---
title: "Deal Screener"
description: "AI adoption signals for investor due diligence — sector benchmarks, momentum leaders, and value opportunities."
format:
html:
page-layout: full
toc: false
execute:
echo: false
warning: false
---
```{python}
import pandas as pd
import numpy as np
from pathlib import Path
from itables import show
import itables.options as opt
opt.maxBytes = 0
REPO = Path(".")
signals_path = REPO / "data" / "research" / "signals.csv"
panel_path = REPO / "data" / "research" / "ai_adoption_panel.csv"
universe_path = REPO / "data" / "universes" / "comparables_universe.csv"
# ── Build signals if not pre-computed ─────────────────────────────
if signals_path.exists():
df = pd.read_csv(signals_path)
else:
# Fallback: compute on the fly
import sys
sys.path.insert(0, str(REPO))
from research.alarms import build_signals_table
df = build_signals_table()
from IPython.display import display, HTML
# Signal badge colors
BADGE_COLORS = {
"AI_LEADER": ("#1a6b3c", "#d4edda", "AI Leader"),
"AI_MOMENTUM": ("#155a8a", "#d1ecf1", "Momentum"),
"AI_LAGGARD": ("#856404", "#fff3cd", "Laggard"),
"VALUE_PLUS_AI": ("#4a235a", "#e8daef", "Value + AI"),
"RISING_STAR": ("#7d3c00", "#fde8d8", "Rising Star"),
}
def render_badges(signal_str):
if not signal_str or pd.isna(signal_str):
return ""
badges = ""
for s in str(signal_str).split("|"):
s = s.strip()
if s in BADGE_COLORS:
fg, bg, label = BADGE_COLORS[s]
badges += (
f'{label}'
)
return badges
# Summary counts
all_sigs = "|".join(df["signals"].fillna("")).split("|")
n_leader = all_sigs.count("AI_LEADER")
n_momentum = all_sigs.count("AI_MOMENTUM")
n_laggard = all_sigs.count("AI_LAGGARD")
n_value = all_sigs.count("VALUE_PLUS_AI")
n_rising = all_sigs.count("RISING_STAR")
n_coverage = int((df.get("tier", pd.Series(dtype=str)) == "coverage").sum())
```
::: {.callout-note appearance="minimal"}
**Signals are derived from SEC 10-K NLP analysis.** They reflect AI adoption intensity, not stock price performance. See [Research Methodology](assumptions.html) for full signal definitions. Not investment advice.
:::
## Signal Overview
```{python}
cards_html = f"""
{n_leader}
AI Leaders
Top 20% of sector
{n_momentum}
Momentum
Score rising fast
{n_laggard}
Laggards
Bottom 25% of sector
{n_value}
Value + AI
Momentum + cheap valuation
{n_rising}
Rising Stars
Coverage cos. +10pts YoY
"""
display(HTML(cards_html))
```
## Signal Definitions
| Signal | Criteria | Investor Read |
|--------|----------|---------------|
| **AI Leader** | Top 20% of sector by AI Adoption Score | Established AI moat; benchmark for sector |
| **AI Momentum** | YoY score change > sector 75th percentile | AI investment accelerating; may signal emerging moat |
| **AI Laggard** | Bottom 25% of sector by AI Adoption Score | Transformation risk **or** acquisition opportunity |
| **Value + AI** | AI Momentum + EV/EBITDA below sector median | Under-appreciated AI adopter — core deal screen signal |
| **Rising Star** | Coverage company with ≥10pt YoY gain | Mid-market companies catching up; watch list candidates |
## Deal Screen
Use the search box to filter by company, sector, or signal. Sort any column.
```{python}
# Build display table
display_cols = {
"ticker": "Ticker",
"name": "Company",
"industry": "Sector",
"tier": "Tier",
"ai_adoption_score": "AI Score",
"ai_score_yoy_change": "YoY Δ",
"ai_adoption_score_sector_pctile": "Sector %ile",
"ev_to_ebitda": "EV/EBITDA",
"operating_margin_calc": "Op Margin",
"revenue_growth": "Rev Growth",
"signals": "Signals",
}
available = {k: v for k, v in display_cols.items() if k in df.columns}
screen = df[list(available.keys())].copy()
screen.columns = list(available.values())
# Format numerics
for col, fmt in [("AI Score", 1), ("YoY Δ", 1)]:
if col in screen.columns:
screen[col] = screen[col].round(fmt)
if "Sector %ile" in screen.columns:
screen["Sector %ile"] = (screen["Sector %ile"] * 100).round(0).astype("Int64")
if "EV/EBITDA" in screen.columns:
screen["EV/EBITDA"] = screen["EV/EBITDA"].round(1)
if "Op Margin" in screen.columns:
screen["Op Margin"] = (screen["Op Margin"] * 100).round(1)
if "Rev Growth" in screen.columns:
screen["Rev Growth"] = (screen["Rev Growth"] * 100).round(1)
# Capitalise Tier
if "Tier" in screen.columns:
screen["Tier"] = screen["Tier"].str.capitalize()
# Sort: signal count desc, AI score desc
screen = screen.sort_values(
["AI Score"], ascending=False
).reset_index(drop=True)
show(
screen,
caption="AI Deal Screen — latest 10-K scoring",
scrollX=True,
pageLength=20,
classes="display compact stripe",
columnDefs=[
{"width": "75px", "targets": [0]},
{"width": "170px", "targets": [1]},
{"width": "110px", "targets": [2]},
{"width": "75px", "targets": [3]},
],
)
```
## AI Leaders by Sector
```{python}
if not df.empty and "ai_adoption_score" in df.columns and "industry" in df.columns:
top3 = (
df.sort_values("ai_adoption_score", ascending=False)
.groupby("industry")
.head(3)[["industry", "ticker", "name", "ai_adoption_score", "ai_score_yoy_change", "signals"]]
.sort_values(["industry", "ai_adoption_score"], ascending=[True, False])
)
COLORS = {
"Technology": "#2980b9", "Financials": "#16a085",
"Industrials": "#8e44ad", "Healthcare": "#c0392b",
"Consumer": "#d35400", "Real Estate": "#7f8c8d", "Energy": "#f39c12",
}
def sector_card(sector, grp):
color = COLORS.get(sector, "#555")
rows = ""
for _, r in grp.iterrows():
yoy = r.get("ai_score_yoy_change", np.nan)
yoy_str = f"+{yoy:.1f}" if pd.notna(yoy) and yoy > 0 else (f"{yoy:.1f}" if pd.notna(yoy) else "—")
yoy_color = "#1a6b3c" if yoy_str.startswith("+") else "#c0392b"
badges = render_badges(r.get("signals", ""))
rows += f"""
{r['ticker']}
{r['name']}
{r['ai_adoption_score']:.1f}
{yoy_str}
{badges}
"""
return f"""
{sector}
Ticker
Company
AI Score
YoY Δ
Signals
{rows}
"""
sectors = sorted(top3["industry"].unique())
left_sectors = sectors[::2] # even indices → left column
right_sectors = sectors[1::2] # odd indices → right column
left_html = "".join(sector_card(s, top3[top3["industry"] == s]) for s in left_sectors)
right_html = "".join(sector_card(s, top3[top3["industry"] == s]) for s in right_sectors)
display(HTML(f"""
{left_html}
{right_html}
"""))
```
---
*AI Adoption Scores extracted from SEC EDGAR 10-K filings via NLP. Updated monthly via automated pipeline. Not investment advice.*