Parametric OHLCV Cross Strategy with Tunable Entry and Exit Rules
Article Strategy library · Author: @Mablue (Masoud Azizi)
Summary
Diamond is a configurable long-only trading strategy built around crossings between selected dataframe fields. Entry occurs when a chosen fast field, shifted by a configurable number of bars, crosses above a chosen slow field multiplied by a scale factor. Exit uses a separately configured cross below rule. The selectable fields include OHLCV data, and custom indicators can be added for use in the same framework.
The document includes a five-minute timeframe, example hyperparameter settings, and several reported optimization runs. Results vary by objective and run, with different trade counts and return figures, which illustrates that parameter selection materially changes outcomes; these figures are not a controlled comparison or proof of durable performance. The entry and exit rules are generic rather than tied to a stated market rationale, and the code's default indicator fields need to be populated before use. The strategy also configures ROI targets, a stop loss, and trailing stops, so exits can involve both signal and risk settings.
Key ideas
- Entry and exit signals are generated by configurable crossovers between selected dataframe fields.
- The fast field can be shifted by a tunable number of bars and compared with a scaled slow field.
- The field choices include OHLCV data, with custom indicators available after they are added to the dataframe.
- Reported optimization outcomes vary across objectives and runs, so they do not establish robust performance.
- The strategy combines signal exits with configured ROI, stop-loss, and trailing-stop settings.
Tags
Full text
# Diamond
# Diamond
## Source (GPL-3.0)
```python
# 𝐼𝓉 𝒾𝓈 𝒟𝒾𝒶𝓂𝑜𝓃𝒹 𝒮𝓉𝓇𝒶𝓉𝑒𝑔𝓎.
# 𝒯𝒽𝒶𝓉 𝓉𝒶𝓀𝑒𝓈 𝒽𝑒𝓇 𝑜𝓌𝓃 𝓇𝒾𝑔𝒽𝓉𝓈 𝓁𝒾𝓀𝑒 𝒜𝒻𝑔𝒽𝒶𝓃𝒾𝓈𝓉𝒶𝓃 𝓌𝑜𝓂𝑒𝓃
# 𝒯𝒽𝑜𝓈𝑒 𝓌𝒽𝑜 𝓈𝓉𝒾𝓁𝓁 𝓅𝓇𝑜𝓊𝒹 𝒶𝓃𝒹 𝒽𝑜𝓅𝑒𝒻𝓊𝓁.
# 𝒯𝒽𝑜𝓈𝑒 𝓌𝒽𝑜 𝓉𝒽𝑒 𝓂𝑜𝓈𝓉 𝒷𝑒𝒶𝓊𝓉𝒾𝒻𝓊𝓁 𝒸𝓇𝑒𝒶𝓉𝓊𝓇𝑒𝓈 𝒾𝓃 𝓉𝒽𝑒 𝒹𝑒𝓅𝓉𝒽𝓈 𝑜𝒻 𝓉𝒽𝑒 𝒹𝒶𝓇𝓀𝑒𝓈𝓉.
# 𝒯𝒽𝑜𝓈𝑒 𝓌𝒽𝑜 𝓈𝒽𝒾𝓃𝑒 𝓁𝒾𝓀𝑒 𝒹𝒾𝒶𝓂𝑜𝓃𝒹𝓈 𝒷𝓊𝓇𝒾𝑒𝒹 𝒾𝓃 𝓉𝒽𝑒 𝒽𝑒𝒶𝓇𝓉 𝑜𝒻 𝓉𝒽𝑒 𝒹𝑒𝓈𝑒𝓇𝓉 ...
# 𝒲𝒽𝓎 𝓃𝑜𝓉 𝒽𝑒𝓁𝓅 𝓌𝒽𝑒𝓃 𝓌𝑒 𝒸𝒶𝓃?
# 𝐼𝒻 𝓌𝑒 𝒷𝑒𝓁𝒾𝑒𝓋𝑒 𝓉𝒽𝑒𝓇𝑒 𝒾𝓈 𝓃𝑜 𝓂𝒶𝓃 𝓁𝑒𝒻𝓉 𝓌𝒾𝓉𝒽 𝓉𝒽𝑒𝓂
# (𝒲𝒽𝒾𝒸𝒽 𝒾𝓈 𝓅𝓇𝑜𝒷𝒶𝒷𝓁𝓎 𝓉𝒽𝑒 𝓅𝓇𝑜𝒹𝓊𝒸𝓉 𝑜𝒻 𝓉𝒽𝑒 𝓉𝒽𝑜𝓊𝑔𝒽𝓉 𝑜𝒻 𝓅𝒶𝒾𝓃𝓁𝑒𝓈𝓈 𝒸𝑜𝓇𝓅𝓈𝑒𝓈)
# 𝒲𝒽𝑒𝓇𝑒 𝒽𝒶𝓈 𝑜𝓊𝓇 𝒽𝓊𝓂𝒶𝓃𝒾𝓉𝓎 𝑔𝑜𝓃𝑒?
# 𝒲𝒽𝑒𝓇𝑒 𝒽𝒶𝓈 𝒽𝓊𝓂𝒶𝓃𝒾𝓉𝓎 𝑔𝑜𝓃𝑒?
# 𝒲𝒽𝓎 𝓃𝑜𝓉 𝒽𝑒𝓁𝓅 𝓌𝒽𝑒𝓃 𝓌𝑒 𝒸𝒶𝓃?
# 𝓁𝑒𝓉𝓈 𝓅𝒾𝓅 𝓊𝓃𝒾𝓃𝓈𝓉𝒶𝓁𝓁 𝓉𝒶-𝓁𝒾𝒷 𝑜𝓃 𝒜𝒻𝑔𝒽𝒶𝓃𝒾𝓈𝓉𝒶𝓃
# IMPORTANT: Diamond strategy is designed to be pure and
# cuz of that it have not any indicator population. idea is that
# It is just use the pure dataframe ohlcv data for calculation
# of buy/sell signals, But you can add your indicators and add
# your key names inside catagorical hyperoptable params and
# than you be able to hyperopt them as well.
# thanks to: @Kroissan, @drakes00 And @xmatthias for his patience and helps
# Author: @Mablue (Masoud Azizi)
# github: https://github.com/mablue/
# * freqtrade backtesting --strategy Diamond
# freqtrade hyperopt --hyperopt-loss ShortTradeDurHyperOptLoss --spaces buy sell roi trailing stoploss --strategy Diamond -j 2 -e 10
# * 3/10: 76 trades. 51/18/7 Wins/Draws/Losses. Avg profit 1.92%. Median profit 2.40%. Total profit 0.04808472 BTC ( 48.08%). Avg duration 5:06:00 min. Objective: 1.75299
# freqtrade hyperopt --hyperopt-loss OnlyProfitHyperOptLoss --spaces buy sell roi trailing stoploss --strategy Diamond -j 2 -e 10
# * 10/10: 76 trades. 39/34/3 Wins/Draws/Losses. Avg profit 0.61%. Median profit 0.05%. Total profit 0.01528359 BTC ( 15.28%). Avg duration 17:32:00 min. Objective: -0.01528
# freqtrade hyperopt --hyperopt-loss SharpeHyperOptLoss --spaces buy sell roi trailing stoploss --strategy Diamond -j 2 -e 10
# * 4/10: 15 trades. 10/2/3 Wins/Draws/Losses. Avg profit 1.52%. Median profit 7.99%. Total profit 0.00754274 BTC ( 7.54%). Avg duration 1 day, 0:04:00 min. Objective: -0.90653
# freqtrade hyperopt --hyperopt-loss SharpeHyperOptLossDaily --spaces buy sell roi trailing stoploss --strategy Diamond -j 2 -e 10
# * 7/10: 130 trades. 68/54/8 Wins/Draws/Losses. Avg profit 0.71%. Median profit 0.06%. Total profit 0.03050369 BTC ( 30.50%). Avg duration 10:07:00 min. Objective: -11.08185
# freqtrade hyperopt --hyperopt-loss SortinoHyperOptLoss --spaces buy sell roi trailing stoploss --strategy Diamond -j 2 -e 10
# * 2/10: 10 trades. 7/0/3 Wins/Draws/Losses. Avg profit 5.50%. Median profit 7.05%. Total profit 0.01817970 BTC ( 18.18%). Avg duration 0:27:00 min. Objective: -11.72450
# freqtrade hyperopt --hyperopt-loss SortinoHyperOptLossDaily --spaces buy sell roi trailing stoploss --strategy Diamond -j 2 -e 10
# | * Best | 3/10 | 165 | 98 63 4 | 1.00% | 0.05453885 BTC (54.54%) | 0 days 08:02:00 | 0.00442974 BTC (13.41%) | -41.371 |
# | * Best | 7/10 | 101 | 56 42 3 | 0.73% | 0.02444518 BTC (24.45%) | 0 days 13:08:00 | 0.00107122 BTC (3.24%) | -66.7687 |
# * 7/10: 101 trades. 56/42/3 Wins/Draws/Losses. Avg profit 0.73%. Median profit 0.13%. Total profit 0.02444518 BTC ( 24.45%). Avg duration 13:08:00 min. Objective: -66.76866
# freqtrade hyperopt --hyperopt-loss OnlyProfitHyperOptLoss --spaces buy sell roi trailing stoploss --strategy Diamond -j 2 -e 10
# * 7/10: 117 trades. 74/41/2 Wins/Draws/Losses. Avg profit 1.91%. Median profit 1.50%. Total profit 0.07370921 BTC ( 73.71%). Avg duration 9:26:00 min. Objective: -0.07371
# --- Do not remove these libs ---
from freqtrade.strategy import CategoricalParameter, DecimalParameter, IntParameter, IStrategy
from pandas import DataFrame
# --------------------------------
# Add your lib to import here
import talib.abstract as ta
from functools import reduce
import freqtrade.vendor.qtpylib.indicators as qtpylib
class Diamond(IStrategy):
# ###################### RESULT PLACE ######################
# Config: 5 x UNLIMITED STOCK costume pair list,
# hyperopt : 5000 x SortinoHyperOptLossDaily,
# 34/5000: 297 trades. 136/156/5 Wins/Draws/Losses. Avg profit 0.49%. Median profit 0.00%. Total profit 45.84477237 USDT ( 33.96Σ%). Avg duration 11:54:00 min. Objective: -46.50379
INTERFACE_VERSION: int = 3
# Buy hyperspace params:
buy_params = {
"buy_fast_key": "high",
"buy_horizontal_push": 7,
"buy_slow_key": "volume",
"buy_vertical_push": 0.942,
}
# Sell hyperspace params:
sell_params = {
"sell_fast_key": "high",
"sell_horizontal_push": 10,
"sell_slow_key": "low",
"sell_vertical_push": 1.184,
}
# ROI table:
minimal_roi = {
"0": 0.242,
"13": 0.044,
"51": 0.02,
"170": 0
}
# Stoploss:
stoploss = -0.271
# Trailing stop:
trailing_stop = True
trailing_stop_positive = 0.011
trailing_stop_positive_offset = 0.054
trailing_only_offset_is_reached = False
# timeframe
timeframe = '5m'
# #################### END OF RESULT PLACE ####################
buy_vertical_push = DecimalParameter(0.5, 1.5, decimals=3, default=1, space='buy')
buy_horizontal_push = IntParameter(0, 10, default=0, space='buy')
buy_fast_key = CategoricalParameter(['open', 'high', 'low', 'close', 'volume',
# you can not enable this lines befour you
# populate an indicator for them and set
# the same key name for it
# 'ma_fast', 'ma_slow', {...}
], default='ma_fast', space='buy')
buy_slow_key = CategoricalParameter(['open', 'high', 'low', 'close', 'volume',
# 'ma_fast', 'ma_slow', {...}
], default='ma_slow', space='buy')
sell_vertical_push = DecimalParameter(0.5, 1.5, decimals=3, default=1, space='sell')
sell_horizontal_push = IntParameter(0, 10, default=0, space='sell')
sell_fast_key = CategoricalParameter(['open', 'high', 'low', 'close', 'volume',
# 'ma_fast', 'ma_slow', {...}
], default='ma_fast', space='sell')
sell_slow_key = CategoricalParameter(['open', 'high', 'low', 'close', 'volume',
# 'ma_fast', 'ma_slow', {...}
], default='ma_slow', space='sell')
def populate_indicators(self, dataframe: DataFrame, metadata: dict) -> DataFrame:
# you can add new indicators and enable them inside
# hyperoptable categorical params on the top
# dataframe['ma_fast'] = ta.SMA(dataframe, timeperiod=9)
# dataframe['ma_slow'] = ta.SMA(dataframe, timeperiod=18)
# dataframe['{...}'] = ta.{...}(dataframe, timeperiod={...})
return dataframe
def populate_entry_trend(self, dataframe: DataFrame, metadata: dict) -> DataFrame:
conditions = []
conditions.append(
qtpylib.crossed_above
(
dataframe[self.buy_fast_key.value].shift(self.buy_horizontal_push.value),
dataframe[self.buy_slow_key.value] * self.buy_vertical_push.value
)
)
if conditions:
dataframe.loc[
reduce(lambda x, y: x & y, conditions),
'enter_long']=1
return dataframe
def populate_exit_trend(self, dataframe: DataFrame, metadata: dict) -> DataFrame:
conditions = []
conditions.append(
qtpylib.crossed_below
(
dataframe[self.sell_fast_key.value].shift(self.sell_horizontal_push.value),
dataframe[self.sell_slow_key.value] * self.sell_vertical_push.value
)
)
if conditions:
dataframe.loc[
reduce(lambda x, y: x & y, conditions),
'exit_long']=1
return dataframe
```Shown in full with attribution under the source's licence. Licence: GPL-3.0
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.