^
And my bitcoins were moved off bitfinex so I'm safe, but if bitfinex goes bankrupt...

I can't program if I'm glued to short term trades so I have to ignore prices as much as I can. I can go days without looking now.

Originally Posted by xXOpkillerXx

You do realize that all you can make your program do, you can put it in a deep learning algo right ? I mean, if done right, the learning AI should Not yield lower results than the base settings you give it. But yeah, you gotta know that stuff fairly well ofc

Yeah I'm already plotting code for that. I was under the impression of more of a trial and error tabula rasa bot rather than something more like self adjusting variables. The former is super interesting.