Bitcoin Forum

I've developed a bot for gox trading (hosted on google app engine) which I've been using with reasonable success so far (avg profit 1-3% per day).  I am considering opening it up to outside investors and want to get an idea of the demand for such a service.  I will most likely settle on taking a percentage of the profits that it generates (as opposed to alternate pricing schemes).

It is certainly not as profitable [longer term] as simply holding at this point but it does offer the advantage of capitalizing on the near-constant noise (and is thus a good avenue for diversification).  I will, of course, over time increase its efficiency and customizability but it is already reasonably efficient and customizable.

I doubt that you've developed a trading system that will perform that good over the long term.

If it was that easy everyone would already be using such a system at the stock market.

It is relatively easy. I did also develop a bot like this, and it works quite well. The reason for this is because market is very liquid. You can buy and sell with a small profit every day, just by watching indicators (such as RSI or MACD).

I did daytrading by hand when I had time and the percentage of spread was bigger (ie: during 1 USD days, I could profit 30% weekly).

Percentage of spread I mean: a 10 cent spread at 10 USD, is less than mtgox fees. but 10 cent spread at 1 USD is 10% of spread (thus potential profit in a single sell/rebuy transaction)

Yea, it will be even more profitable when one or both of two things occurs--

1 - gox lowers fees due to competitive pressure.
2 - gox increases the frequency of price updates (or volume increases to the point where this makes sense or just happens--not entirely sure how quotes are updated currently).

But, as I said, it is a decent profit now (considering that it's a 'set it and forget it' scenario).  And, the more capital I have to throw around, the more so it will be.

Another direction I will most likely take this: a sort of meta-trader [broker] platform; i.e., executing trades on the exchange with most favorable rates (or locally).

If it works so well, why would you want to share it haha

Two reasons--

1 - I don't have sufficient capital to maximize its potential. i.e., I am sure that I could be throwing around a lot more coinage without effecting the market dynamics much (if at all).
2 - I will take a percentage of the profits for use of the bot.

Think of it like a mutual or hedge fund with myself (and my robots) as the manager.

I am all for it :D

I am willing to throw some thousand USD on it, if it proves to work properly.

I suspect you will find that with any substantial amount of coins, it will be much harder to get the same old, good, profit. You will have to take worse deals if you really want to move larger sums.

You can keep track on the mtgox prices by reloading the buy/sell page constantly,  It can go up and down extremely fast I've found.  The charts don't give you good resolution and it's at least a 30s lag.

http://bitcoin.clarkmoody.com/time-sales/
or
http://bitcoincharts.com/
or
lots of others, search these forums

Please don't hammer the MtGox servers harder than necessary.

Keep in mind that mtgox has some nice APIs that report in realtime all trades and whatnot.

You can see it working with several pages around the web (bitcoincharts, bitcoinwatch, bitcoinmonitor...) and Gribble (the irc bot, that reports in real-time all exchange transactions made on #bitcoin-market on freenode network)

Interesting idea.
You could use the GLBSE stock exchange to handle the collection of investments and paying the profits as dividends.

I'd be interested. Are you ready to deliver yet? :D

MtGox has a WebSockets API, which is feeding you depth/trade data in real-time, which is what my bot is using. No need to hammer the servers with HTTP requests.

WebSockets would be nice but unfortunately AppEngine doesn't support it yet.

Maybe it does not support them natively on the sending side - but have you checked the receiving side too?

All my web searching has indicated that neither sending nor receiving is possible.  They do have:
http://code.google.com/appengine/docs/python/channel/overview.html

which Mt Gox may want to consider supporting for App Engine clients (at least until they get Web Sockets).  Should be a pretty easy modification to the server code to support this protocol.

It's not a thing that they have to support for it to work. WebSockets are just a layer of communication over simple socket connection. You can even try to telnet the service and send in some header data and it will work. There already is some code in Python for handling WebSockets: https://github.com/mtah/python-websocket (https://github.com/mtah/python-websocket)

Or this, written specially for GAE: http://code.google.com/p/typhoonae/wiki/WebSockets (http://code.google.com/p/typhoonae/wiki/WebSockets)

Google AppEngine does not allow socket connections.  I saw 'typhoonae' but it is merely a proposal (or suggested implementation).  It requires modifications to the SDK (which are, of course, fine for testing locally) but will not work when your code is deployed to App Engine itself (which uses a standard SDK).

Also, 'AppEngine Channels' is a competitor [an equivalent but different (yet another protocol)] to WebSockets and Gox would have to explicitly support it if he wants to (i.e., via [albeit simple] code).

i'm looking into what bots are good options to throw some coins at.

How would you provide investors details of earnings, and what would be your policies on deposits and withdrawals?

The problem is here. Any trading strategy must be above buy-and-hold as a minimum criteria for consideration. And there are millions of strategies which are above buy-and-hold for only a very small amount of time (e.g. they all perform perfect on the backtesting data, but will make huge loss in a few days or weeks).

Though, there is one exception. Since the bitcoin market is just emerging, it may be inefficient and thus even simple algorithms would be profitable.

It's not a problem if you look at it as an investment of USD with a profit in USD.  Sure, it's not as profitable as simply buying and holding BTC when BTC explodes; but it is a first-order derivative and when the price of BTC is stable, it performs much better than BTC itself.  i.e., it's doing *very* well compared to BTC when the price of BTC holds at $15 for the last few weeks as it has.  deltaBTC = 0, deltaDerivative > 0.

You can also look at this as a less-risky investment than BTC itself (i.e., more diversified--akin to a mutual fund) since at any given point in time, your position is half USD/half BTC (on average).

The plan was to merely make it a fund on GLBSE.  You can buy and sell shares of that fund just like BTC itself (or mutual funds or what have you).  All calculations would be done in USD and converted to BTC on payout.