tessida's picture
Update app.py
8f5c870 verified
Raw History Blame Contribute Delete
19.4 kB
import pandas as pd
import numpy as np
import plotly.express as px
import plotly
from IPython.display import display
import ipywidgets as widgets
import gradio as gr
import plotly.graph_objects as go
df = pd.DataFrame()
bettype_list = []
def format_selection_num(value):
"""
Format selection_num for display:
- 500.0 -> "500"
- 499.5 -> "499.5"
"""
if pd.isna(value):
return ""
try:
numeric_value = float(value)
if numeric_value.is_integer():
return str(int(numeric_value))
return str(numeric_value).rstrip("0").rstrip(".")
except Exception:
return str(value)
def clean_uploaded_df(raw_df):
"""
Standardize column names and normalize selection_direction for consistent logic.
"""
raw_df.columns = raw_df.columns.str.lower()
if "selection_direction" in raw_df.columns:
raw_df["selection_direction"] = raw_df["selection_direction"].astype(str).str.strip().str.title()
return raw_df
def prepare_ui(file):
global df
df = pd.read_csv(file.name)
df = clean_uploaded_df(df)
min_date = df["placeddatetime"].min()
max_date = df["placeddatetime"].max()
bet_types = ["All"] + sorted(df["bet_type"].dropna().unique())
return min_date, max_date, gr.update(choices=bet_types, value="All")
def load_csv(file):
global df, bettype_list
if file is None:
return (
gr.update(choices=[], interactive=False),
gr.update(choices=[], interactive=False),
gr.update(choices=[], value=[], interactive=False)
)
df = pd.read_csv(file.name)
df = clean_uploaded_df(df)
market_options = sorted(df["sourcemarkettypename2"].dropna().unique())
bettype_list = sorted(df["bet_type"].dropna().unique())
return (
gr.update(choices=market_options, value=None, interactive=True),
gr.update(choices=[], value=None, interactive=False),
gr.update(choices=["All"] + bettype_list, value="All", interactive=True),
)
def filter_table(start_date, end_date, selected_bet_type):
try:
start = start_date
end = end_date
except Exception:
return pd.DataFrame(columns=[
"Event_Name", "Name", "Line", "Risk", "Bets", "Handle", "Sharp Handle"
])
mask = (df["placeddatetime"] >= start) & (df["placeddatetime"] <= end)
filtered_df = df[mask].copy()
if selected_bet_type != "All":
filtered_df = filtered_df[filtered_df["bet_type"] == selected_bet_type].copy()
if filtered_df.empty:
return pd.DataFrame(columns=[
"Event_Name", "Name", "Line", "Risk", "Bets", "Handle", "Sharp Handle"
])
# Build display line from selection_direction + selection_num
filtered_df["line_display"] = (
filtered_df["selection_direction"].astype(str).str.strip().str.title()
+ " "
+ filtered_df["selection_num"].apply(format_selection_num)
)
# Main aggregation using lineid
agg_df = (
filtered_df.groupby("lineid", as_index=False)
.agg(
LineID=("lineid", "first"),
Event_Name=("eventname", "first"),
Name=("player_name", "first"),
Line=("line_display", "first"),
Risk=("pot_risk", "sum"),
Handle=("handle", "sum"),
Bets=("geallbetid", "nunique")
)
)
# Sharp Handle
target_brackets = ["Marker", "No Value", "Concerning"]
no_value_df = filtered_df[
filtered_df["customerbracket"].isin(target_brackets)
]
handle_no_value_df = (
no_value_df.groupby("lineid", as_index=False)
.agg(
line_id=("lineid", "first"),
handle_no_value=("handle", "sum")
)
)
agg_df = agg_df.merge(handle_no_value_df, on="lineid", how="left")
agg_df["handle_no_value"] = agg_df["handle_no_value"].fillna(0)
# Select top 100 by most negative pot risk
top_10 = agg_df.sort_values("Risk").head(100)
top_10["Sharp Handle"] = top_10.apply(
lambda row: f"${int(row['handle_no_value']):,} ({round((row['handle_no_value'] / row['Handle']) * 100)}%)"
if row["handle_no_value"] > 0 and row["Handle"] != 0 else "$0 (0%)",
axis=1
)
top_10["Risk"] = top_10["Risk"].round(0).apply(
lambda x: f"${int(x):,}" if x >= 0 else f"-${abs(int(x)):,}"
)
top_10["Handle"] = top_10["Handle"].round(0).apply(
lambda x: f"${int(x):,}" if x >= 0 else f"-${abs(int(x)):,}"
)
top_10["handle_no_value"] = top_10["handle_no_value"].round(0)
return top_10[[
"Event_Name", "Name", "Line", "Risk", "Bets", "Handle", "Sharp Handle"
]]
def get_players(market_type):
if market_type is None or df.empty:
return []
players = df[df["sourcemarkettypename2"] == market_type]["player_name"].dropna().unique()
return sorted(players)
def update_players(market_type):
players = get_players(market_type)
return gr.update(choices=players, value=None, interactive=True)
def generate_stacked_bar(data, x, y, category, title, y_label, normalize=False):
segment_colors = {
"New": "white",
"Normal": "gray",
"Retail": "black",
"Outstanding": "green",
"VIP": "purple",
"Monitored": "yellow",
"Live Advantage": "orange",
"Sharp": "red",
"No Value": "#8B0000",
"Marker": "blue",
"Concerning": "pink",
"Other": "brown"
}
fig = go.Figure()
if normalize:
data = data.copy()
totals = data.groupby([x, "direction"])[y].transform("sum").replace(0, 1)
data[y] = data[y] / totals * 100
data[y] = data.apply(
lambda row: -abs(row[y]) if row["direction"] == "Under" else abs(row[y]),
axis=1
)
for cat in sorted(data[category].dropna().unique()):
sub_df = data[data[category] == cat]
fig.add_trace(go.Bar(
x=sub_df[x],
y=sub_df[y],
name=cat,
marker_color=segment_colors.get(cat, "gray")
))
fig.add_hline(y=0, line_dash="dash", line_color="gray", opacity=0.6)
fig.add_annotation(
xref="paper", yref="paper",
x=-0.1, y=1.05,
text="Over",
showarrow=False,
font=dict(size=12, color="green"),
align="left"
)
fig.add_annotation(
xref="paper", yref="paper",
x=-0.1, y=-0.05,
text="Under",
showarrow=False,
font=dict(size=12, color="red"),
align="left"
)
line_levels = sorted(data[x].dropna().unique(), key=lambda v: float(v))
fig.update_layout(
barmode="relative",
title=title,
xaxis_title=x,
yaxis_title=y_label + (" (%)" if normalize else ""),
yaxis_tickprefix=("%" if normalize else ""),
xaxis_tickangle=-45,
bargap=0.25,
autosize=True,
margin=dict(l=40, r=20, t=60, b=60),
xaxis=dict(
categoryorder="array",
categoryarray=line_levels,
automargin=True
),
yaxis=dict(automargin=True),
dragmode=False,
modebar_remove=[
"zoom", "zoomIn", "zoomOut", "pan", "select", "lasso2d", "autoScale2d"
]
)
return fig
def generate_charts(market_type, player_name, bettype_filters, normalize):
if not market_type or not player_name or df.empty:
return None, None, pd.DataFrame(), pd.DataFrame()
filtered = df[
(df["sourcemarkettypename2"] == market_type) &
(df["player_name"] == player_name)
].copy()
if bettype_filters and bettype_filters != "All":
filtered = filtered[filtered["bet_type"] == bettype_filters].copy()
if filtered.empty:
return None, None, pd.DataFrame(), pd.DataFrame()
# Use selection_num directly as line.
# Use selection_direction directly as direction.
filtered["line"] = filtered["selection_num"]
filtered["direction"] = filtered["selection_direction"].astype(str).str.strip().str.title()
filtered.dropna(subset=["direction", "line"], inplace=True)
summary = (
filtered
.groupby(["line", "direction", "customerbracket"], as_index=False)
.agg(
pot_risk=("pot_risk", "sum"),
bet_count=("geallbetid", "nunique")
)
)
summary["pot_risk"] = summary.apply(
lambda row: -row["pot_risk"] if row["direction"] == "Over" else row["pot_risk"],
axis=1
)
summary["bet_count"] = summary.apply(
lambda row: -row["bet_count"] if row["direction"] == "Under" else row["bet_count"],
axis=1
)
fig_risk = generate_stacked_bar(
summary,
"line",
"pot_risk",
"customerbracket",
f"Liability for {player_name} - {market_type}",
"Potential Risk",
normalize
)
fig_bets = generate_stacked_bar(
summary,
"line",
"bet_count",
"customerbracket",
f"Bet Count for {player_name} - {market_type}",
"Bet Count",
normalize
)
table_df = (
filtered
.assign(direction=filtered["direction"].fillna(""))
.groupby(["line", "direction"], as_index=False)
.agg(
handle=("handle", "sum"),
potential_risk=("pot_risk", "sum"),
average_odds=("odds", "mean"),
bet_count=("geallbetid", "nunique")
)
)
table_df["average_odds"] = table_df["average_odds"].apply(lambda x: f"{x:.2f}")
table_df["bet_count"] = table_df["bet_count"].round(0).astype(int)
table_df["handle"] = table_df["handle"].round(0).astype(int).apply(
lambda x: f"${x:,}"
)
table_df["potential_risk"] = table_df["potential_risk"].round(0).astype(int).apply(
lambda x: f"${x:,}" if x >= 0 else f"-${abs(int(round(x))):,}"
)
table_df.columns = ["Line", "Selection", "Handle", "Risk", "Avg. Odds", "Bet Count"]
# No change requested here.
# Net liability table continues to use selection_num directly.
net_lines = sorted(filtered["selection_num"].dropna().unique())
net_rows = []
for idx, line in enumerate(net_lines):
under_risk = filtered[
(filtered["selection_direction"] == "Under") &
(filtered["selection_num"] >= line)
]["pot_risk"].sum()
over_recoup = filtered[
(filtered["selection_direction"] == "Over") &
(filtered["selection_num"] >= line)
]["handle"].sum()
under_net = under_risk + over_recoup
over_risk = filtered[
(filtered["selection_direction"] == "Over") &
(filtered["selection_num"] <= line)
]["pot_risk"].sum()
under_recoup = filtered[
(filtered["selection_direction"] == "Under") &
(filtered["selection_num"] <= line)
]["handle"].sum()
over_net = over_risk + under_recoup
# Interval exposure: liability if player finishes between this line and the next
if idx < len(net_lines) - 1:
next_line = net_lines[idx + 1]
interval_range = f"{format_selection_num(line)}–{format_selection_num(next_line)}"
over_win = filtered[
(filtered["selection_direction"] == "Over") &
(filtered["selection_num"] <= line)
]["pot_risk"].sum()
under_win = filtered[
(filtered["selection_direction"] == "Under") &
(filtered["selection_num"] >= next_line)
]["pot_risk"].sum()
over_handle_loss = filtered[
(filtered["selection_direction"] == "Over") &
(filtered["selection_num"] > line)
]["handle"].sum()
under_handle_loss = filtered[
(filtered["selection_direction"] == "Under") &
(filtered["selection_num"] < next_line)
]["handle"].sum()
interval_net = over_win + under_win + over_handle_loss + under_handle_loss
interval_exposure = (
f"${int(round(interval_net)):,}"
if interval_net >= 0
else f"-${abs(int(round(interval_net))):,}"
)
else:
interval_range = None
interval_exposure = None
net_rows.append({
"line": line,
"under_net_liability": (
f"${int(round(under_net)):,}"
if under_net >= 0
else f"-${abs(int(round(under_net))):,}"
),
"over_net_liability": (
f"${int(round(over_net)):,}"
if over_net >= 0
else f"-${abs(int(round(over_net))):,}"
),
"interval_range": interval_range,
"interval_exposure": interval_exposure
})
net_liability_df = pd.DataFrame(net_rows)
net_liability_df.columns = [
"Line", "Under Net Risk", "Over Net Risk", "Interval Range", "Interval Exposure"
]
return fig_risk, fig_bets, table_df, net_liability_df
with gr.Blocks(
title="Top NFL Outright Markets by Liability",
css="""
#top-table-container {
height: 400px; /* adjust to taste */
overflow-y: auto;
}
"""
) as demo:
file_input = gr.File(label="Upload CSV File", file_types=[".csv"])
gr.Markdown("Upload a CSV file, then use the controls below to filter bets by date and bet type.")
gr.Markdown("### Top NFL Outright Markets by Line Liability")
with gr.Row():
start_input = gr.Textbox(label="Start Date (YYYY-MM-DD)")
end_input = gr.Textbox(label="End Date (YYYY-MM-DD)")
bet_type_input = gr.Dropdown(label="Bet Type", choices=["All"], value="All")
# ✅ Wrapped in scrollable container
with gr.Group(elem_id="top-table-container"):
table = gr.Dataframe(
label="Top 10 Liability Summary",
interactive=False
)
file_input.change(
fn=prepare_ui,
inputs=file_input,
outputs=[start_input, end_input, bet_type_input]
)
start_input.change(
fn=filter_table,
inputs=[start_input, end_input, bet_type_input],
outputs=[table]
)
end_input.change(
fn=filter_table,
inputs=[start_input, end_input, bet_type_input],
outputs=[table]
)
bet_type_input.change(
fn=filter_table,
inputs=[start_input, end_input, bet_type_input],
outputs=[table]
)
gr.Markdown("### 🏈 NFL Outright Liability Dashboard")
with gr.Row():
market_dropdown = gr.Dropdown(choices=[], label="Market Type", interactive=False)
player_dropdown = gr.Dropdown(choices=[], label="Player Name", interactive=False)
bettype_dropdown = gr.Dropdown(choices=["All"], value="All", label="Bet Types", interactive=False)
submit_button = gr.Button("Submit", variant="primary")
with gr.Row():
chart_risk = gr.Plot(label="Liability (pot_risk)")
chart_bets = gr.Plot(label="Bet Count")
normalize_toggle = gr.Checkbox(label="Normalize bar height (percent scale)", value=False)
with gr.Row():
data_table = gr.DataFrame(
label="Liability Summary",
interactive=False
)
with gr.Row():
net_liability_table = gr.DataFrame(
label="Net Liability by Line",
interactive=False
)
file_input.change(
fn=load_csv,
inputs=file_input,
outputs=[market_dropdown, player_dropdown, bettype_dropdown]
)
market_dropdown.change(
fn=update_players,
inputs=market_dropdown,
outputs=player_dropdown
)
submit_button.click(
fn=generate_charts,
inputs=[market_dropdown, player_dropdown, bettype_dropdown, normalize_toggle],
outputs=[chart_risk, chart_bets, data_table, net_liability_table]
)
normalize_toggle.change(
fn=generate_charts,
inputs=[market_dropdown, player_dropdown, bettype_dropdown, normalize_toggle],
outputs=[chart_risk, chart_bets, data_table, net_liability_table]
)
# with gr.Blocks(title="Top NFL Outright Markets by Liability") as demo:
# file_input = gr.File(label="Upload CSV File", file_types=[".csv"])
# gr.Markdown("Upload a CSV file, then use the controls below to filter bets by date and bet type.")
# gr.Markdown("### Top NFL Outright Markets by Line Liability")
# with gr.Row():
# start_input = gr.Textbox(label="Start Date (YYYY-MM-DD)")
# end_input = gr.Textbox(label="End Date (YYYY-MM-DD)")
# bet_type_input = gr.Dropdown(label="Bet Type", choices=["All"], value="All")
# #table = gr.Dataframe(label="Top 10 Liability Summary", interactive=False, height=150 )
# with gr.Group(elem_id="top-table-container"):
# table = gr.Dataframe(label="Top 10 Liability Summary", interactive=False)
# file_input.change(
# fn=prepare_ui,
# inputs=file_input,
# outputs=[start_input, end_input, bet_type_input]
# )
# start_input.change(
# fn=filter_table,
# inputs=[start_input, end_input, bet_type_input],
# outputs=[table]
# )
# end_input.change(
# fn=filter_table,
# inputs=[start_input, end_input, bet_type_input],
# outputs=[table]
# )
# bet_type_input.change(
# fn=filter_table,
# inputs=[start_input, end_input, bet_type_input],
# outputs=[table]
# )
# gr.Markdown("### 🏈 NFL Outright Liability Dashboard")
# with gr.Row():
# market_dropdown = gr.Dropdown(choices=[], label="Market Type", interactive=False)
# player_dropdown = gr.Dropdown(choices=[], label="Player Name", interactive=False)
# bettype_dropdown = gr.Dropdown(choices=["All"], value="All", label="Bet Types", interactive=False)
# submit_button = gr.Button("Submit", variant="primary")
# with gr.Row():
# chart_risk = gr.Plot(label="Liability (pot_risk)")
# chart_bets = gr.Plot(label="Bet Count")
# normalize_toggle = gr.Checkbox(label="Normalize bar height (percent scale)", value=False)
# with gr.Row():
# data_table = gr.DataFrame(
# label="Liability Summary",
# interactive=False
# )
# with gr.Row():
# net_liability_table = gr.DataFrame(
# label="Net Liability by Line",
# interactive=False
# )
# file_input.change(
# fn=load_csv,
# inputs=file_input,
# outputs=[market_dropdown, player_dropdown, bettype_dropdown]
# )
# market_dropdown.change(
# fn=update_players,
# inputs=market_dropdown,
# outputs=player_dropdown
# )
# submit_button.click(
# fn=generate_charts,
# inputs=[market_dropdown, player_dropdown, bettype_dropdown, normalize_toggle],
# outputs=[chart_risk, chart_bets, data_table, net_liability_table]
# )
# normalize_toggle.change(
# fn=generate_charts,
# inputs=[market_dropdown, player_dropdown, bettype_dropdown, normalize_toggle],
# outputs=[chart_risk, chart_bets, data_table, net_liability_table]
# )
demo.launch()