freqtrade_origin/docs/strategy-customization.md
2020-02-06 21:53:03 +03:00

24 KiB

Strategy Customization

This page explains where to customize your strategies, and add new indicators.

Install a custom strategy file

This is very simple. Copy paste your strategy file into the directory user_data/strategies.

Let assume you have a class called AwesomeStrategy in the file AwesomeStrategy.py:

  1. Move your file into user_data/strategies (you should have user_data/strategies/AwesomeStrategy.py
  2. Start the bot with the param --strategy AwesomeStrategy (the parameter is the class name)
freqtrade trade --strategy AwesomeStrategy

Develop your own strategy

The bot includes a default strategy file. Also, several other strategies are available in the strategy repository.

You will however most likely have your own idea for a strategy. This document intends to help you develop one for yourself.

To get started, use freqtrade new-strategy --strategy AwesomeStrategy. This will create a new strategy file from a template, which will be located under user_data/strategies/AwesomeStrategy.py.

!!! Note This is just a template file, which will most likely not be profitable out of the box.

Anatomy of a strategy

A strategy file contains all the information needed to build a good strategy:

  • Indicators
  • Buy strategy rules
  • Sell strategy rules
  • Minimal ROI recommended
  • Stoploss strongly recommended

The bot also include a sample strategy called SampleStrategy you can update: user_data/strategies/sample_strategy.py. You can test it with the parameter: --strategy SampleStrategy

Additionally, there is an attribute called INTERFACE_VERSION, which defines the version of the strategy interface the bot should use. The current version is 2 - which is also the default when it's not set explicitly in the strategy.

Future versions will require this to be set.

freqtrade trade --strategy AwesomeStrategy

For the following section we will use the user_data/strategies/sample_strategy.py file as reference.

!!! Note "Strategies and Backtesting" To avoid problems and unexpected differences between Backtesting and dry/live modes, please be aware that during backtesting the full time-interval is passed to the populate_*() methods at once. It is therefore best to use vectorized operations (across the whole dataframe, not loops) and avoid index referencing (df.iloc[-1]), but instead use df.shift() to get to the previous candle.

!!! Warning "Warning: Using future data" Since backtesting passes the full time interval to the populate_*() methods, the strategy author needs to take care to avoid having the strategy utilize data from the future. Some common patterns for this are listed in the Common Mistakes section of this document.

Customize Indicators

Buy and sell strategies need indicators. You can add more indicators by extending the list contained in the method populate_indicators() from your strategy file.

You should only add the indicators used in either populate_buy_trend(), populate_sell_trend(), or to populate another indicator, otherwise performance may suffer.

It's important to always return the dataframe without removing/modifying the columns "open", "high", "low", "close", "volume", otherwise these fields would contain something unexpected.

Sample:

def populate_indicators(self, dataframe: DataFrame, metadata: dict) -> DataFrame:
    """
    Adds several different TA indicators to the given DataFrame

    Performance Note: For the best performance be frugal on the number of indicators
    you are using. Let uncomment only the indicator you are using in your strategies
    or your hyperopt configuration, otherwise you will waste your memory and CPU usage.
    :param dataframe: Raw data from the exchange and parsed by parse_ticker_dataframe()
    :param metadata: Additional information, like the currently traded pair
    :return: a Dataframe with all mandatory indicators for the strategies
    """
    dataframe['sar'] = ta.SAR(dataframe)
    dataframe['adx'] = ta.ADX(dataframe)
    stoch = ta.STOCHF(dataframe)
    dataframe['fastd'] = stoch['fastd']
    dataframe['fastk'] = stoch['fastk']
    dataframe['blower'] = ta.BBANDS(dataframe, nbdevup=2, nbdevdn=2)['lowerband']
    dataframe['sma'] = ta.SMA(dataframe, timeperiod=40)
    dataframe['tema'] = ta.TEMA(dataframe, timeperiod=9)
    dataframe['mfi'] = ta.MFI(dataframe)
    dataframe['rsi'] = ta.RSI(dataframe)
    dataframe['ema5'] = ta.EMA(dataframe, timeperiod=5)
    dataframe['ema10'] = ta.EMA(dataframe, timeperiod=10)
    dataframe['ema50'] = ta.EMA(dataframe, timeperiod=50)
    dataframe['ema100'] = ta.EMA(dataframe, timeperiod=100)
    dataframe['ao'] = awesome_oscillator(dataframe)
    macd = ta.MACD(dataframe)
    dataframe['macd'] = macd['macd']
    dataframe['macdsignal'] = macd['macdsignal']
    dataframe['macdhist'] = macd['macdhist']
    hilbert = ta.HT_SINE(dataframe)
    dataframe['htsine'] = hilbert['sine']
    dataframe['htleadsine'] = hilbert['leadsine']
    dataframe['plus_dm'] = ta.PLUS_DM(dataframe)
    dataframe['plus_di'] = ta.PLUS_DI(dataframe)
    dataframe['minus_dm'] = ta.MINUS_DM(dataframe)
    dataframe['minus_di'] = ta.MINUS_DI(dataframe)
    return dataframe

!!! Note "Want more indicator examples?" Look into the user_data/strategies/sample_strategy.py. Then uncomment indicators you need.

Strategy startup period

Most indicators have an instable startup period, in which they are either not available, or the calculation is incorrect. This can lead to inconsistencies, since Freqtrade does not know how long this instable period should be. To account for this, the strategy can be assigned the startup_candle_count attribute. This should be set to the maximum number of candles that the strategy requires to calculate stable indicators.

In this example strategy, this should be set to 100 (startup_candle_count = 100), since the longest needed history is 100 candles.

    dataframe['ema100'] = ta.EMA(dataframe, timeperiod=100)

By letting the bot know how much history is needed, backtest trades can start at the specified timerange during backtesting and hyperopt.

!!! Warning startup_candle_count should be below ohlcv_candle_limit (which is 500 for most exchanges) - since only this amount of candles will be available during Dry-Run/Live Trade operations.

Example

Let's try to backtest 1 month (January 2019) of 5m candles using the an example strategy with EMA100, as above.

freqtrade backtesting --timerange 20190101-20190201 --ticker-interval 5m

Assuming startup_candle_count is set to 100, backtesting knows it needs 100 candles to generate valid buy signals. It will load data from 20190101 - (100 * 5m) - which is ~2019-12-31 15:30:00. If this data is available, indicators will be calculated with this extended timerange. The instable startup period (up to 2019-01-01 00:00:00) will then be removed before starting backtesting.

!!! Note If data for the startup period is not available, then the timerange will be adjusted to account for this startup period - so Backtesting would start at 2019-01-01 08:30:00.

Buy signal rules

Edit the method populate_buy_trend() in your strategy file to update your buy strategy.

It's important to always return the dataframe without removing/modifying the columns "open", "high", "low", "close", "volume", otherwise these fields would contain something unexpected.

This will method will also define a new column, "buy", which needs to contain 1 for buys, and 0 for "no action".

Sample from user_data/strategies/sample_strategy.py:

def populate_buy_trend(self, dataframe: DataFrame, metadata: dict) -> DataFrame:
    """
    Based on TA indicators, populates the buy signal for the given dataframe
    :param dataframe: DataFrame populated with indicators
    :param metadata: Additional information, like the currently traded pair
    :return: DataFrame with buy column
    """
    dataframe.loc[
        (
            (qtpylib.crossed_above(dataframe['rsi'], 30)) &  # Signal: RSI crosses above 30
            (dataframe['tema'] <= dataframe['bb_middleband']) &  # Guard
            (dataframe['tema'] > dataframe['tema'].shift(1)) &  # Guard
            (dataframe['volume'] > 0)  # Make sure Volume is not 0
        ),
        'buy'] = 1

    return dataframe

!!! Note Buying requires sellers to buy from - therefore volume needs to be > 0 (dataframe['volume'] > 0) to make sure that the bot does not buy/sell in no-activity periods.

Sell signal rules

Edit the method populate_sell_trend() into your strategy file to update your sell strategy. Please note that the sell-signal is only used if use_sell_signal is set to true in the configuration.

It's important to always return the dataframe without removing/modifying the columns "open", "high", "low", "close", "volume", otherwise these fields would contain something unexpected.

This will method will also define a new column, "sell", which needs to contain 1 for sells, and 0 for "no action".

Sample from user_data/strategies/sample_strategy.py:

def populate_sell_trend(self, dataframe: DataFrame, metadata: dict) -> DataFrame:
    """
    Based on TA indicators, populates the sell signal for the given dataframe
    :param dataframe: DataFrame populated with indicators
    :param metadata: Additional information, like the currently traded pair
    :return: DataFrame with buy column
    """
    dataframe.loc[
        (
            (qtpylib.crossed_above(dataframe['rsi'], 70)) &  # Signal: RSI crosses above 70
            (dataframe['tema'] > dataframe['bb_middleband']) &  # Guard
            (dataframe['tema'] < dataframe['tema'].shift(1)) &  # Guard
            (dataframe['volume'] > 0)  # Make sure Volume is not 0
        ),
        'sell'] = 1
    return dataframe

Minimal ROI

This dict defines the minimal Return On Investment (ROI) a trade should reach before selling, independent from the sell signal.

It is of the following format, with the dict key (left side of the colon) being the minutes passed since the trade opened, and the value (right side of the colon) being the percentage.

minimal_roi = {
    "40": 0.0,
    "30": 0.01,
    "20": 0.02,
    "0": 0.04
}

The above configuration would therefore mean:

  • Sell whenever 4% profit was reached
  • Sell when 2% profit was reached (in effect after 20 minutes)
  • Sell when 1% profit was reached (in effect after 30 minutes)
  • Sell when trade is non-loosing (in effect after 40 minutes)

The calculation does include fees.

To disable ROI completely, set it to an insanely high number:

minimal_roi = {
    "0": 100
}

While technically not completely disabled, this would sell once the trade reaches 10000% Profit.

Stoploss

Setting a stoploss is highly recommended to protect your capital from strong moves against you.

Sample:

stoploss = -0.10

This would signify a stoploss of -10%.

For the full documentation on stoploss features, look at the dedicated stoploss page.

If your exchange supports it, it's recommended to also set "stoploss_on_exchange" in the order_types dictionary, so your stoploss is on the exchange and cannot be missed due to network problems, high load or other reasons.

For more information on order_types please look here.

Ticker interval

This is the set of candles the bot should download and use for the analysis. Common values are "1m", "5m", "15m", "1h", however all values supported by your exchange should work.

Please note that the same buy/sell signals may work with one interval, but not the other. This setting is accessible within the strategy by using self.ticker_interval.

Metadata dict

The metadata-dict (available for populate_buy_trend, populate_sell_trend, populate_indicators) contains additional information. Currently this is pair, which can be accessed using metadata['pair'] - and will return a pair in the format XRP/BTC.

The Metadata-dict should not be modified and does not persist information across multiple calls. Instead, have a look at the section Storing information

Storing information

Storing information can be accomplished by creating a new dictionary within the strategy class.

The name of the variable can be chosen at will, but should be prefixed with cust_ to avoid naming collisions with predefined strategy variables.

class Awesomestrategy(IStrategy):
    # Create custom dictionary
    cust_info = {}
    def populate_indicators(self, dataframe: DataFrame, metadata: dict) -> DataFrame:
        # Check if the entry already exists
        if "crosstime" in self.cust_info[metadata["pair"]:
            self.cust_info[metadata["pair"]["crosstime"] += 1
        else:
            self.cust_info[metadata["pair"]["crosstime"] = 1

!!! Warning The data is not persisted after a bot-restart (or config-reload). Also, the amount of data should be kept smallish (no DataFrames and such), otherwise the bot will start to consume a lot of memory and eventually run out of memory and crash.

!!! Note If the data is pair-specific, make sure to use pair as one of the keys in the dictionary.

Additional data (DataProvider)

The strategy provides access to the DataProvider. This allows you to get additional data to use in your strategy.

All methods return None in case of failure (do not raise an exception).

Please always check the mode of operation to select the correct method to get data (samples see below).

Possible options for DataProvider

  • available_pairs - Property with tuples listing cached pairs with their intervals (pair, interval).
  • ohlcv(pair, timeframe) - Currently cached ticker data for the pair, returns DataFrame or empty DataFrame.
  • historic_ohlcv(pair, timeframe) - Returns historical data stored on disk.
  • get_pair_dataframe(pair, timeframe) - This is a universal method, which returns either historical data (for backtesting) or cached live data (for the Dry-Run and Live-Run modes).
  • orderbook(pair, maximum) - Returns latest orderbook data for the pair, a dict with bids/asks with a total of maximum entries.
  • market(pair) - Returns market data for the pair: fees, limits, precisions, activity flag, etc. See ccxt documentation for more details on Market data structure.
  • runmode - Property containing the current runmode.

Example: fetch live ohlcv / historic data for the first informative pair

if self.dp:
    inf_pair, inf_timeframe = self.informative_pairs()[0]
    informative = self.dp.get_pair_dataframe(pair=inf_pair,
                                             timeframe=inf_timeframe)

!!! Warning "Warning about backtesting" Be carefull when using dataprovider in backtesting. historic_ohlcv() (and get_pair_dataframe() for the backtesting runmode) provides the full time-range in one go, so please be aware of it and make sure to not "look into the future" to avoid surprises when running in dry/live mode).

!!! Warning "Warning in hyperopt" This option cannot currently be used during hyperopt.

Orderbook

if self.dp:
    if self.dp.runmode in ('live', 'dry_run'):
        ob = self.dp.orderbook(metadata['pair'], 1)
        dataframe['best_bid'] = ob['bids'][0][0]
        dataframe['best_ask'] = ob['asks'][0][0]

!!! Warning The order book is not part of the historic data which means backtesting and hyperopt will not work if this method is used.

Available Pairs

if self.dp:
    for pair, ticker in self.dp.available_pairs:
        print(f"available {pair}, {ticker}")

Get data for non-tradeable pairs

Data for additional, informative pairs (reference pairs) can be beneficial for some strategies. Ohlcv data for these pairs will be downloaded as part of the regular whitelist refresh process and is available via DataProvider just as other pairs (see above). These parts will not be traded unless they are also specified in the pair whitelist, or have been selected by Dynamic Whitelisting.

The pairs need to be specified as tuples in the format ("pair", "interval"), with pair as the first and time interval as the second argument.

Sample:

def informative_pairs(self):
    return [("ETH/USDT", "5m"),
            ("BTC/TUSD", "15m"),
            ]

!!! Warning As these pairs will be refreshed as part of the regular whitelist refresh, it's best to keep this list short. All intervals and all pairs can be specified as long as they are available (and active) on the used exchange. It is however better to use resampling to longer time-intervals when possible to avoid hammering the exchange with too many requests and risk being blocked.

Additional data (Wallets)

The strategy provides access to the Wallets object. This contains the current balances on the exchange.

!!! Note Wallets is not available during backtesting / hyperopt.

Please always check if Wallets is available to avoid failures during backtesting.

if self.wallets:
    free_eth = self.wallets.get_free('ETH')
    used_eth = self.wallets.get_used('ETH')
    total_eth = self.wallets.get_total('ETH')

Possible options for Wallets

  • get_free(asset) - currently available balance to trade
  • get_used(asset) - currently tied up balance (open orders)
  • get_total(asset) - total available balance - sum of the 2 above

Additional data (Trades)

A history of Trades can be retrieved in the strategy by querying the database.

At the top of the file, import Trade.

from freqtrade.persistence import Trade

The following example queries for the current pair and trades from today, however other filters can easily be added.

if self.config['runmode'] in ('live', 'dry_run'):
    trades = Trade.get_trades([Trade.pair == metadata['pair'],
                               Trade.open_date > datetime.utcnow() - timedelta(days=1),
                               Trade.is_open == False,
                ]).order_by(Trade.close_date).all()
    # Summarize profit for this pair.
    curdayprofit = sum(trade.close_profit for trade in trades)

Get amount of stake_currency currently invested in Trades:

if self.config['runmode'] in ('live', 'dry_run'):
    total_stakes = Trade.total_open_trades_stakes()

Retrieve performance per pair. Returns a List of dicts per pair.

if self.config['runmode'] in ('live', 'dry_run'):
    performance = Trade.get_overall_performance()

Sample return value: ETH/BTC had 5 trades, with a total profit of 1.5% (ratio of 0.015).

{'pair': "ETH/BTC", 'profit': 0.015, 'count': 5}

!!! Warning Trade history is not available during backtesting or hyperopt.

Prevent trades from happening for a specific pair

Freqtrade locks pairs automatically for the current candle (until that candle is over) when a pair is sold, preventing an immediate re-buy of that pair.

Locked pairs will show the message Pair <pair> is currently locked..

Locking pairs from within the strategy

Sometimes it may be desired to lock a pair after certain events happen (e.g. multiple losing trades in a row).

Freqtrade has an easy method to do this from within the strategy, by calling self.lock_pair(pair, until). until must be a datetime object in the future, after which trading will be reenabled for that pair.

Locks can also be lifted manually, by calling self.unlock_pair(pair).

To verify if a pair is currently locked, use self.is_pair_locked(pair).

!!! Note Locked pairs are not persisted, so a restart of the bot, or calling /reload_conf will reset locked pairs.

!!! Warning Locking pairs is not functioning during backtesting.

Pair locking example
from freqtrade.persistence import Trade
from datetime import timedelta, datetime, timezone
# Put the above lines a the top of the strategy file, next to all the other imports
# --------

# Within populate indicators (or populate_buy):
if self.config['runmode'] in ('live', 'dry_run'):
   # fetch closed trades for the last 2 days
    trades = Trade.get_trades([Trade.pair == metadata['pair'],
                               Trade.open_date > datetime.utcnow() - timedelta(days=2),
                               Trade.is_open == False,
                ]).all()
    # Analyze the conditions you'd like to lock the pair .... will probably be different for every strategy
    sumprofit = sum(trade.close_profit for trade in trades)
    if sumprofit < 0:
        # Lock pair for 12 hours
        self.lock_pair(metadata['pair'], until=datetime.now(timezone.utc) + timedelta(hours=12))

Print created dataframe

To inspect the created dataframe, you can issue a print-statement in either populate_buy_trend() or populate_sell_trend(). You may also want to print the pair so it's clear what data is currently shown.

def populate_buy_trend(self, dataframe: DataFrame, metadata: dict) -> DataFrame:
    dataframe.loc[
        (
            #>> whatever condition<<<
        ),
        'buy'] = 1

    # Print the Analyzed pair
    print(f"result for {metadata['pair']}")

    # Inspect the last 5 rows
    print(dataframe.tail())

    return dataframe

Printing more than a few rows is also possible (simply use print(dataframe) instead of print(dataframe.tail())), however not recommended, as that will be very verbose (~500 lines per pair every 5 seconds).

Specify custom strategy location

If you want to use a strategy from a different directory you can pass --strategy-path

freqtrade trade --strategy AwesomeStrategy --strategy-path /some/directory

Derived strategies

The strategies can be derived from other strategies. This avoids duplication of your custom strategy code. You can use this technique to override small parts of your main strategy, leaving the rest untouched:

class MyAwesomeStrategy(IStrategy):
    ...
    stoploss = 0.13
    trailing_stop = False
    # All other attributes and methods are here as they
    # should be in any custom strategy...
    ...

class MyAwesomeStrategy2(MyAwesomeStrategy):
    # Override something
    stoploss = 0.08
    trailing_stop = True

Both attributes and methods may be overriden, altering behavior of the original strategy in a way you need.

Common mistakes when developing strategies

Backtesting analyzes the whole time-range at once for performance reasons. Because of this, strategy authors need to make sure that strategies do not look-ahead into the future. This is a common pain-point, which can cause huge differences between backtesting and dry/live run methods, since they all use data which is not available during dry/live runs, so these strategies will perform well during backtesting, but will fail / perform badly in real conditions.

The following lists some common patterns which should be avoided to prevent frustration:

  • don't use shift(-1). This uses data from the future, which is not available.
  • don't use .iloc[-1] or any other absolute position in the dataframe, this will be different between dry-run and backtesting.
  • don't use dataframe['volume'].mean(). This uses the full DataFrame for backtesting, including data from the future. Use dataframe['volume'].rolling(<window>).mean() instead
  • don't use .resample('1h'). This uses the left border of the interval, so moves data from an hour to the start of the hour. Use .resample('1h', label='right') instead.

Further strategy ideas

To get additional Ideas for strategies, head over to our strategy repository. Feel free to use them as they are - but results will depend on the current market situation, pairs used etc. - therefore please backtest the strategy for your exchange/desired pairs first, evaluate carefully, use at your own risk. Feel free to use any of them as inspiration for your own strategies. We're happy to accept Pull Requests containing new Strategies to that repo.

We also got a strategy-sharing channel in our Slack community which is a great place to get and/or share ideas.

Next step

Now you have a perfect strategy you probably want to backtest it. Your next step is to learn How to use the Backtesting.