macrosynergy.pnl.naive_pnl#

“Naive” profit-and-loss (PnL) calculations with basic signal options, disregarding transaction costs.

class NaivePnL(df, ret, sigs, cids=None, bms=None, start=None, end=None, blacklist=None)[source]#

Bases: object

Computes and collects illustrative PnLs with limited signal options and disregarding transaction costs.

Parameters:

df (Dataframe) – standardized DataFrame with the following necessary columns: ‘cid’, ‘xcat’, ‘real_date’ and ‘value’.
ret (str) – return category.
sigs (List[str]) – signal categories. Able to pass in multiple possible signals to the Class’ constructor and their respective vintages will be held on the instance’s DataFrame. The signals can subsequently be referenced through the self.make_pnl() method which receives a single signal per call.
cids (List[str]) – cross sections that are traded. Default is all in the dataframe.
bms (str, List[str]) – list of benchmark tickers for which correlations are displayed against PnL strategies.
start (str) – earliest date in ISO format. Default is None and earliest date in df is used.
end (str) – latest date in ISO format. Default is None and latest date in df is used.
blacklist (dict) – cross-sections with date ranges that should be excluded from the dataframe.

make_pnl(sig, sig_op='zn_score_pan', sig_add=0, sig_mult=1, sig_neg=False, pnl_name=None, rebal_freq='daily', rebal_slip=0, vol_scale=None, leverage=1.0, min_obs=261, iis=True, sequential=True, neutral='zero', thresh=None, entry_barrier=None, exit_barrier=None, winsorize_first=False, normalized_weights=False)[source]#

Calculate daily PnL and add to class instance.

Parameters:

sig (str) – name of raw signal that is basis for positioning. The signal is assumed to be recorded at the end of the day prior to position taking.
sig_op (str) – signal transformation options; must be one of ‘zn_score_pan’, ‘zn_score_cs’, ‘binary’, or ‘raw’. The default is ‘zn_score_pan’. See notes below for further details.
sig_add (float) – add a constant to the signal after initial transformation. This allows to give PnLs a long or short bias relative to the signal score. Default is 0.
sig_mult (float) – multiply a constant to the signal after initial tranformation and after sig_add has been added too. Default is 1.
sig_neg (str) – if True the PnL is based on the negative value of the transformed signal. Default is False.
pnl_name (str) – name of the PnL to be generated and stored. Default is None, i.e. a default name is given. The default name will be: ‘PNL_<signal name>[_<NEG>]’, with the last part added if sig_neg has been set to True. Previously calculated PnLs of the same name will be overwritten. This means that if a set of PnLs are to be compared, each PnL requires a distinct name.
rebal_freq (str) – re-balancing frequency for positions according to signal must be one of ‘daily’ (default), ‘weekly’, ‘monthly’, ‘quarterly’, or ‘annual’. The re-balancing is only concerned with the signal value on the re-balancing date which is delimited by the frequency chosen. Additionally, the re-balancing frequency will be applied to make_zn_scores() if used as the method to produce the raw signals.
rebal_slip (str) – re-balancing slippage in days. Default is 1 which means that it takes one day to re-balance the position and that the new positions produce PnL from the second day after the signal has been recorded.
vol_scale (bool) – ex-post scaling of PnL to annualized volatility given. This is for comparative visualization and not out-of-sample. Default is none.
leverage (float) – leverage applied to the raw signal when a vol_scale is not defined. Default is 1.0, i.e., position size is 1 or 100% of implied risk capital.
min_obs (int) – the minimum number of observations required to calculate zn_scores. Default is 252.
iis (bool) – if True (default) zn-scores are also calculated for the initial sample period defined by min_obs, on an in-sample basis, to avoid losing history.
sequential (bool) – if True (default) score parameters (neutral level and standard deviations) are estimated sequentially with concurrently available information only.
neutral (str) – method to determine neutral level. Default is ‘zero’. Alternatives are ‘mean’ and “median”.
thresh (float) – threshold value beyond which scores are winsorized, i.e. contained at that threshold. Therefore, the threshold is the maximum absolute score value that the function is allowed to produce. The minimum threshold is one standard deviation. Default is no threshold.
entry_barrier (float) – Threshold in terms of absolute signal value to enter a position in a binary strategy. This prevents binary strategies from excessive position flipping. Default is None, i.e., the binary strategy always takes its full position, no matter how small the signal value.
exit_barrier (float) – Threshold in terms of absolute signal value to exit a position in a binary strategy. In conjunction with an entry barrier, this determines the probability of position liquidations when the sign of the signal does not change. The value must be below the entry barrier. Without an entry barrier there can be no exit barrier. Default is None, which means that a position is only liquidated of the sign of the signal flips.
winsorize_first (bool) – if True, the signal is winsorized before any signal manipulation is applied. This means that the signal is clipped to a specified range before any further processing is done. Default is False, meaning that the signal is winsorized after all other transformations have been applied.
normalized_weights (bool) – if True, the PnL is computed using normalized weights, meaning that the PnL is computed as the mean of the signal-adjusted returns across all cross-sections. Default is False, meaning that the PnL is computed as the sum of the signal-adjusted returns across all cross-sections.

Notes

When sig_op = “zn_score_pan”, raw signals are transformed into zn-scores around a neutral value where pooled panel statistics are calculated.

When sig_op = “zn_score_cs”, raw signals are transformed into zn-scores around a neutral value where statistics are calculated by cross section alone.

When sig_op = “binary”, transforms signals into uniform long/shorts (1/-1) across all cross sections.

When sig_op = “raw”, no transformation is applied to the signal.

Entry and exit barriers are only applicable when the signal operation is binary.