OHLC public method bug? https://www.kraken.com/help/api#get-ohlc-datahttps://api.kraken.com/0/public/OHLC?pair=XBTCZEUR&interval=60&since=1381449600 api asked for data since=1381449600returns data starting from 1413583200the param since is ignored. Also... could you consider to move your api docs to github? it will allows better integration with all apps which are accessing your API. It is also well versioned. Some markets already did that. We'll consider the github idea - thanks for the suggestion. The other thing isn't a bug. OHLC data does not go back beyond a certain number of frames and depending on the interval it won't go back past 24 hours. The "since" value here is mainly used to received incremental updates. Choosing a longer interval will yield OHLC data going further back. You can build your own OHLC from the trades data. Start with "since" of 0 to get trade data from the beginning. https://api.kraken.com/0/public/Trades?pair=XXBTZEUR&since=0Then from the returned data, grab the "last" time ID (i.e. ... ,"last":"1383839436659595694"}}) and use that ID for the "since" value in the new poll: https://api.kraken.com/0/public/Trades?pair=XXBTZEUR&since=1383839436659595694and so on until you've fetched all the trade data. If that doesn't clear things up, let me know. Dargo, I'm already familiar with processing of raw tick trades to OHLC. The since param in trades is valuable as it helps to enforce data integrity, I use it every few minutes  I've asked for the method OHLC because it would allow to DIRECT usage by js charting libraries. So regarding the OHLC method, as I understand, the since param cannot be used to get deep historical OHLC but only quite recent OHLC data? It's a bit limitation comparing to since parameter in trades method, which obviously overloads the market API much heavier (due to the volume of rows) than OHLC data would overload. |
|
|
|
OHLC public method bug? https://www.kraken.com/help/api#get-ohlc-datahttps://api.kraken.com/0/public/OHLC?pair=XBTCZEUR&interval=60&since=1381449600 api asked for data since=1381449600returns data starting from 1413583200the param since is ignored. Also... could you consider to move your api docs to github? it will allows better integration with all apps which are accessing your API. It is also well versioned. Some markets already did that. |
|
|
|
|
"instant order" from definition is order which doesn't care about price, so it will try to buy/sell at any price. It will eat any orders till the order book will be empty, and when it happens it should just place the order without filling it. Not a best solution, but it is the way how instant order works.
|
|
|
|
@ Dargo:
another question:
is it possible to have a second account at kraken? I want to test a trading bot, but of course not with all my bitcoins/money I have at kraken. I could withdraw bitcoins and money, but on the other hand I want to trade manually with all my bitcoins/money during testing the bot. So is it possible to have a second account for bot-testing? Or do you have another suggestion?
You will have to use a different email address, but yes you can create a second account. Otherwise just use the same name, address, etc. so it's clear that it's a second account of yours. very nice information. How the verification process goes in that way? Can I apply for the same Tier as I already have on my main email address and it will be verified using previously provided documents? or I need to pass the full process for each Tier again? |
|
|
|
Count increment for order add/cancel calls: +0 is it correct that I can sell my 1 bitcoin by placing one hundred orders, each for 0.01 BTC, all at the same time, and I will not be blocked? this seems to be potential vulnerability for ddos-like actions on you api interface. |
|
|
|
What you are speaking of isn't a bug, its because we have a very low volume compared to other exchanges such as BitStamp.
About the fee, you can simply "buy" 0.606 BTC.
I believe you should rethink such design approach if you want to catch some real market share. |
|
|
|
bernard75, you are perfectly correct. I'm already verified but I don't even think to trade / develop API interface as long as the fee structure is not 100% clearly stated. It is not even possible to estimate the trading capabilities without that information.
Now. When you are on ANXPRO, you can see the fees for each individual order when you select the different options. And this is exactly the problem. You can see the fee for individual order but you can't see the fee structure, that means you are not able to estimate your future trades fees, so you are not able to estimate profits/losses. |
|
|
|
|
bernard75, you are perfectly correct. I'm already verified but I don't even think to trade / develop API interface as long as the fee structure is not 100% clearly stated. It is not even possible to estimate the trading capabilities without that information.
|
|
|
|
New API Rate: The API rate limit has been increased for account Tiers 1-3 Edit: Tier 3.
Old rate: 1 call every 5 seconds with a burst of 10.
New rate: 1 call every 3 seconds with a burst of 15.
The details, in terms of the count system for the rate limit, are as follows (changes in bold):
Max count before lockout: 15 (lockout is for 15 minutes)
Count increment for order add/cancel calls: +0
Count increment for ledger/trade history calls: +2
Count increment for all other calls: +1
Count reduced by 1 every: 3 seconds
thanks for update, I assume that things marked as strong are the one updated. I would suggest also to write a simple conclusion, what is safe (not considering burst etc.) interval for trades and for other calls. trades/ledger calls: 6s (reduced from 10s) add/cancel calls: 0s (didn't remember how it was before) other calls: 3s (reduced from 5s) correct me if I'm wrong. I believe this is the way how most of (at least not only kraken specific) trading engines will process interval parameters. |
|
|
|
ANXPRO charges trading fees as low as 0.05%, and we don't charge any transaction fees to trade bitcoins. Sure, we will update the OP. No, but when you are on ANXPRO, you can see the fees before every trade you execute, depending on the type of the order. While placing market order or fill or kill order the fee is written as 0.60%, when placing limit order the fee is written as 0.30%. It is far higher then mentioned 0.05%. I suggest you to build a page "fee structure" and link it in the platform site footer, as the footer is the place where everybody would look for such information in the first place. Currently fee structure is very confusing. |
|
|
|
|
thanks for explanation, so it means that current "safe" trades history method interval is 10s.
|
|
|
|
could you update the first post. it still states the site is in "beta" stage. also the fees seems to be quite confusing. I see different fees in the anxpro.com/#trade window depending on the order type (market, limit, fill or kill), They varies from 0.30% to 0.60%. Yet in the anxbtc FAQ I see the updated (New fees in effect on June 1) information regarding fee which states: ANX customers are no longer charged transaction fees when using the ANXBTC platform to buy and sell bitcoins. Could you clarify? or link comprehensive document related to fees? Regards charges trading fees as low as 0.05%, and we don't charge any transaction fees to trade bitcoins. Sure, we will update the OP. so the "0.30%/0.60% fee" also had to be updated in the trading platform. thanks for the information. |
|
|
|
could you update the first post. it still states the site is in "beta" stage. also the fees seems to be quite confusing. I see different fees in the anxpro.com/#trade window depending on the order type (market, limit, fill or kill), They varies from 0.30% to 0.60%. Yet in the anxbtc FAQ I see the updated (New fees in effect on June 1) information regarding fee which states: ANX customers are no longer charged transaction fees when using the ANXBTC platform to buy and sell bitcoins. Could you clarify? or link comprehensive document related to fees? Regards |
|
|
|
I pulled this from the other thread. Hi Dargo,
Can you clarify some points regarding the API?
1) What is the request/throttle limit before getting banned, if any? (e.g. 1 request per second)
It's a bit complicated since there are a variety of factors that can change things, but the basic limits are as follows. The limit for sustained requests is 1 request every 5 seconds. The limit for sustained orders is about 1 per minute. The sustained limits can be exceeded in short bursts. Keep in mind that this is subject to change as we see how people use the API. Please increase the rate limits. One request per 5 seconds is pretty low but I can work around it. One request per second would be more resonable. However one order per minute is very low. Overall I place less than 60 orders per hour, but sometimes I may need to place 4 or 5 orders in a minute. I run into the rate limit too often. +1 I think our API limit should be higher too. It's an ongoing project of mine to try to convince others on the team about this and get the limits raised. I've had partial success and a raise of the limit should be coming. I'll check on the progress of that and let everybody know. Dargo, Is there any other rate limit for trades ("Get recent trades") method? I already set 6 seconds for it but still get "EAPI:Rate limit exceeded" from time to time. If the api rate is method specific could we get a full table which describe each method, burst, etc.? |
|
|
|
what is the difference between anxbtc.com and anxpro.com?
Basically, you can trade cryptos for fiat on ANXBTC, and cryptos on ANXPRO. I see I can trade fiat on anxpro also. could you elaborate more deeply... does anxbtc and anxpro have common: 1. market depth? 2. trades history? 3. user accounts? 4. user verification process? 5. fees? 6. api interface? 7. supported currency pairs? after very short look it seems that anxpro is just another(better) implementation for you market. |
|
|
|
Not very useful.
Hello, Let me explain more about the fact that volume in different currency pair is similar . Other exchanges have separate order books and they have one order book for BTCEUR, another for BTCUSD. Trades, pricing, depth is all independent. However ANX has a sophisticated engine that streams in FX rates in real-time, and blends all FIAT currencies into a single order book. Orders placed in BTCEUR can match against orders placed in BTCUSD. It's based on a very fair algorithm that converts each existing order into the equivalent price for any incoming order. USD orders are settled in USD, and EUR in EUR. Some exchanges have tried to do multi-currency by first converting orders into a common currency, which is terrible for the customer as they suffer FX slippage. In the case of ANX a USD order rests in USD, and a EUR in EUR. If they match, both parties receive their native currency. So when you call "/api/2/{currency_pair}/money/trade/fetch" for different currency pairs, you are actually querying against a single order book. The data is completely transparent - the "primary" flat tells you what is going on. If a EUR trade matches against a USD trade, the USD and EUR will both be reported in the history call with primary="true". Most charting and analytics sites continue to think of ANX as just the same as the other sites, and just choose one of our currencies to report, which is unfortunate as it doesn't accurately reflect our trades and blended order book. Yes, I already know that. The problem is, like you said, the representation. It's not useful and it doesn't look professional. It looks fake. How is this documented in your API documentation? What are your plans to improve the situation? +1 Currently when looking at bitcoincharts.com market list, the ANX looks like a fake volume market. You should do something with this, not just wait for community to adapt to your interface. |
|
|
|
I pulled this from the other thread. Hi Dargo,
Can you clarify some points regarding the API?
1) What is the request/throttle limit before getting banned, if any? (e.g. 1 request per second)
It's a bit complicated since there are a variety of factors that can change things, but the basic limits are as follows. The limit for sustained requests is 1 request every 5 seconds. The limit for sustained orders is about 1 per minute. The sustained limits can be exceeded in short bursts. Keep in mind that this is subject to change as we see how people use the API. Please increase the rate limits. One request per 5 seconds is pretty low but I can work around it. One request per second would be more resonable. However one order per minute is very low. Overall I place less than 60 orders per hour, but sometimes I may need to place 4 or 5 orders in a minute. I run into the rate limit too often. +1 |
|
|
|
Kraken-tools for the privileged  in case of any concerns... R and Rbitcoin are open source. using it doesn't require any magical skill: 1. install R 2. run command: install.packages("Rbitcoin") 3. run script  |
|
|
|
Hey Dargo Could you please talk to your developers to increase the number of calls from the API or find another solution? I'm using the website " www.coinreporting.com". It's a very useful website to have an overview of all the trades I made. Unfortunately there is at the moment only one way to import the Kraken Trades. The Import works with the API. But the problem is that it only allows 50 calls, so I'm only able to import my last 50 trades.. but I need something like x thousand for my trades. =/ I don't know much about API's, but I think you made the 50 limit for a reason and I think you won't rase this to 10 thousand or something like this. So maybe there is another solution. How about a csv export or something simular for the trade history? Hi Serpens, I've been pushing for CSV export for a long time now. Our developers are somewhat snowed under with projects at the moment though, so it may not happen until we hire more developers (which we're in the process of doing right now). As for the API, you can call all your trade history, but you probably can't do it directly through coinreporting.com. You need a script to do it because it'll require many calls to get all the data for an active trader. I believe you can only get historical data for 50 trades at a time, so you have to make many coordinated calls to get all your data. There's a sample script for doing this here: https://github.com/matuszed/KrakenToolsKeep in mind that you may have to adjust the 'max' and 'window' parameters (set at 10/10 seconds in the script) if you hit the rate limit. I realize this won't help those who don't have at least basic familiarity with running script, but it's the best solution for now until we get CSV export. Alternatively, you can download whole private trades history to csv using R and Rbitcoin package, in a little bit more elegant script, performed in HPC, with appropriate data types conversion library(Rbitcoin)
options(digits.secs=4) # kraken time precision
req <- NULL
resDT <- data.table() # storage for results
repeat{
Rbitcoin:::antiddos("kraken", antispam_interval=11, verbose=1)
resList <- market.api.query(market = 'kraken', url = 'https://api.kraken.com/0/private/TradesHistory',
req = req,
key = '',
secret = '')
if(length(resList$error)>0) stop(paste("kraken market api error:",resList$error))
if((length(resList$result$trades)>0 & is.null(req)) | (length(resList$result$trades)>1 & !is.null(req))){
if(!is.null(req)) resList$result$trades[[1]] <- NULL #filter out row previously loaded
req <- list(end = names(resList$result$trades[length(resList$result$trades)]))
resDT <- rbindlist(list(
resDT,
data.table(txid = names(resList$result$trades), rbindlist(resList$result$trades)[,c("price","cost","fee","vol","margin") := lapply(.SD, as.numeric),.SDcols=c("price","cost","fee","vol","margin")][,`:=`(time=as.POSIXct(time,origin="1970-01-01",tz="UTC"))])
))
cat(paste0("\n",as.character(Sys.time()),": added ",length(resList$result$trades)," rows"))
} else {
cat(paste0("\n",as.character(Sys.time()),": completed"))
break # empty batch
}
# next batch
}
# save to csv (csv2 for EU, csv for US)
write.csv2(resDT,'private_trades_history.csv',row.names=FALSE)
# aggregate to orders using R data.table and then save
dt <- resDT[,list(time_start = min(time), time_end = max(time), tx_count = length(unique(txid)), price = sum(price*vol)/sum(vol), cost = sum(cost), fee = sum(fee), vol = sum(vol)), by=c("ordertxid","pair","type")]
write.csv2(dt,'private_filled_orders_history.csv',row.names=FALSE) |
|
|
|
|
Show Posts
