def data_api(pair, start_date, end_date, time_resolution):
'''
Summary:
----------
Retrieves crypto data via Binance API
Params:
----------
pair : str
crypto pair to retrieve (i.e. BTC/USDT)
start_date : str
starting date in dataset (YYYY-MM-DD)
end_date : str
ending date in dataset (YYYY-MM-DD)
time_resolution : str
resolution in data set (1h, 1d, 1w)
Outputs:
----------
crypto : dataframe
financial data in OHLCV form
'''
crypto = get_crypto_data(pair,
start_date,
end_date,
time_resolution=time_resolution)
return crypto
def test_get_crypto_data():
# test that multiple exchanges work
from fastquant import CRYPTO_EXCHANGES
exchange_pairs = {
"binance": "BTC/USDT",
"coinbasepro": "BTC/USD",
"bithumb": "XRP/KRW",
"kraken": "BTC/USD",
"kucoin": "BTC/USDT",
"bitstamp": "BTC/USD",
}
for ex in CRYPTO_EXCHANGES:
crypto_df = get_crypto_data(
exchange_pairs[ex], DATE_START, DATE_END, exchange=ex
)
assert isinstance(crypto_df, pd.DataFrame)
return values.ewm(alpha=1/n, adjust=False).mean()
def atr(df, n=50):
data = df.copy()
high = data['high']
low = data['low']
close = data['close']
data['tr0'] = abs(high - low)
data['tr1'] = abs(high - close.shift())
data['tr2'] = abs(low - close.shift())
tr = data[['tr0', 'tr1', 'tr2']].max(axis=1)
atr = wwma(tr, n)
return atr
#max # of records is 500
outter_df = get_crypto_data("BTC/USDT", "2019-01-01", "2021-02-21")
#EVWMA
Short_EVWMA = pd.DataFrame(TA.EVWMA(outter_df,12))
Long_EVWMA = pd.DataFrame(TA.EVWMA(outter_df,26))
Short_EVWMA.columns = ['EVWMA_12']
Long_EVWMA.columns = ['EVWMA_26']
ATR = pd.DataFrame(atr(outter_df))
ATR.columns = ['ATR']
#p 209 of ttr doc
MACD_EVWMA = pd.DataFrame(Short_EVWMA['EVWMA_12'] - Long_EVWMA['EVWMA_26'])
MACD_EVWMA.columns = ['MACD-line']
Signal_EVWMA = pd.DataFrame(ta.ema(MACD_EVWMA["MACD-line"], length=9))
def test_get_crypto_data():
crypto_df = get_crypto_data(CRYPTO_SYMBOL, DATE_START, DATE_END)
assert isinstance(crypto_df, pd.DataFrame)
def crpyto_comp_plotting(dfs: List[pd.DataFrame],
title: str,
e_vars: EbayVariables,
start_date: datetime,
end_date: datetime,
roll: int = 0,
min_msrp: int = 100) -> None:
"""
Parameters
----------
dfs :
title :
e_vars :
roll :
min_msrp :
Returns
-------
"""
dfs = deepcopy(dfs)
min_date = datetime.now()
max_date = datetime.now() - timedelta(365)
for df in dfs:
df = df[df['Sold Date'] >= start_date]
df = df[df['Sold Date'] <= end_date]
df_min = df['Sold Date'].min()
df_max = df['Sold Date'].max()
if df_min < min_date:
min_date = df_min
if df_max > max_date:
max_date = df_max
min_date = str(min_date - timedelta(1)).split(' ')[0]
max_date = str(max_date).split(' ')[0]
# Etherium Pricing
# print(min_date, max_date)
eth_crypto_full = get_crypto_data("ETH/USDT", min_date, max_date)
eth_crypto_full = eth_crypto_full.close
eth_prices = eth_crypto_full.div(eth_crypto_full[0])
# eth_prices = eth_prices.mul(100)
# Bitcoin Pricing
btc_crypto = get_crypto_data("BTC/USDT", min_date, max_date)
btc_prices = btc_crypto.close
btc_prices = btc_prices.div(btc_prices[0])
btc_prices = btc_prices.mul(100)
colors = [
'#000000', '#7f0000', '#808000', '#008080', '#000080', '#ff8c00',
'#2f4f4f', '#00ff00', '#0000ff', '#ff00ff', '#6495ed', '#ff1493',
'#98fb98', '#ffdab9'
]
plt.figure() # In this example, all the plots will be in one figure.
plt.ylabel("GPU: Price/hashrate - ETH: Price")
plt.xlabel("Sale Date")
plt.tick_params(axis='y')
plt.tick_params(axis='x', rotation=30)
for i, df in enumerate(dfs):
color = i % (len(colors) - 1)
df = prep_df(df)
df = df[df['Sold Date'] >= start_date]
df = df[df['Sold Date'] <= end_date]
hash_rate = 1
# Source https://cryptoage.com/en/2380-the-current-table-with-the-hash-rate-of-videocards-for-2021.html
if '3060 Ti' in df['item'].iloc[0]:
hash_rate = 59
elif '3060' in df['item'].iloc[0]:
hash_rate = 37.5
elif '3070' in df['item'].iloc[0]:
hash_rate = 59
elif '3080 Ti' in df['item'].iloc[0]:
hash_rate = 64
elif '3080' in df['item'].iloc[0]:
hash_rate = 100
elif '3090' in df['item'].iloc[0]:
hash_rate = 111
med_prices = df.groupby(
['Sold Date'])['Total Price'].median() / hash_rate / eth_prices
# med_prices = 100 * med_prices / med_prices[0]
# print(df['item'].iloc[0], 'R Squared:', r2_score(eth_prices[-(len(med_price_scaled)):], med_price_scaled))
# print(df['item'].iloc[0], 'R Squared:', r2_score(med_prices, eth_prices[-(len(med_prices)):]))
# med_mad = robust.mad(df.groupby(['Sold Date'])['Total Price']/ msrps[i] * 100)
# print(med_mad)
if roll > 0:
med_prices = med_prices.rolling(roll, min_periods=1).mean()
min_msrp = min(min_msrp, min(med_prices))
plt.plot(med_prices, colors[color], label=df['item'].iloc[0])
# plt.fill_between(med_price_scaled, med_price_scaled - med_mad, med_price_scaled + med_mad, color=colors[ci])
plt.plot(eth_prices, label='Etherium')
# plt.plot(btc_prices, label='Bitcoin')
# plt.ylim(bottom=min_msrp)
plt.legend()
# plt.tight_layout()
plt.subplots_adjust(bottom=0.2)
# plt.tight_layout()
if roll > 0:
plt.title(f"{title} {roll} Day Rolling Average - % MSRP")
plt.savefig(f"Images/{title} {roll} Day Rolling Average - % MSRP")
else:
plt.title(f"{title} - % MSRP")
plt.savefig(f"Images/{title} - % MSRP")
if e_vars.show_plots: plt.show()
# Plotting the non-scaled graph
plt.figure() # In this example, all the plots will be in one figure.
fig, ax1 = plt.subplots()
plt.ylabel(f"Median Sale Price ({e_vars.ccode})")
plt.xlabel("Sale Date")
plt.tick_params(axis='y')
plt.tick_params(axis='x', rotation=30)
if roll > 0:
plt.title(f"{title} {roll} Day Rolling Average - {e_vars.ccode}")
else:
plt.title(f"{title} - {e_vars.ccode}")
for i, df in enumerate(dfs):
color = i % (len(colors) - 1)
df = df[df['Sold Date'] >= datetime(2021, 1, 1)]
med_price = df.groupby(['Sold Date'])['Total Price'].median()
if roll > 0:
med_price = med_price.rolling(roll, min_periods=1).mean()
min_msrp = min(min_msrp, min(med_price))
plt.plot(med_price, colors[color], label=df['item'].iloc[0])
# plt.ylim(bottom=min_msrp)
formatter = ticker.FormatStrFormatter(f'{e_vars.ccode}%1.0f')
ax1.yaxis.set_major_formatter(formatter)
plt.legend()
plt.tight_layout()
# plt.savefig(f"Images/{title} - {e_vars.ccode}")
# if e_vars.show_plots: plt.show()
plt.show()