fetch_ohlcv returning strange results

*First of all i understand this is under development

i’m a little bit confused about how the fetch_ohlcv returns data. I have an algo that i drip feed bars every day (or hour etc). For example, every day i want to give a OHLCV daily bar to the algo to process, the algo stores a list of all previous daily candles.

So to clarify, i would have a timer that runs every day, pulls the last full closed daily bar up to the millisecond and feeds it to my algo.

I don’t want a bar in progress. I want the close of the bar to be the current price

• OS: Windows 10
• Programming Language: Python 3.5
• CCXT version: 1.10.413
• Exchange: bitmex
• Method: fetch_ohlcv

import ccxt

exchange = ccxt.bitmex()
exchange_markets = exchange.load_markets()
ticker = 'BTC/USD'

print('current exchange time')
print(exchange.iso8601(exchange.milliseconds()))
print()

#the latest close of the ticker - in the following lines i'm expecting the close price of the bar to be this value
print('latest close: ' + str(exchange.fetchTicker(symbol=ticker)['close']))
#this seems to get really old data, 2016 - I would expect the most recent 100 bars by default?
bars = exchange.fetch_ohlcv(ticker,'1d')
print('bars in array: ' + str(len(bars)))
print(bars[-1])
print(exchange.iso8601(bars[-1][0])) 
print()


print('latest close: ' + str(exchange.fetchTicker(symbol=ticker)['close']))
#does this get the latet 1day closed bar?
bars = exchange.fetch_ohlcv(ticker,'1d', since=exchange.milliseconds())
print('bars in array: ' + str(len(bars)))
print('# the -open- is higher than the -high- ?')
print(bars[-1])
print(exchange.iso8601(bars[-1][0])) 
print()


print('latest close: ' + str(exchange.fetchTicker(symbol=ticker)['close']))
#I would expect this to return one full day of ticks in the form of 1 day bar
bars = exchange.fetch_ohlcv(ticker,'1d', since=exchange.milliseconds() - (24 * 60 * 60 * 1000)) # 1 day
print('#returns two bars, but i only put since 1 day of milliseconds')
print('bars in array: ' + str(len(bars)))
print('second second bar in array')
print('is this a bar in progress or a closed bar?')
print(bars[1])
print('# the date is the 2017-12-26 - 1 day in the future, how is this possible')
print(exchange.iso8601(bars[1][0]))
print('first bar in array')
print(bars[0])
print(exchange.iso8601(bars[0][0]))

This is an example output

current exchange time
2017-12-25T04:54:04.260Z

latest close: 14161.0
bars in array: 100
[1451779200000, 435.99, 438, 429.25, 435, 729519]
2016-01-03T00:00:00.000Z

latest close: 14123.5
bars in array: 1
\# the -open- is higher than the -high- ?
[1514246400000, 14227.5, 14185.5, 14123.5, 14136.5, 368345]
2017-12-26T00:00:00.000Z

latest close: 14118.5
\#returns two bars, but i only put since 1 day of milliseconds
bars in array: 2
second second bar in array
is this a bar in progress or a closed bar?
[1514246400000, 14227.5, 14387, 13232, 14108.5, 225518513]
\# the date is the 2017-12-26 - 1 day in the future, how is this possible
2017-12-26T00:00:00.000Z
first bar in array
[1514160000000, 14719.5, 14788.5, 12521.5, 14227.5, 1662515600]
2017-12-26T00:00:00.000Z

[kroitor]

I tried the last example and it is still fetching the last partial bar, do i need to get the latest build?

What about the first one?

the last bar has a timestamp [1514246400000, 14227.5, 14480, 13232, 14040.5, 315800080] 2017-12-26T00:00:00.000Z

>>> exchange = ccxt.bitmex()
>>> since = exchange.milliseconds() - 86400 * 1000 * 2
>>> limit = 2
>>> result = bitmex.fetch_ohlcv ('BTC/USD', '1d', since, limit)
>>> len(result)
2
>>> [exchange.iso8601(ohlcv[0]) for ohlcv in result]
['2017-12-24T00:00:00.000Z', '2017-12-25T00:00:00.000Z']

>>> limit = None
>>> result = bitmex.fetch_ohlcv ('BTC/USD', '1d', since, limit)
>>> len(result)
3
>>> [exchange.iso8601(ohlcv[0]) for ohlcv in result]
['2017-12-24T00:00:00.000Z', '2017-12-25T00:00:00.000Z', '2017-12-26T00:00:00.000Z']

↑ are you sure you’re doing it right? From the above it should be obvious.

[kroitor]

@DaneEdwards thanks! We really appreciate it! I’m sure many people will find it useful! You might also want to take a look at our examples/py for built-in rate-limiting, async generators and other interesting use-cases.