Quote from: fujiwara on April 15, 2013, 08:27:23 AM

Great! I even managed to set it up. But what about strategy.py - any examples?

Here is an example: https://gist.github.com/prof7bit/5395900

Read everything below carefully and don't run it if you not 100% understand what it does and what its implications are, including temporary loss when BTC price goes down! I'm not giving any support or guarantee for this, its only example code, I'm using it myself but I'm not going to give support for any trading bot code, you can only run a trading bot if you 100% understand what it does!

* Save the above file in _balancer.py (I'm using underscores because these filenames are in .gitignore)

* Update goxtool to the very latest version (important because there is a new slot in strategy.py!)

* Then start goxtool:

./goxtool.py --strategy=_balancer.py --protocol=socketio --use-http

* bring your account into 50/50 balance manually (exactly half of the money in USD, other half in BTC at current price, use a pocket calculator to have really exact numbers, then cancel all open orders)

* Then press p (this will place 2 orders above and below price. If they seem wrong press c to cancel them again and find the error. if anything is wrong press c to cancel orders and effectively stop the bot)

The bot will trade if one of the orders has been filled, it will then cancel the other one and place two fresh orders above and below last trade price. The bot will maintain a 50/50 ratio of your USD and BTC account balances. Please read about portfolio rebalancing to understand the implications and why you would want (or not want) to do this. This should only be regarded as exemplary code, study it to learn how to use the trading API, don't just run it blindly. I repeat: I will not be responsible for any damage you do to your money!

If you look closely you will see that it is working around some limitations of the gox streaming API. Trade messages and user_order messages will not arrive in the order we would ideally like to have, I cannot trigger any action from within slot_trade() because at this time the number of open orders will not yet have been updated (fortunately the wallet *will* be updated already), so I need to jump through one additional hoop. If you design a bot yourself then print debug output until you understand the exact order in which the messages arrive and which information is available at which time.