ATP [Advanced Trading Platform]

[MobGod]

How do we get this to work for alt-coins on
https://www.cryptsy.com/

[1714038814]

1714038814

[1714038814]

1714038814

[1714038814]

1714038814

[daybyter]

API is here:

https://www.cryptsy.com/pages/api

Maybe I'm the only one trying to implement it so far. At least the #cryptsy channel at freenode is very quiet so far...

[Nova!]

API is here:

https://www.cryptsy.com/pages/api

Maybe I'm the only one trying to implement it so far. At least the #cryptsy channel at freenode is very quiet so far...

Looks a lot like BTC-e API.  I can probably modify the BTC-e exchange module to communicate with cryptsy if you'd like to see it as a module.

[Nova!]

Crap wrong product, sorry.  Consider it added to my TODO list on the ATPx/ Expressvest product.

[MobGod]

Ok so it doesn't work correct? Your going to try and get it working?

[Askit2]

Well I know it should work out eventually but the bot has purchased me into the wrong side of the trade at MTGoX pretty badly. Compounding my frustration is the fact that I can't seem to log in to gox. So even as I watch the ticker climb 20$ in 2 days I can't manage to get on to stop the losses manually.

I know this bot can make money in an up market. Well sort of as there will be some periods that are not just up. Ironically when it looks like a rocket to the stars not as much. The bot works and I really don't want to complain about what it does well. It buys a ton of Fiat. Ironically that looks great heading down. If only I could get it to blow out of the Fiat closer to the bottom. But if it sells any the weight drops off. This makes smaller sales. Then the market heads up. I guess a bigger problem isn't that it heads down, but that I don't get enough trades in at a 15% gain to make any difference when the market heads up. Manually I would expect at 10%+ gains it would be prudent to take a profit on half of what is in fiat. I have no idea how to give the ATP trading logic that would do that. well aside from risk 3 and max loss 1.0. I can't leave the bot like that as it will buy the same way compounding the issue.

I guess another option would be a fork that had some logic to detect a Fiat gain % and start increasing the weight more as the gain was higher. Then again that would be against the logic of the ATP currently in that it tries not to make multiple trades into a down market or at least not increasingly larger trades. I get not trading excessively into an up market. I also understand that in the ~24 hours Gox was down the bot had ample opportunity to trade enough had the ticker stayed up more then a couple hours of the day for me.

Is there a way to force a sell of all of a position? I can use the API but keep being cloud flaired by gox. Oddly some part of Gox is working correctly. Just not a part I can force a trade with.

Not sure if it would work but if I remove all logic to buy in theory it should only get true for selling. Maybe I can force a sale but it still wouldn't be able to be left on. At this point I want to stop the bleeding. Maybe I will try it. At worst I get another crash. This one not Gox related.

EDIT: It doesn't fatally crash on no ask logic. It does print a ton of errors. I expected that so far euro is fixed waiting for USD to pull back a little.

EDIT2: Conflicting logic serves the same purpose without errors. Just telling the bot to wait for an impossible situation is enough to stop trading.

[jago25_98]

Let's say we have the message:

"Arbitrage Engine has detected an after fee profit opportunity of 123.15897301% on currency pair RUR/EUR on BTC-e"

How would one take best advantage of that?

The other messages are interesting in that they conflict

Code:

2013-07-12 12:27:51,200 : MtGox Advance/Decline Spread algorithm has determined that the USD market is trending down.
2013-07-12 12:27:51,201 : MtGox VWAP Cross algorithm has determined that the USD market is trending flat.
2013-07-12 12:27:51,208 : MtGox Trend following trading agent has decided no USD action will be taken at this time.

ah, a nonce in the works!

Code:

2013-07-12 12:32:12,137 : Arbitrage could not trade with a balance of RUR 0 on BTC-e
2013-07-12 12:32:12,257 : ERROR: Caught unexpected exception, shutting down BTC-e trend following trading agent now!. Details are listed below.
com.xeiam.xchange.ExchangeException: BTCE returned an error: invalid nonce parameter; 66521328 66521328
	at com.xeiam.xchange.btce.service.polling.BTCEBasePollingService.checkResult(BTCEBasePollingService.java:71)
	at com.xeiam.xchange.btce.service.polling.BTCEPollingAccountService.getAccountInfo(BTCEPollingAccountService.java:53)
	at org.aido.atp.AccountManager.refreshAccounts(AccountManager.java:121)
	at org.aido.atp.AccountManager.getBalance(AccountManager.java:105)
	at org.aido.atp.TrendTradingAgent.evalBid(TrendTradingAgent.java:494)
	at org.aido.atp.TrendTradingAgent.run(TrendTradingAgent.java:346)
	at java.lang.Thread.run(Thread.java:722)

[Aido]

Let's say we have the message:

"Arbitrage Engine has detected an after fee profit opportunity of 123.15897301% on currency pair RUR/EUR on BTC-e"

How would one take best advantage of that?

Arbitrage code in develop branch is very much experimental and the message is not true. As part of testing I hardcoded a very high profit percentage to make sure a trade would take place regardless of the true value.

[jago25_98]

Let's say we have the message:

"Arbitrage Engine has detected an after fee profit opportunity of 123.15897301% on currency pair RUR/EUR on BTC-e"

How would one take best advantage of that?

Arbitrage code in develop branch is very much experimental and the message is not true. As part of testing I hardcoded a very high profit percentage to make sure a trade would take place regardless of the true value.

No worries, I'm not trading live (actually opened some empty new accounts just for testing). I do love it though, gives me some satisfaction seeing saying something even if it does give a worrying counter to another trade strategy elsewhere...

Couldn't see that hard coding commented in:
https://github.com/aido/AidoATP/blob/master/src/main/java/org/aido/atp/ArbitrageEngine.java
but then I don't do java and I could have missed it.

My question though, and this was aimed more at the traders here rather than programmers was more about how to trade manually to take advantage of an arbitrage situation in general and that was meant as an example primarily. I see a scenario earlier in this thread but the person is using the example of 2 people who are friends and trust each other.

One way I can see if we have a signal for RUR/USD would be to go to a forex broker and click buy on that. But really for arb I was thinking more about Bitcoin and then it gets complicated with transaction delays and costs in for example sending USD from Bitstamp to USD.

Of course, http://bitcoin-analytics.com/ does alot of it for us... but for a fee. There are also other arb bots... but I don't see how they work without sending funds from one exchange to the other

[Askit2]

You can always manually send them after the arbitrage.

[jago25_98]

You can always manually send them after the arbitrage.

Yes of course, that's what I'd like to envisage... but how would you best do that t for a EUR/RUR buy signal? - and also bearing in mind getting funds to the various exchanges is often a problem too (btc-e?)

[Aido]

Couldn't see that hard coding commented in:
https://github.com/aido/AidoATP/blob/master/src/main/java/org/aido/atp/ArbitrageEngine.java

Your link is for the master branch on github. The message about "Arbitrage Engine has detected an after fee profit opportunity of 123.15897301% on currency pair RUR/EUR on BTC-e" is from the develop branch

Code:

// Temporary value for testing purposes until I figure out how the hell to calculate profit
double profit = 100;

but then I don't do java

Then that would make two of us 

but I don't see how they work without sending funds from one exchange to the other

For the record, AidoATP (and it's parent IsisATP) uses intra-exchange arbitrage not inter-exchange. So no funds sent between exchanges.

However, as rough as the arbitrage code is, it should still signal an opportunity that may be profitable if acted upon manually.

[Askit2]

Is there a simple way to add 1 option to the ATP so that maybe the next version allows high risk without the exponential reduction system?
My thought would be add exponential reduction as an option that users could switch at runtime. This would add a question on first run and an option to the config or registry.

It isn't that the system is bad. It usually works great. Not much sold until the market changes etc. I ask because ATP seems to like trading me solidly into fiat as much as it can. I have increased my maximum local to far above what max BTC is set at now. It had been right on for it. I think, but can't really say for sure, that the weighting for my site or API or Gox issues will not get down as low below average as it gets above average. I am trying to say that although it buys (with much more logic required) and sells (simple logic) the Sales of bitcoin stack 2 deep at times but the buys for bitcoin only ever get to 1 in a row. This leads me to think that since the market fell below the last buy of bitcoins but the weight was low the exponential weighting system has limited me to 1 order in a row for sales.

I hope increasing the maximum local will help correct this. I think it will as the weight should be times the capped local amount. It isn't like the logic I have is bad. the logic itself works. Well for 2 of 3 types of markets. It works going down. It works going flat. It doesn't work so great going up. I can't find logic that would dump my fiat when the market heads up. I know its called a stop loss. What I am trying to setup is similar in idea except that you don't keep as much fiat. This limits my gains on the way down. But should radically curb my losses on the way up.

As an example. Using my old maximum local of 8 and max BTC of .1 I had ~8$ and .28BTC when the market headed down from 100+. I was there because I had decided to net out half my fiat earlier that day. The market went down to 68 ish. ATP didn't sell much apparently because after around 24 hours of being below 72 and trading flatish I had ~18$ and less bitcoins. I was still above a 13% gain though and really wanted to sell to cash in. MTGoX was being a douche so I kept seeing cloudflare and couldn't login to trade manually. Now the market over the next 2 days went up to 90 ish. I was sitting at a loss of 10% when it hit 90.

I know had the bot sold no BTC when the market was down I would have still had a gain. I also know that when it hit bottom and I was up ~15% I would have loved to sell manually. I waited for the bot being sure that if I can't have it manage things pretty well on its own I really haven't gotten things setup correctly. When the market was heading up I did try to correct the losses I was sure I would get. Cloudflare is my nemesis. Well as Jim Cramer says never turn a gain into a loss.

I have made two changes. I am now averaging over a much larger time period and I have increased maximum local. Hopefully I can see less fiat and more smart trades.

[Aido]

I suppose I could add an "extreme risk" option that would disable the the exponential reduction in trade amount as a trend continues i.e. trade all the way up or down a trend until you have either all BTC or all fiat (depending on trend direction). It should be fairly easy to code.

However, the sunshine outside is not conducive to coding and I cannot commit to when this will be done. If I have a bit of time later this evening I may have a look at it.

[Askit2]

I think maybe my example was bad.
Extreme risk sounds like maximum risk.

What I was hoping for a separate setting that one could use with an already tuned risk profile. Like Conservative or high risk. But it would take more work now that I think about it. You would have to read in a value, run through a logical if, then have a variable in the final equation that would allow one if result to have exponential weighting and one to always return I would guess 0 ish.

I can keep using maximum risk as it will do something a little similar. I can't leave either on all the time likely but I was thinking that just disabling the exponential reduction for a working otherwise setup would give me a little more control on what I want the bot to do.

Enjoy the weather. If I get my weighting right this should be fairly un needed.

[Aido]

what I'm thinking is having 4 risk levels.

• 1 = Low risk = current conservative risk
• 2 = Medium risk = current high risk
• 3 = High risk = current maximum risk
• 4 = Maximum risk = current maximum risk with exponential reduction in trade amount disabled

This way, no change to existing users configuration as no change to risk levels 1 to 3. If someone wants to experiment with the additional risk level just edit their config and change risk level to 4.

[Aido]

New commit in develop branch (see previous post)

commit 1f3d28943ac40659de61dbb789930415a1a63977

    Added another level of risk

[MobGod]

any news on adding cryptsy ?

[Askit2]

Aido
I didn't see anything in the changes to explain what I am seeing on the console.

On BTC selling calculations everything looks great.
Blah blah blah weight,
Blah amount less then minimum
BLah blah blah didn't trade

On buying bitcoins it looks wrong
Bah blah blah weight

I look in the log and I see a value of USD I have. It just never seems to calculate an amount of bitcoins to buy.

Switched back from development version with 4 options and Bam a buy on less weight then the other version. Odd

EDIT: am on strategy 2. Not sure if that helps or hinders but that is what I was on.

EDIT2: Aido I think I was wrong before about my logic being fine. I made an assumption that the max local wasn't the most ATP would calculate based on but a limit to how much it could use at once. It appears that I was a bit off in that. Either that or doubling the max had an unexpected side effect of allowing multiple buys in a row.

[Aido]

any news on adding cryptsy ?

AidoATP currently relies on the XChange library for all communications with the different exchanges. If cryptsy isn't supported by XChange then it won't be supported by AidoATP.

Cryptsy currently isn't supported by XChange