Thanks for trying it out!

Well there was a bug in Gekko since around friday: Because of wrong error handling every try to buy/sell would result in Gekko trying to refetch trade data with the error message above (instead of buying/selling again). This is fixed now so you should download the latest version.

Also check that the default currency in your Mt. Gox account is USD. Gekko will support all currencies supported by Mt. Gox in the near future. But for now you have to set the default currency to USD or disable real trading.

----

I am currently adding a portfolio manager that will take care of a lot of trade things that are really messy now:

• Proper buy and sell order on BTC-e and Bitstamp by first checking the lowest ask / highest bid. And checking afterwards wether the order went through and if not retry.
• Better error handling, verifying real trade settings
• Better profit tracking based on your real portfolio instead of virtual balance

Community Request

It is not fully working yet but with this portfolio manager I am proposing a universal way at which different exchanges could be implemented the same way through different (javascript) projects. I am talking about the `portfolio managers expectations` and the `trading method expectations` as described in docs/Exchanges.md. I think standards like this would really speed up the development of different projects. If anybody has any ideas of things that need to be different / added to the document please let me know (or submit a pull request if you are a developer).

On the options panel / page there are settings for the thresholds. When you click the add-on button in Chrome you should see a table, the last column are the thresholds. Whenever the threshold of the most recent tick (so the top right box) is below the sell threshold the bot will sell all your BTC, when its above the buy threshold it will buy BTC.

^ that is the basic story, minor variances depend on which version you got, how many thresholds need to pass (in config), etc.

Update

• The profit calculating options are now configurable.
• Gekko now supports Bitstamp (alpha stage).

This means that Gekko now supports the three biggest exchanges. The auto buying and selling should be considered alpha (but is more stable than the BTC-e counterpart I think).


Thanks!

Gekko is really early stage at the moment and one of the biggest issues at the moment are automatic placing orders on exchanges that don't support MKT orders through their API. Gekko also isn't keeping track of the orders, it's basically fire and forget (really unacceptable for a trade bot). Gekko doesn't have this problem at Mt. Gox because it will place a MKT order and Mt. Gox will figure out what the market price is and make sure the order gets filled.

So the biggest todo will be to improve this and make sure that when Gekko detects a trend it will make sure all it's orders are filled at all supported exchanges.

I'm really curious on you are going to solve some of the problems I'm running into (like altcoins data for example, see couple of post above). My goal for Gekko, besides learning, is to improve the quality of the open source trade bots out there .

I'm going to try to be there, can't promise anything.

Thanks for the research! I don't think it's a good idea to put a scraper in Gekko: It's not really nice towards the website and if they change a tiny thing in the HTML everyone has got a non working trade bot until I fix Gekko and everyone updates (manually).

The good news is that `ltc.block-explorer.com` pulls the data from an API (as you've shown). However I can't find any documentation on the API, so I don't think it's wise to use this private API for the same reasons. I think I will email the guys behind the site to ask.


Cool, thanks for testing! Let me know if Gekko runs in to any trouble. I've also got Gekko running on BTC-e since the last update (almost 24h) and it just passed a threshold:



(and the math seems to add up)


As for the feature: Simple request and I hope I'm able to do it tonight, else tomorrow

Big update!

I've implemented some new features yesterday, the bot has been running overnight but it still needs a lot of more testing to determine how stable it is.

Gekko is now able to watch trends at BTC-e

It appears that it is not possible to get historical data from BTC-e, so when you set it to watch btc-e Gekko will use data from bitcoincharts instead. You can set Gekko to watch and trade at these different currencies at the moment: USD, EUR and RUR.

Gekko cannot handle altcoins at the moment because I can't find any datasource (API) for them.

Gekko can now automatically mail you on new advice

This currently only works for Gmail and Google apps accounts. To enable this you need to set your email in address, you also have to provide your email password because Gekko will use your email address to send mails to yourself.

Make sure you've looked at the code before you provide your email address and password. Under the hood Gekko uses the emailjs module to send the email. Gekko is in NO way able to read your email or whatever and your email address and password are only sent to the gmail servers (over SSL).

(I don't like this solution very much, but I'm also not comfortable with shipping an API key for a service like Sendgrid. If anybody has a better idea that works cross platform, please let me know)

You can set the new features in Gekko's config.

Helping with Gekko

To stabilise Gekko and remove all bugs we need a lot of testing. If you are interested in Gekko you could set it up in your system and let it run (without auto trading disabled for example), this would speed up the process by a lot!


The way Gekko is setup is that it implements trading methods. Right now it only knows about EMA. With the rules you are describing the best way to go would be to create a new trading method. Before we can go at this we do need all the rules clearly defined. I would have to give this idea some thought.


Thanks! Note that Gekko is at this moment in no way an advanced trading bot. I also would like bitstamp in there, but because of the problem you are describing and the fact that there is no NPM module for bitstamp at this moment (I would have to create my own, not hard at all but it takes some time).

I'm still figuring out a decent way to solve the problem you are describing (BTC-e has the same problem). I was thinking of determining the price based on the last trades and the way the trend is heading at what speed. But that would be pretty dangerous as Gekko is not only speculating what is going to happen but also at what price it should buy.

I think there will be a threshold that determines (when buying for example) a higher price (more sure to get filled but less profitable). I don't think there is historical data about the order book which makes backtesting not really possible (have to look into this per exchange). I also think Gekko should monitor the orders, if they don't get filled after x minutes it will re try at a different price. (It isn't even doing that right now).

I'm open to suggestions!

About the algo: EMAs popularity is caused by Goomboo. But Gekko is currently setup so that new trading methods could get implemented rather easily: Describe your trading method in Gekko's methods, receive a object which fetches trade data and emit an advice event when you want to buy or sell.

I'm open to suggestions, what method would you like to see in Gekko?

Good idea, also would be really easy to implement. Added it to my list.


Yep, it's on the list of exchanges I need to add!


It is saying that `mtgox-apiv2` (a dependency of Gekko) needs a more recent version of nodejs. It needs a version above 0.10 and you have 0.6.19 installed.

Quick tip: you can install something like nvm to easiliy manage and install new versions of nodejs.


So you mean when the lag at Mt. Gox is really high trade at other exchanges beacause Mt. Gox is lagging? Because what information can we get from knowing that the lag is really big? Also all the times Mt. Gox is not under stress the the lag is low.

Every day I'm getting the feeling more and more people feel the same way. Gekko currently does not support this but will in the future, stay tuned!

Sorry misread your first message, levelDB looks way better then trying to implement flatfile really bad myself. Storing trade history is on my todo list and I think I will use levelDB for it. Thanks for the tip!


Thanks!

Currently Gekko is basing all his advice on what is happening at Mt. Gox (even when you have automatic trading set at BTC-e, because for BTC every exchange will follow the trends happening at Mt. Gox). I'm planning a new feature where you can configure on what exchange (and what currency) to watch for trends

Note that other features (mostly backtesting) are higher on my todo list at the moment.


Haha thanks man!

Interesting idea, but all EMA bots at the moment are based on the Goomboo's Journal. Goomboo, as a trader, suggested that the more simple rules such a trading system has, the more profit it will generate in the long run.

I'm currently working on implementing a backtesting feature for Gekko, when this is done it's pretty easy to test such a system against a lot of data (say the last 12 months). That way it will be a lot easier to say wether such a system will have more success. (Note that are a lot of trading tools already available that support backtesting).

Ow that I understand

BTC-e doesn't let me do a 'market price', I have to state at which price I want to place an order. When you have enabled auto trading and Gekko identifies a trend a couple of things will happen:

• If you configured Mt. Gox -> buy/sell at 'market price'
• If you configured BTC-e -> look at all the BTC-e trades in the last 20 seconds and calculate average price, now buy sell with this price.

This is just a workaround to first grab the latest price and use that. Do you suggest a special strategy for 'strong orders'? Because right now Gekko will never do a strong order on Mt. Gox because it can do a market price. I can make it configurable so that Mt. Gox will also have a 'strong order' option, but what price should Gekko use or how should it calculate a price?

Gekko is being actively developed at the moment

Gekko now reports the profit it generates or would have generated (wether you have real trading enabled or not)! Gekko always simulates trading based on the EMA advice and reports about it every interval.


Read more in the profit reporting section in the readme. All configuration has also moved to a handy single config file.

Note that it is possible to make profit / lose money without doing any trades, due BTC changing in price


I don't quite understand it yet, because I can't find what you mean by strong order / market price strategy (I know very little about trading, I learned last week what a order book is for example). If you can point me to some links I will research it and try to put it into Gekko

Upcoming features

• Backtesting different EMA settings
• More exchanges

I may be interested if it's in Amsterdam. But monday is not a possibility for me, just like any other city than Amsterdam.

Almost: EMA means that when calculating a new EMA it will take three factors into account: the current price, the last calculated EMA value and a weight factor*. This does mean that for the first EMA you want to use to be accurate it needs to have build up a history of EMAs to compare the current price with.

*The weight factor is just how important the current price is VS how important the last EMA is. So therefor a higher weight like 21 (a EMA21 means the weight is 21) means the last EMA is more important and it will move slower towards the price (lag behind a trend).

So EMA trade bots calculate two different EMAs and compare them to determine if a shortEMA is moving away from the longEMA (using the threshold) and thus know when the market is in trend.


100% agreed, but I would add that Goomboo said that in January 2012, a time when the market was a lot smaller and different.

Good point, Gekko is going to need some sort of interface. Right now I'm thinking of creating a bookmarklet which you can run on clarkmoody that will connect to a local running Gekko and plot the advice (or the buys/sells) on the hourly graph. This would take the least amount of work and you can just use the clarkmoody graphs which are pretty good IMO.

So this would mean: user goes to clarkmoody, click a bookmark and see Gekko's advice / actions as shapes drawn on the graph.

I'm sorry but I still don't completely understand your suggestion for trade methods (not a lot of trade experience here):

1. Do you mean buy/sell on current market price on BTC-e? Because btc-e doesn't let you do an order at market price. This is how Gekko implements it right now when you configure buying/selling on btc-e:

• Watch trends on Mt. Gox, when new trend (up/down):
• Fetch the latest trades from btc-e (since prices are different from Mt. Gox), calculate btc-e price based on all trades in the last 20 seconds.
• Buy/sell using this calculated price.

2. Yep this needs to added to Gekko.

Thanks for the tips


Edit For everyone who wants to help out testing Gekko but having some installation problems: I've made a tutorial on how to install and run Gekko on windows.

Installing Gekko on Windows

Gekko Update:

Gekko now also supports automatic trading via BTC-e in addition to Mt. Gox! I've also updated the readme, to clarify how to run Gekko.

edit: I've just added BTC-e and I am my Gekko is using it, but it's alpha stage to say the least.

I will add more exchanges when I have some spare time.


Thanks for the link. I haven't heard that much about leveldb before but will look into it, sounds pretty cool. But isn't leveldb the same kind of dependency? Right now Gekko is running fine on my laptop (OSX), my pc (Windows 7) and my server (Ubuntu). It appears) that installing leveldb for windows, for example, requires building from source using Microsoft Visual Studio (not something everyone wanting to spin Gekko for a testrun is waiting for I think).


Awesome library! If I'm understanding it correctly the libary provides easy access to a lot of exchanges right? That's definitly something Gekko needs right now.

The biggest problem for me is that I have no experience in Java, so it's pretty hard for me to read the library. Gekko is just a small project at the moment, it's just a few hundred lines of code right now. Is it possible to connect to your framework using a communication layer?

I guess the easiest to get this method (which is described in more detail by Goomboo) into your library is to just port it to Java. If you don't have a lot of experience in javascript (or nodejs), you have the advantage to look at two additional projects who implement the same trading method:

• GoxTradingBotTobli
• GoxTradingBot

I've also just looked at how to get Gekko to talk to all exchanges using nodejs, and a lot of people have written modules for exchanges already:

• MtGox (api 2)
• BTC-e
• CampBx
• Snow, Ripple
• bitbox
• bitfloor

It should be pretty much just plugging them into Gekko, because all connect logic is already written.

This page states that the nonce is capped at 4294967294.

Thanks for your feedback

Great idea about storing historical data. Redis is great but it would mean a pretty big extra dependency for everyone who is not a developer and just wants to try/run gekko. Maybe if the difference in performance is not that big I could just go for flat file?

At the moment Gekko is just a simple command line tool (no frontend / GUI). Because it uses slow EMA, websockets to Mt. Gox would be overkill I think right? Eg. If it only needs a couple of trades every 60 minutes, not depending on anywhere near realtime.

I indeed want to abstract out all logic around buying and selling at the different exchanges so I can just plug in a new exchange. Right now the trading methods (there is only one yet ) are abstracted out so that the trading method can only advice and Gekko (who listens to the advice) holds the keys to the Mt. Gox API.

Thanks for the feedback. I will start working on getting Gekko on different exchanges.

I am thinking on calculating the EMAs based on trades done at Mt. Gox (with a fallback when Mt. Gox is down or something), but buying and selling on a configurable exchange (bitstamp, btc-e, Vircurex, campbx, etc). Because I think that at the the moment Mt. Gox sets the trading trend and all other exchanges follow it with a delay.

For those interested I've just released my own tradebot Gekko. It is also written in javascript but it runs on nodejs and not in the browser. I'm hoping we can take javascript based trade bots to the next level!

Gekko currently does EMA and I'm planning to add more methods in the future.

Gekko's thread