Jesse

Utilities

Risk management and statistic calculations are essential for a successful algo trading career.

Hence, Jesse offers a collection of commonly used utility functions that life easier for quants.

anchor_timeframe

Returns the anchor timeframe. Useful for writing dynamic strategies using multiple timeframes.

anchor_timeframe(timeframe)

Properties:

  • timeframe: str

Return Type: str

Example:

bigger_timeframe = anchor_timeframe('1h') # prints '4h'

are_cointegrated

Uses unit-root test on residuals to test for a cointegrated relationship between two price return series.

TIP

Notice that for the formula to make sense price_returns_1 and price_returns_2 must be "price returns" and not the mere prices of the two assets. Hence you need to convert your asset prices to returns using the prices_to_returns utility.

The cutoff parameter points to the p-value threshold used in the formula.

are_cointegrated(
    price_returns_1: np.ndarray, price_returns_2: np.ndarray, cutoff=0.05
) -> bool

Properties:

  • price_returns_1: np.ndarray
  • price_returns_1: np.ndarray
  • cutoff: float | default=0.05

Return Type: bool

crossed

Helper for the detection of crosses

crossed(series1, series2, direction=None, sequential=False)

Properties:

  • series1: np.ndarray
  • series2: float, int, np.ndarray
  • direction: str - default: None - above or below

Return Type: bool | np.ndarray

combinations_without_repeat

Creates an array containing all combinations of the passed arrays individual values without repetitions. Useful for the optimization mode.

combinations_without_repeat(a: np.ndarray, n: int = 2) -> np.ndarray

Properties:

  • a: np.ndarray
  • n: int - default: 2

Return Type: np.ndarray

estimate_risk

Estimates the risk per share

estimate_risk(entry_price, stop_price)

Properties:

  • entry_price: float
  • stop_price: float

Return Type: float

kelly_criterion

Returns the Kelly Criterionopen in new window.

kelly_criterion(win_rate, ratio_avg_win_loss)

Properties:

  • win_rate: float
  • ratio_avg_win_loss: float

Return Type: float

limit_stop_loss

Limits the stop-loss price according to the max allowed risk percentage. (How many percent you're OK with the price going against your position)

limit_stop_loss(entry_price, stop_price, trade_type, max_allowed_risk_percentage)

Properties:

  • entry_price: float
  • stop_price: float
  • trade_type: str
  • max_allowed_risk_percentage: float

Return Type: float

numpy_candles_to_dataframe

Helper to convert numpy to financial dataframe

numpy_candles_to_dataframe(candles: np.ndarray, name_date="date", name_open="open", name_high="high",
                               name_low="low", name_close="close", name_volume="volume")

Properties:

  • candles: np.ndarray
  • name_date: str
  • name_open: str
  • name_high: str
  • name_low: str
  • name_close: str
  • name_volume: str

Return Type: pd.DataFrame

qty_to_size

Converts a quantity to its corresponding position-size. Example: Requesting 2 shares at the price of $50 would return $100.

qty_to_size(qty, price)

Properties:

  • qty: float
  • price: float

Return Type: float

risk_to_qty

Calculates the quantity, based on the percentage of the capital you're willing to risk per trade.

TIP

This is probably the most important helper function that you're going to need in your strategies. Those of you whom are familiar with compounding risk would love this function.

We made a websiteopen in new window for you just to play with this simple but important formula.

WARNING

There might be situations where this helper returns a qty exceeding the available capital leading to an exception. The reason for this is a very close stop loss (often due to the usage of the ATR). You can check this with the calculator above. That's not a error, but expected behavior of the formula. You might want to add a logic limiting the qty to a maximum percentage of the capital.

risk_to_qty(capital, risk_per_capital, entry_price, stop_loss_price, precision=3, fee_rate=0)

Properties:

  • capital: float
  • risk_per_capital: float
  • entry_price: float
  • stop_loss_price: float
  • precision: int - default: 3
  • fee_rate: float - default: 0

Return Type: float

Example:

def go_long(self):
    # risk 1% of the capital($10000) for a trade entering at $100 with the stop-loss at $80
    risk_perc = 1
    entry = 100
    stop = 80
    profit = 150
    capital = 10000
    # or we could access capital dynamically:
    capital = self.balance
    qty = utils.risk_to_qty(capital, risk_perc, entry, stop)

    self.buy = qty, entry
    self.stop_loss = qty, stop
    self.take_profit = qty, profit

In real trading, you usually need to include the exchange fee in qty calculation to make sure you don't spend more than the existing capital (in which case Jesse would raise an error):

# so instead of 
qty = utils.risk_to_qty(capital, risk_perc, entry, stop)

# it's better to do
qty = utils.risk_to_qty(capital, risk_perc, entry, stop, self.fee_rate)

See Also: fee_rate

risk_to_size

Calculates the size of the position based on the amount of risk percentage you're willing to take.

risk_to_size(capital_size, risk_percentage, risk_per_qty, entry_price)

Properties:

  • capital_size: float
  • risk_percentage: float
  • risk_per_qty: float
  • entry_price: float

Return Type: float

signal_line

Returns the moving average of the series. Useful to create so called signal lines of indicators.

signal_line(series, period=10, matype=0)

Properties:

  • series: np.array
  • period: int - default = 10
  • matype: int - default = 0

See hereopen in new window for available matypes

Return Type: np.array

size_to_qty

Converts a position-size to the corresponding quantity. Example: Requesting $100 at the price of $50 would return 2.

size_to_qty(position_size, price, precision=3, fee_rate=0)

Properties:

  • price: float
  • position_size: float
  • precision: int - default: 3
  • fee_rate: float - default: 0

Return Type: float

streaks

Returns the streaks of the series. A positive number stands for a positive streak and a negativ number for a negative streak. By default it uses the first discrete difference.

streaks(series: np.array, use_diff=True) -> np.array

Properties:

  • series: np.array
  • use_diff: bool

Return Type: np.array[bool]

strictly_decreasing

Returns whether a series in strictly decreasing or not.

strictly_increasing(series, lookback)

Properties:

  • series: np.array
  • lookback: int

Return Type: bool

strictly_increasing

Returns whether a series in strictly increasing or not.

strictly_increasing(series, lookback)

Properties:

  • series: np.array
  • lookback: int

Return Type: bool

subtract_floats

Subtracts two floats without the rounding issue in Python

subtract_floats(float1: float, float2: float) -> float

Properties:

  • float1: float
  • float2: float

Return Type: float

sum_floats

Sums two floats without the rounding issue in Python

sum_floats(float1: float, float2: float) -> float

Properties:

  • float1: float
  • float2: float

Return Type: float

prices_to_returns

Converts a series of asset prices to returns.

In case you're wondering why you need to use price returns for price series analysis instead of the price values, check out this answer on Quant Stackexchangeopen in new window.

TIP

Notice that the first return value for the first index cannot be calculated hence it equals nan.

prices_to_returns(price_series: np.ndarray) -> np.ndarray

Properties:

  • price_series: np.ndarray

Return Type: np.ndarray

wavelet_denoising

Denoises / filters timeseries data. First deconstructs and then reconstructs based on a threshold.

TIP

Based on PyWaveletsopen in new window. See its docs for more information. It's recommended to consider only haar, db, sym, coif wavelet basis functions, as these are relatively suitable for financial data.

wavelet_denoising(raw: np.ndarray, wavelet: str ='haar', level: int = 1, mode: str = 'symmetric', smoothing_factor: float = 0, threshold_mode: str = 'hard') -> np.ndarray

Properties:

Return Type: np.ndarray

z_score

A Z-score is a numerical measurement that describes how many standard deviations far away the data is comparing to the mean of the data.

TIP

Notice that for the formula to make sense price_returns must be "price returns" and not the mere prices of the two assets. Hence you need to convert your asset prices to returns using the prices_to_returns utility.

z_score(price_returns: np.ndarray) -> np.ndarray

Properties:

  • price_returns: np.ndarray

Return Type: np.ndarray