What bot to use?

Anyone know how to get orderbook snapshot histories on altcoin markets?

Polo has an API call specifically for historical data
>poloniex.com/public?command=returnTradeHistory¤cyPair=BTC_NXT&start=1410158341&end=1410499372
Start and end are specified as unix timestamps. Returns up to 50,000 trades within the time period

Bittrex doesn't have anything similar afaik.

looks cool, I've done some beginner level java and C# can this be pulled off with javascript/php?

How do you write such beautiful UI?

>javascript/php?
I'm not big into web development so really don't know how far you can push those two, but either way I wouldn't recommend it.
Doing TA for a dozen pairs in real time is pretty cpu intensive, so performance is important.

Friend is a graphics designer so she's helping me out a bit.
Both of those are winforms actually, believe it or not.
Don't like WPF cuz markup syntax.

Very interesting, which language?
I've been researching on writing my bot and feeding it some btc as a test.

Are you doing only TA or maybe ML algo?

I just grab 24h ath and 24atl compare it to current price + some max spread between current and 24h ath, and then buy with trailing stop loss of 1%. Only volumes > 150 btc. No sell order if its a net loss (better to hodl in this market). Srry an on phone.

C#, interface is winforms cuz im oldschool like that
It's purely TA, still trying various predictors and strategies.

>I just grab 24h ath and 24atl compare it to current price + some max spread between current and 24h ath, and then buy with trailing stop loss
I don't get it, what's your buy trigger? Where's the price supposed to be relative to the 24h ath and atl?
Is it a mean-reverse or..?

Dont try to understand my logic, its prob shitty. It gave me some good returns last week, but thats because the market was in upswing generally.

Snd theres no 'buy trigger' it just buys at ask price the stocks that my scanner returns (likely to moon) if theres funds in my wallet left.

That's purdy. I'm diving in with LSTMbots.

Ah, I see, it's a different approach.
What I do is monitor the price in real time and trigger a buy if certain conditions are met.

Still I'm curious how your strategy works. I've got price data off polo for the past 7 days or so, so I want to put it to the test.
Where's the current price supposed to be relative to the 24h ath and atl for it to be considered "likely to moon"?

Basically, the closer current price to ATL the better, and the ATH shouldnt be too high. Those things mean low spread (not dumped) and close to lowest resistance point. Thats the extent of my logic lmfao

I have only drafted my bot last week, like you i have to funetine it etc but sadly i like time during the day. I know some programming, once the api is down the real issue is the trading logic...

nice shilling

So it's basically a mean-reverse of sorts.
Will run it on the data I have and see what happens.

The trailing stop-loss is exactly what I'm using now to trigger sell btw, but I do also have a standard stop-loss at -15%, just to make sure it doesn't hodl dooming coins.

>once the api is down the real issue is the trading logic...
Sadly you're right. It took me about 2 weeks to implement the API, and that was about 3 months ago.
Ever since then I've been experimenting with trade strategies and still haven't found anything reliable.

I started off looking at the moving averages, I'd buy when things looked bullish and sell when they started to calm down (usually 3-5% profit, sometimes less)

but my bot wasn't making enough trades, hodlers at heart, so now I'm looking at using neural nets to try and ride the waves down as far as as possible before buying, and then ride the waves up as much as possible before selling. Lots of 8-12% trades in a day on various altcoins. I've got about 6TB of market data over the last month or so.

>6TB
Lolwut

I get data.

Price patterns with ANN's, tried that. Made some good trades but the fees ate all the profit.
Will test how it performs with a longer timeframe. Price movements are higher so it should be able to cover for the fees as well.
Have a few more things to try before I go back to that tho.

One thing I noticed is I got much better results when I fed it price offset (from SMA) rather then the normalized price values themselves.

Just looking at price is kinda lame from my results. Better results when you consider things like volume of buys/sells within a range of percentages from the ask, frequency of trades, volume of trades, etc. I think some people call this 'attention', or 'interest'.

Why don't you do some research into the lemonparty before you knock them. Check it out. They are going to revolutionize the blockchain.

lemonparty.org