"""Generate synthetic price data."""
import datetime
import random
from pathlib import Path
from typing import Dict, Iterable
import pandas as pd
from tradeexecutor.state.identifier import TradingPairIdentifier
from tradeexecutor.state.types import USDollarPrice
from tradingstrategy.chain import ChainId
from tradingstrategy.timebucket import TimeBucket
from tradingstrategy.types import USDollarAmount
[docs]def generate_ohlcv_candles(
bucket: TimeBucket,
start: datetime.datetime,
end: datetime.datetime,
start_price: int | float=1800,
daily_drift=(0.95, 1.05),
high_drift=1.05,
low_drift=0.90,
random_seed=1,
pair_id=1,
exchange_id=1,
) -> pd.DataFrame:
"""Generate some sample time series data.
The output candles are deterministic: the same input parameters result to the same output parameters.
:param bucket:
Time bucket to use for the candles
:param start:
Start time for the candles
:param end:
End time for the candles
:param start_price:
Starting price for the candles
:param daily_drift:
Tuple of (min, max) daily drift for the candles
:param high_drift:
High drift for the candles
:param low_drift:
Low drift for the candles
:param random_seed:
Random seed to use for the candles
:param pair_id:
Pair ID to use for the candles
:param exchange_id:
Exchange ID to use for the candles
"""
random_gen = random.Random(random_seed)
time_delta = bucket.to_timedelta()
open = start_price
now = start
assert pair_id is not None
data = []
while now < end:
close = random_gen.uniform(open * daily_drift[0], open * daily_drift[1])
high = random_gen.uniform(open, open * high_drift)
low = random_gen.uniform(open, open * low_drift)
data.append({
"pair_id": pair_id,
"timestamp": now,
"open": open,
"close": close,
"high": high,
"low": low,
"buy_volume": 0,
"sell_volume": 0,
"buys": 0,
"sells": 0,
"start_block": int(now.timestamp()),
"end_block": int(now.timestamp()),
})
open = close
now += time_delta
df = pd.DataFrame(data)
df.set_index("timestamp", drop=False, inplace=True)
df["volume"] = df["buy_volume"] + df["sell_volume"] # Convert from Uni v2 style volume
return df
[docs]def generate_multi_pair_candles(
time_bucket: TimeBucket,
start: datetime.datetime,
end: datetime.datetime,
pairs: Dict[TradingPairIdentifier, USDollarPrice],
random_seed=1,
) -> pd.DataFrame:
"""Generate synthetic tarding data for multiple trading pairs.
:param pairs:
Map of trading pairs and their starting prices
"""
segments = []
for pair, price in pairs.items():
df = generate_ohlcv_candles(
time_bucket,
start,
end,
start_price=price,
pair_id=pair.internal_id,
exchange_id=pair.internal_exchange_id,
random_seed=random_seed,
)
segments.append(df)
return pd.concat(segments)
[docs]def generate_fixed_price_candles(
bucket: TimeBucket,
start: datetime.datetime,
end: datetime.datetime,
pair_price_map: Dict[TradingPairIdentifier, USDollarAmount],
) -> pd.DataFrame:
"""Generate flat prices for several assets.
Creates fake fixed price data where prices are stable over a period of time.
"""
data = []
for pair, price in pair_price_map.items():
time_delta = bucket.to_timedelta()
open = price
close = price
high = price
low = price
now = start
while now < end:
data.append({
"pair_id": pair.internal_id,
"timestamp": now,
"open": open,
"close": close,
"high": high,
"low": low,
"buy_volume": 0,
"sell_volume": 0,
"buys": 0,
"sells": 0,
"start_block": int(now.timestamp()),
"end_block": int(now.timestamp()),
})
open = close
now += time_delta
df = pd.DataFrame(data)
df.set_index("timestamp", drop=False, inplace=True)
return df
[docs]def load_ohlcv_parquet_file(path: Path, chain_id: ChainId, exchange_id: int, pair_id: int) -> pd.DataFrame:
"""Load OHLCV data directly from a parquet file.
- Assume one file for one pair
- Columns: timestamp, open, close, high, low, volume
"""
assert isinstance(path, Path)
assert path.exists()
df = pd.read_parquet(path)
df.set_index("timestamp", drop=False, inplace=True)
# Fill data we need for the universe
# https://stackoverflow.com/a/69822548/315168
df.loc[:, "pair_id"] = pair_id
df.loc[:, "exchange_id"] = exchange_id
df.loc[:, "chain_id"] = chain_id.value
return df