Chrome Browser extension: MtGox trading bot

[Lucko]

- Experimental feature: "The Crazy Ivan" - reverse the EMA-logic: The bot will sell when it's supposed to buy, and buy when it's supposed to sell!

I don't get it? Why? Can you elaborate? Thanks

Well, it's highly experimental and added mostly for fun (to put something in the new experimental section - I plan to add some more rules and settings there later that diverts from the original EMA-bot strategy).

From the description on the options-page:
"When the EMA-strategy is reversed, the bot will buy when it's supposed to sell, and sell when it should buy! There are absolutely NO theories supporting this strategy, however, when the market is ranging and you feel that the bot is always 180 degrees out of phase with the swings, it might work. Maybe..."

Thanks and thanks for the bot.

[1714600856]

1714600856

[1714600856]

1714600856

[1714600856]

1714600856

[1714600856]

1714600856

[Lucko]

If you are looking for experimental staff. What about hi frequency trading. When EMA is stable sell at EMA+1$ and buy at EMA-1$ or maybe not EMA but something else.

EDIT: No trigger for this should probably be Bolinger and stable EMA. Since Bolinger gets wider as soon as a price start moving again and EMA 10/21 must be the primery strategy again.

[jimbob007]

Would adding a setting in for the fee help at all? Such that it will only sell if it covers the fee? That way it wouldn't sell at a loss. Of course this only works if the price is rising. This might mess with the EMA portion of the bot, but I'm not completely sure if this would help at all. But I am loving the bot! Great work on it and thanks!

[spazcoin]

Tobbelino, I think the "Crazy Ivan" reference to Red October is hilarious!  Thanks for including that off-the-wall strategy as another thing to play with.  Now that I've read through this forum and GoomBoo's Journal, I'll start testing out your bot next week.

[WolfgangDS]

So I put in the API and the secret, set up the numbers.  If I just wait, it'll automatically set up a trade when the market reaches the specified conditions?  Noob question I know, but I just want to make sure I understand how this thing works.

[whydifficult]

So I put in the API and the secret, set up the numbers.  If I just wait, it'll automatically set up a trade when the market reaches the specified conditions?  Noob question I know, but I just want to make sure I understand how this thing works.

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.

[Lucko]

Tobbelino, I think the "Crazy Ivan" reference to Red October is hilarious!  Thanks for including that off-the-wall strategy as another thing to play with.  Now that I've read through this forum and GoomBoo's Journal, I'll start testing out your bot next week.

It is not only Red October. It is real submarine phrase. And it is call that way because Soviet navy mostly did them.

But if we are using submarine names my suggestion could be called Angles and Dangles...

[TobbeLino]

If you are looking for experimental staff. What about hi frequency trading. When EMA is stable sell at EMA+1$ and buy at EMA-1$ or maybe not EMA but something else.

EDIT: No trigger for this should probably be Bolinger and stable EMA. Since Bolinger gets wider as soon as a price start moving again and EMA 10/21 must be the primery strategy again.

The problem is that the trading fee at MtGox is so high, it's probably hard to make money from some kind of high freq trading. But maybe I'll look into it in the future. As for using Bollinger bands as an indicator, I'm not sure what data is required to calculate them. The bot currently only has access to a limited amount of data (the price of the first trade at each sample point). But maybe that's good for some kind of trending/ranging-indication...

[TobbeLino]

Would adding a setting in for the fee help at all? Such that it will only sell if it covers the fee? That way it wouldn't sell at a loss. Of course this only works if the price is rising. This might mess with the EMA portion of the bot, but I'm not completely sure if this would help at all. But I am loving the bot! Great work on it and thanks!

This has been proposed by many. But it's a risky strategy imo (based on the highly emotional assumption that the price WILL go up, and every down-trend is just very temporary). It may put you in a very bad position, and all it really does it that it "hides" your loss and make it worse. It will also cripple the EMA-strategy (the strategy is symmetric and should not prefer "buying" over "selling").

But maybe something will show up in the "experimental section" in the future
The bot can fetch the fee from the user info from MtGox, and the latest trades could also be fetched, so it's possible to use the last paid price in the strategy. But be warned: Your last paid price really has nothing to to with where the price will go from here!

[Lucko]

If you are looking for experimental staff. What about hi frequency trading. When EMA is stable sell at EMA+1$ and buy at EMA-1$ or maybe not EMA but something else.

EDIT: No trigger for this should probably be Bolinger and stable EMA. Since Bolinger gets wider as soon as a price start moving again and EMA 10/21 must be the primery strategy again.

The problem is that the trading fee at MtGox is so high, it's probably hard to make money from some kind of high freq trading. But maybe I'll look into it in the future. As for using Bollinger bands as an indicator, I'm not sure what data is required to calculate them. The bot currently only has access to a limited amount of data (the price of the first trade at each sample point). But maybe that's good for some kind of trending/ranging-indication...

http://www.metastock.com/customer/support/kb/faq.aspx?fc=120

http://en.wikipedia.org/wiki/Bollinger_Bands

[Grover]

The "Crazy Ivan" bot strategy is used in some FOREX bots, namely Fapturbo.

[wamatt]

The "Crazy Ivan" bot strategy is used in some FOREX bots, namely Fapturbo.

"Fapturbo" http://www.fapturbo.com/

good lord.... that's looks totally legit bro

[dwolfman]

I'm really hoping that, someday, bots like this which are easy to use would get an option to work with some exchange OTHER than MtGox. 

Like CampBX for example. 

[maverick1337]

Would adding a setting in for the fee help at all? Such that it will only sell if it covers the fee? That way it wouldn't sell at a loss. Of course this only works if the price is rising. This might mess with the EMA portion of the bot, but I'm not completely sure if this would help at all. But I am loving the bot! Great work on it and thanks!

This has been proposed by many. But it's a risky strategy imo (based on the highly emotional assumption that the price WILL go up, and every down-trend is just very temporary). It may put you in a very bad position, and all it really does it that it "hides" your loss and make it worse. It will also cripple the EMA-strategy (the strategy is symmetric and should not prefer "buying" over "selling").

But maybe something will show up in the "experimental section" in the future
The bot can fetch the fee from the user info from MtGox, and the latest trades could also be fetched, so it's possible to use the last paid price in the strategy. But be warned: Your last paid price really has nothing to to with where the price will go from here!

Just a quick note on your bot, not sure if this is intentional or not, but whether you check or uncheck "disabled on start," when you close out of the browser and boot it up the bot is always disabled. Basically it's always disabled on start whether you check or uncheck that box.

[TobbeLino]

Just a quick note on your bot, not sure if this is intentional or not, but whether you check or uncheck "disabled on start," when you close out of the browser and boot it up the bot is always disabled. Basically it's always disabled on start whether you check or uncheck that box.

You're right! There was some weirdness about this setting!
I've just uploaded a fixed version (v0.2.2.3)

[whydifficult]

I have a proposal for everyone working on (a fork of) the chrome browser extension. In an effort to improve the quality of our bots / projects and to speed up development I am creating a new way (sort of mini standard) at how our different projects could handle different exchanges efficiently:

(Born out of the requests by a lot of users to enable different exchanges, I figured this is about the right time since AFAIK out of all JS projects only Gekko supports exchanges besides Mt. Gox right now)

Gekko is going to use a 'portfolio manager' to handle all advices from the bot (like "buy now" or "sell now") and route them to the exchange the user prefers. I am proposing a way at which different exchanges could be implemented the same way by using this 'portfolio manager'. 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).

Note that the document is in early draft.

This would mean some rewriting of how the current chrome bot works BUT if you implement this system you will basically get all differences between all exchanges that I add to Gekko with minimal effort* (you would only need to write your own API wrapper because of the difference between Node and Chrome) but all the differences how each API returns stuff are abstracted out by me. Once Gekko's portfolio manager is there you could also port all that code into the chrome bot.

*Once the portfolio manager is added to Gekko this will be Mt. Gox, BTC-e and Bitstamp right away (the three biggest exchanges at this moment).

[San1ty]

What technical indicator are you actually using? It looks like the implementation of MACD but stopped halfway? Could you like me to a wiki explaining what your indicator is?

[TobbeLino]

What technical indicator are you actually using? It looks like the implementation of MACD but stopped halfway? Could you like me to a wiki explaining what your indicator is?

Goomboo's Journal: https://bitcointalk.org/index.php?topic=60501

[San1ty]

What technical indicator are you actually using? It looks like the implementation of MACD but stopped halfway? Could you like me to a wiki explaining what your indicator is?

Goomboo's Journal: https://bitcointalk.org/index.php?topic=60501

Yeah that looks like someone that wanted to use MACD and forgot a part.

http://en.wikipedia.org/wiki/MACD

I would refrain from using that method as I have doubts that it is more efficient.



Also look at how much more responsive the MACD crossover is vs regular EMA crossover.

[TobbeLino]

Yeah that looks like someone that wanted to use MACD and forgot a part.

http://en.wikipedia.org/wiki/MACD

I would refrain from using that method as I have doubts that it is more efficient.

Also look at how much more responsive the MACD crossover is vs regular EMA crossover.

This bot was never attempting to implement a MACD indicator, but a simpler one using only two EMAs of different lengths (as I understand it, this is like using only the MACD line's zero crossovers as indicator).

You can read the whole analysis and discussion about this in the tread I previously referred to (Goomboo's Journal: https://bitcointalk.org/index.php?topic=60501).
I'm no expert in trading strategies and technical analysis myself, and I do not have a quantitative experience of all the many strategies that exists, but to me Goomboo seems to be a quite experienced trader, and I'm quite sure he didn't try to use MACD and simply "forgot" the signal line. His philosophy is instead: The simpler strategies tend to beat the more complex ones in the long run. Again, I can not judge whether this is a good or bad philosophy, but I like it

The signal–line crossovers in MACD may very well respond faster, but it will also generate more false signals.

But of course, it would be very nice to also have MACD indicators implemented in the bot so that the users can choose.
Maybe some day