def _exit_by_max_holding_time(
trades: Trades,
df: pd.DataFrame,
max_holding_time: pd.Timedelta,
exit_condition: Callable[[pd.DataFrame, pd.Series, pd.Timestamp],
bool],
) -> Trades:
indices = [] # type: List[pd.Timestamp]
exits = [] # type: List[Tuple[float, int]]
for i in trades.ids:
trade = trades.get_trade(i)
if trade.sum() == 0:
continue
idx = trade.index[0]
start = max(idx, df.index[0])
end = min(idx + max_holding_time, df.index[-1])
df_exit = df.loc[start:end]
transaction = _exit_transaction(df_exit, trade, exit_condition)
indices.append(transaction.timestamp)
exits.append((transaction.amount, i))
df = pd.DataFrame(index=indices, data=exits, columns=["amount", "_id"])
return from_dataframe(df, symbol, currency_unit)
def count_trades(trades: Trades, mkt: MarketData) -> Tuple[int, int, int]:
"""Count winning, losing and total number of trades.
Args:
trades : Trades to evaluate. Each trade is evaluated
only if it contains more than one transactions,
because we can define pl in that case.
mkt: Market data. The index should contains all trades' index.
Returns:
total count, win count, lose count
"""
pls = [
_calc_pl(trades.get_trade(i), mkt) for i in trades.ids
if len(trades.get_trade(i).index) > 1
]
total = len(trades.ids)
win = sum([pl > 0.0 for pl in pls])
lose = sum([pl < 0.0 for pl in pls])
return total, win, lose
def skip_entry_by_hours(trades: Trades, hours: List[int]) -> Trades:
"""Skip entry by hours.
Args:
trades: Trades
hours: Hours which will be filtered out from entry.
Result:
Trades
"""
deleted_ids = []
for i in trades.ids:
entry = trades.get_trade(i).index[0]
if entry.hour in hours:
deleted_ids.append(i)
return trades[~trades["_id"].isin(deleted_ids)]
def skip_entry_by_spread(trades: Trades, mkt: AskBidMarketData,
max_spread: float) -> Trades:
"""Skip entry by spread.
Args:
trades: Trades
mkt: Market data for ask/bid prices
max_spread: More than the value, skip entry
Result:
Trades
"""
assert max_spread >= 0.0
assert trades.index.unique().isin(mkt.index).all()
spread = mkt.spread
deleted_ids = [] # type: List[int]
for i in trades.ids:
entry = trades.get_trade(i).index[0]
if spread.at[entry] > max_spread:
deleted_ids.append(i)
return trades[~trades["_id"].isin(deleted_ids)]
def _exit(
trades: Trades,
df: pd.DataFrame,
exit_condition: Callable[[pd.DataFrame, pd.Series], pd.Series],
) -> pd.Series:
indices = [] # type: List[pd.Timestamp]
exits = [] # type: List[Tuple[float, int]]
for i in trades.ids:
trade = trades.get_trade(i)
if trade.sum() == 0:
continue
idx = trade.index[0]
df_exit = df[idx <= df.index]
transaction = _exit_transaction(df_exit, trade, exit_condition)
indices.append(transaction.timestamp)
exits.append((transaction.amount, i))
df = pd.DataFrame(index=indices, data=exits, columns=["amount", "_id"])
return from_dataframe(df, symbol)