Making a trading bot

[Eildosa]

Hi,

TLDR : help me with the finance aspect (how do detect buy and sell opportunity) and I'll share the bot with you

I'm a java JEE/python engineer, I recently tried to make a trading bot, without success.
I can code allright, but I know nothing about finance, trading indicators, etc.

My idea was to detect a fall and sell when it happened and then detect a rise and buy at its start.

I took the burst coin data from poloniex and quickly smoothed the price chart and detected rise and fall by doing an average of the 5 last value.



And I thought, "DAMN I GOT IT, SO EASY, MONEY MONEY MONEY!!!"

but then I went in real time, and oups, the smoothing algorythym uses future value, so the end of the curve, the present, is false and always changing.
video of what I said :

https://youtu.be/eZLQm28lbVk

So here I am.
Is my strategy even worth it (selling during fall, re-buyin when it rises)?
Do you know a better strategy?
Some financial indicator could be helpfull? (MACD and such)

Thanks.

[krisnajsadrak]

you can try to use RSI,, for me its work, higher time frame , can minimize the false signal
buy when the RSI hit oversold area,, and sell when RSI hit overbought area combine with EMA 50
as pro-trader say,, keep it simple 

[stormcleric]

Hi,

TLDR : help me with the finance aspect (how do detect buy and sell opportunity) and I'll share the bot with you

I'm a java JEE/python engineer, I recently tried to make a trading bot, without success.
I can code allright, but I know nothing about finance, trading indicators, etc.

My idea was to detect a fall and sell when it happened and then detect a rise and buy at its start.

I took the burst coin data from poloniex and quickly smoothed the price chart and detected rise and fall by doing an average of the 5 last value.



And I thought, "DAMN I GOT IT, SO EASY, MONEY MONEY MONEY!!!"

but then I went in real time, and oups, the smoothing algorythym uses future value, so the end of the curve, the present, is false and always changing.
video of what I said :

https://youtu.be/eZLQm28lbVk

So here I am.
Is my strategy even worth it (selling during fall, re-buyin when it rises)?
Do you know a better strategy?
Some financial indicator could be helpfull? (MACD and such)

Thanks.

You're getting the whole topic wrong.

You must first learn to trade. Then maybe you can code a successful bot.

It's very likely you'll understand why the strategy you posted above won't work as soon as you read a bit about this world.

There are other bots already developed. Have a look at them.

[echris]

The previous poster is giving you good advice. I am using a bot successfully. But first, I had to learn to trade manually. Bots help you earn faster but you have to set the coins, settings etc.. based on market conditions. I read Gunthar's (of gunbot trading bot for Poloniex Bittrex and Kraken) posts https://bitcointalk.org/index.php?topic=1715214.msg17180502#msg17180502 and l practiced until I understood. Good luck!

[Eildosa]

Went to udemy yesterday and bought Stock Trading Ninja: Complete System For Trading Success for 10$
We'll see, but this course smells bullshit.

9 out of 10 win during trades !!!
Everyone of you will master this!!!

blah blah blah, seems too good to be true.

Edit : heh, this gunbot dude explained how his bot works, I can make my own now.

[Red-Apple]

well the only thing you are missing is the fact that there is no logic behind any of the altcoins movements! they are what we around here call "Pump and Dump" altcoins.
these are very small markets with a lot of whales who have a lot of money at their disposal. there are also a lot of pump groups who have a lot of sheep following them!!!
they give a signal and the sheep start buying that shitcoin and losing money while the pumpers sell and make money. you see this on the chart as a big rise out of nowhere and for no reason!

Is my strategy even worth it (selling during fall, re-buyin when it rises)?

this sounds good though. i have been doing this from time to time. if you can find the rises you can always make a quick profit. but the problem is being fast to act on it both for selling and buying.

[aventus]

Went to udemy yesterday and bought Stock Trading Ninja: Complete System For Trading Success for 10$
We'll see, but this course smells bullshit.

blah blah blah, seems too good to be true.

25 years of trading financial markets makes me say you are right: 100% bullshit for n00bs. You will never make real money. As for the ratio, mine was more 2/3 of loosing trades. But huge profits for small losses.

[Eildosa]

there is no logic behind any of the altcoins movements!

I don't know, I recently learned that when the price goes out of the bollinger band then it suggest a change in trend.

What if everyone knows this and decide to buy/sell because of this indicator, won't it act on the price and make the indicator right.

[Eildosa]

So I coded the interface with poloniex and made this :

candlestick plotted from poloniex data + EMA24 and EMA48 (5min period) and bollinger bands computed from poloniex data
https://ibb.co/gVSfha

If I follow gunbot logic I should buy when the value is below the two EMAs and sell when it's above.
In order to maximize profit I also computed the bollinger band.

My logic is the following :

when price is below the 2 EMAs lines AND "close enougth" from the lower bollinger band : buy
when price is above the 2 EMAs lines AND "close enougth" from the upper bollinger band : sell

I think I'm going to make a simulation with genetic selection on a 1 year of burst coin prices and let it play.
the 3 changing parameters would be EMA1 windows, EMA2 windows, bollinger window.
 
I'll also have to add failsafe, if the price rises too much and settle higher then the bot will be stuck, same is true for the opossite.
I should also make a "scan" module, that can quicly analyse every currency pair on poloniex and tell me which one are optimal for this strategy (oscillating around a value)