adelorenzo/stable
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Update btanalysis.py

Browse Source
This commit is contained in:
Guitheg 2021-12-09 17:20:05 +01:00
parent 3e7ecf931b
commit 553c5e9606
1 changed files with 35 additions and 27 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

58
freqtrade/data/btanalysis.py
View File

@ -390,19 +390,18 @@ def calculate_max_drawdown(trades: pd.DataFrame, *, date_col: str = 'close_date'
low_val = max_drawdown_df.loc[idxmin, 'cumulative'] low_val = max_drawdown_df.loc[idxmin, 'cumulative']
return abs(min(max_drawdown_df['drawdown'])), high_date, low_date, high_val, low_val return abs(min(max_drawdown_df['drawdown'])), high_date, low_date, high_val, low_val
def calculate_trades_mdd(data: dict, trades: pd.DataFrame, *, date_col: str = 'close_date', def calculate_trades_mdd(data: dict, trades: pd.DataFrame) -> float :
value_col: str = 'profit_ratio'
) -> float:
""" """
Calculate MDD (Max DrawDown) : Calculate Trades MDD (Max DrawDown) :
Give the max drawdown given each trades and the history candles data. Give the max drawdown given each trades and the history candles.
The intervals dates used to calculate the max drawdown are the trades intervals dates.
Args: Args:
trades (pd.DataFrame): [description] :param data (dict) : dictionnary of candle dataframe per pair used to calculate the mdd.
date_col (str, optional): [description]. Defaults to 'close_date'. :param trades (pd.DataFrame): trades used to find the intervals dates.
value_col (str, optional): [description]. Defaults to 'profit_ratio'.
Returns: Returns:
float: [description] :return: float: Give the maximum drawdown among each trades.
:raise: ValueError if trade-dataframe was found empty.
""" """
if len(trades) == 0: if len(trades) == 0:
raise ValueError("Trade dataframe empty") raise ValueError("Trade dataframe empty")
@ -411,36 +410,47 @@ def calculate_trades_mdd(data: dict, trades: pd.DataFrame, *, date_col: str = 'c
for pair, df in data.items(): for pair, df in data.items():
# Gather the opening and closing trade dates into one Dates DataFrame
open_close_trades = trades.loc[trades['pair']==pair][["open_date","close_date"]] open_close_trades = trades.loc[trades['pair']==pair][["open_date","close_date"]]
open_close_trades = pd.concat( open_close_trades = pd.concat(
[open_close_trades.rename(columns={'open_date':'date'})[['date']], [open_close_trades.rename(columns={'open_date':'date'})[['date']],
open_close_trades.rename(columns={'close_date':'date'})[['date']]] open_close_trades.rename(columns={'close_date':'date'})[['date']]]
).sort_values(by='date') ).sort_values(by='date')
open_close_trades['open_close_mark'] = np.ones(len(open_close_trades), dtype=int)
# Mark the dates and join it to the current candle dataframe.
# This allow to determine the open and close trade dates in the current
# candle dataframe.
open_close_trades['open_close_mark'] = 1
data_join = df.set_index('date').join(open_close_trades.set_index('date')) data_join = df.set_index('date').join(open_close_trades.set_index('date'))
del open_close_trades del open_close_trades
close_trades = trades.loc[trades['pair']==pair][["close_date"]] # Gather, mark and join only the opening trade dates into the current candle
close_trades = close_trades.rename(columns={'close_date':'date'}) # dataframe.
close_trades['close_mark'] = np.ones(len(close_trades), dtype=int) # This allow to classify trades using the cumsum and split by classes
data_join = data_join.join(close_trades.set_index('date')) # with groupby in order to process a cummax on each trades independantly.
del close_trades open_trades = trades.loc[trades['pair']==pair][["open_date"]]
open_trades = open_trades.rename(columns={'open_date':'date'})
open_trades['open_mark'] = 1
data_join = data_join.join(open_trades.set_index('date'))
del open_trades
data_join[["open_close_mark",'close_mark']] = data_join[ # Set all unmarked date to 0
["open_close_mark",'close_mark']].fillna(0).astype(int) data_join[["open_close_mark",'open_mark']] = data_join[
["open_close_mark",'open_mark']].fillna(0).astype(int)
# Mark with one all dates between an opening date trades and a closing date trades.
data_join['is_in_trade'] = data_join.open_close_mark.cumsum()&1 # &1 <=> %2 data_join['is_in_trade'] = data_join.open_close_mark.cumsum()&1 # &1 <=> %2
data_join.loc[data_join['open_close_mark'] == 1, 'is_in_trade'] = 1 data_join.loc[data_join['open_close_mark'] == 1, 'is_in_trade'] = 1
# Perform a cummax in each trades independtly
data_join['close_cummax'] = 0 data_join['close_cummax'] = 0
data_join["open_close_mark"] = data_join["open_close_mark"] - data_join["close_mark"]
data_join['close_cummax'] = data_join.groupby( data_join['close_cummax'] = data_join.groupby(
data_join.open_close_mark.cumsum() data_join['open_mark'].cumsum()
).close.cummax() )['close'].cummax()
data_join.loc[data_join['is_in_trade'] == 0, 'close_cummax'] = 0 data_join.loc[data_join['is_in_trade'] == 0, 'close_cummax'] = 0
data_join = data_join.rename(columns={'close_mark':'drawdown'}) # Compute the drawdown at each time of each trades
data_join = data_join.rename(columns={'open_mark':'drawdown'})
data_join.loc[data_join['is_in_trade'] == 1, 'drawdown'] = \ data_join.loc[data_join['is_in_trade'] == 1, 'drawdown'] = \
(data_join['close_cummax'] - data_join['close']) \ (data_join['close_cummax'] - data_join['close']) \
/ data_join['close_cummax'] / data_join['close_cummax']
@ -449,8 +459,6 @@ def calculate_trades_mdd(data: dict, trades: pd.DataFrame, *, date_col: str = 'c
trades_mdd_pair_list.append(mdd_pair) trades_mdd_pair_list.append(mdd_pair)
trades_mdd_pair_list = np.array(trades_mdd_pair_list) trades_mdd_pair_list = np.array(trades_mdd_pair_list)
return trades_mdd_pair_list.max() return trades_mdd_pair_list.max()
def calculate_csum(trades: pd.DataFrame, starting_balance: float = 0) -> Tuple[float, float]: def calculate_csum(trades: pd.DataFrame, starting_balance: float = 0) -> Tuple[float, float]: