Author

Topic: Algorithmic market movement prediction - what do people use? (Read 931 times)

dsp
newbie
Activity: 52
Merit: 0
Thanks, but isn't that something one would gain a significant advantage from lower network latency? I guess you could could come up with some method do calculate "significant" time periods...

I was thinking about hooking it to a web site and letting people set their spreads. Etc and let it manage a few BTC for them, With MTgox new API you can delegate what you will let some one do on your account. This type of bot if it controls enough capital can help to stabilize the price. As it is always in direct opposition to the price movement.

dsp
newbie
Activity: 52
Merit: 0
Thanks, but isn't that something one would gain a significant advantage from lower network latency? I guess you could could come up with some method do calculate "significant" time periods...

Na just check the price every 10 seconds. miliseconds dont matter in scalping. Plus if you hit got that often you will get you account canned.
hero member
Activity: 602
Merit: 502
Thanks, but isn't that something one would gain a significant advantage from lower network latency? I guess you could could come up with some method do calculate "significant" time periods...

Network latency is not really relevant here. The number of trades per day are really low.
legendary
Activity: 1666
Merit: 1057
Marketing manager - GO MP
Thanks, but isn't that something one would gain a significant advantage from lower network latency? I guess you could could come up with some method do calculate "significant" time periods...
dsp
newbie
Activity: 52
Merit: 0
I am wondering what people use to control the behavior of lets say a trade bot.

Fourier Analysis? Wavelets?
Matching the data to an elliott wave?

What other methods are there besides high frequency stuff which could be useful without buying a server next to mtgox. Lets say I want to make only a few transactions a day or only place orders. Any suggestions on where to read up on this?

I'm using a bot written in access less than 300 lines, it reactive, not predictive. Basicly is a scalping bot using a Cash to BTC ratio. It keeps lets say 50% in each, this is another variable. If it drops 5% it will buy %2.5 relative to the total value managed. If it increase 5% it will sell 2.5%. I have been playing with the ratios and how much I let it control. If the bot goes out of threshold based on last price is triggers an order. There is a little more secret sauce I'm working on but that's the basic premise. Larger the magnitude of movement the larger the Buy Or Sell. The idea is no matter what happens the bot lives to make another trade.
full member
Activity: 210
Merit: 100
I thought you need a lot of money to put up many buy and sell walls.
legendary
Activity: 1666
Merit: 1057
Marketing manager - GO MP
I am wondering what people use to control the behavior of lets say a trade bot.

Fourier Analysis? Wavelets?
Matching the data to an elliott wave?

What other methods are there besides high frequency stuff which could be useful without buying a server next to mtgox. Lets say I want to make only a few transactions a day or only place orders. Any suggestions on where to read up on this?
Jump to: