Spaces:
Sleeping
Sleeping
File size: 4,425 Bytes
cbdae8d | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 | import pandas as pd
import plotly.graph_objects as go
from data.schema import Out
# Realised P&L is context, so it stays neutral; the forecast
# lines share one hue; exceedances use the critical status
# color plus an x marker and a legend label.
PNL_COLOR = "#b4b2a9"
VAR_COLOR = "#2a78d6"
ES_COLOR = "#184f95"
EXCEEDANCE_COLOR = "#d03b3b"
EXPECTED_COLOR = "#52514e"
def _select(forecasts, confidence):
selected = forecasts[
forecasts[Out.CONFIDENCE] == confidence
]
if selected.empty:
raise ValueError(
f"No backtest results at confidence {confidence}."
)
return selected.sort_values(Out.DATE)
def _break_at_gaps(df, gap_days):
"""
Insert an empty row after every gap longer than `gap_days`
so lines are not drawn across periods without data.
"""
if not gap_days:
return df
gaps = df[Out.DATE].diff() > pd.Timedelta(days=gap_days)
breaks = pd.DataFrame(
{Out.DATE: df.loc[gaps, Out.DATE] - pd.Timedelta(days=1)}
)
return (
pd.concat([df, breaks])
.sort_values(Out.DATE)
.reset_index(drop=True)
)
def plot_backtest_pnl(
forecasts,
confidence,
gap_days=None,
):
"""
Realised P&L against the forecast -VaR and -ES,
with exceedances marked.
"""
df = _select(forecasts, confidence)
exceeded = df[df[Out.EXCEEDANCE]]
df = _break_at_gaps(df, gap_days)
fig = go.Figure()
fig.add_trace(
go.Bar(
x=df[Out.DATE],
y=df[Out.REALIZED_PNL],
name="Realized P&L",
marker_color=PNL_COLOR,
hovertemplate="%{y:,.0f}<extra>Realized P&L</extra>",
)
)
fig.add_trace(
go.Scatter(
x=df[Out.DATE],
y=-df[Out.VAR],
mode="lines",
name=f"-VaR {confidence:.1%}",
line=dict(color=VAR_COLOR, width=2),
hovertemplate="%{y:,.0f}<extra>-VaR</extra>",
)
)
fig.add_trace(
go.Scatter(
x=df[Out.DATE],
y=-df[Out.ES],
mode="lines",
name=f"-ES {confidence:.1%}",
line=dict(color=ES_COLOR, width=2, dash="dot"),
hovertemplate="%{y:,.0f}<extra>-ES</extra>",
)
)
fig.add_trace(
go.Scatter(
x=exceeded[Out.DATE],
y=exceeded[Out.REALIZED_PNL],
mode="markers",
name=f"Exceedance ({len(exceeded)})",
marker=dict(
color=EXCEEDANCE_COLOR,
symbol="x",
size=11,
line=dict(width=1, color=EXCEEDANCE_COLOR),
),
hovertemplate="%{y:,.0f}<extra>Exceedance</extra>",
)
)
fig.update_layout(
title=(
f"Out-of-sample backtest: realized P&L vs "
f"forecast VaR ({confidence:.1%})"
),
xaxis_title="Date",
yaxis_title="P&L",
template="plotly_white",
hovermode="x unified",
bargap=0.3,
)
return fig
def plot_cumulative_exceedances(
forecasts,
confidence,
gap_days=None,
):
"""
Observed exceedances over time against the number a
correctly calibrated model would produce.
"""
df = _select(forecasts, confidence).reset_index(drop=True)
df[Out.EXCEEDANCES] = df[Out.EXCEEDANCE].astype(int).cumsum()
df[Out.EXPECTED_EXCEEDANCES] = (1 - confidence) * (df.index + 1)
df = _break_at_gaps(df, gap_days)
observed = df[Out.EXCEEDANCES]
expected = df[Out.EXPECTED_EXCEEDANCES]
fig = go.Figure()
fig.add_trace(
go.Scatter(
x=df[Out.DATE],
y=expected,
mode="lines",
name="Expected",
line=dict(color=EXPECTED_COLOR, width=2, dash="dash"),
hovertemplate="%{y:.2f}<extra>Expected</extra>",
)
)
fig.add_trace(
go.Scatter(
x=df[Out.DATE],
y=observed,
mode="lines",
name="Observed",
line=dict(color=VAR_COLOR, width=2, shape="hv"),
hovertemplate="%{y}<extra>Observed</extra>",
)
)
fig.update_layout(
title=f"Cumulative exceedances ({confidence:.1%})",
xaxis_title="Date",
yaxis_title="Exceedances",
template="plotly_white",
hovermode="x unified",
)
return fig
|