MaxDrawdown

Description

The worst (most negative) drawdown ever observed since the start (or last reset()):

\[ \text{MaxDrawdown}[t] = \min_{k \le t}\ \text{Drawdown}[k] \]

Monotonically non-increasing in time. Composes Drawdown + CumMin.

Notes

• For the trailing-window equivalent see RollingMaxDrawdown.

NaN handling

Policy: ignore. A NaN in any input at index t causes the function to skip that step: output at t is NaN and internal state is unchanged. Subsequent finite samples are processed as if step t had not occurred.

Examples

Usage example

import numpy as np
import plotly.graph_objects as go
from plotly.subplots import make_subplots
from screamer import MaxDrawdown

np.random.seed(0)
price = 100 * np.exp(np.cumsum(np.random.normal(0.0005, 0.02, size=300)))
mdd = MaxDrawdown()(price)              # worst drawdown seen so far, only gets deeper

fig = make_subplots(rows=2, cols=1, shared_xaxes=True,
                    row_heights=[0.55, 0.45], vertical_spacing=0.08)
fig.add_trace(go.Scatter(y=price, mode="lines", name="price"), row=1, col=1)
fig.add_trace(go.Scatter(y=mdd, mode="lines", name="max drawdown",
                         line=dict(color="red"), fill="tozeroy"), row=2, col=1)
fig.update_layout(title="Worst drawdown so far (MaxDrawdown)",
                  margin=dict(l=20, r=20, t=60, b=20),
                  legend=dict(orientation="h", yanchor="bottom", y=1.02, xanchor="right", x=1))
fig.update_yaxes(title_text="price", row=1, col=1)
fig.update_yaxes(title_text="max drawdown", row=2, col=1)
fig.show()