Compare

Compare

Overview

Four-step workflow: frame → fetch → table → return. Act on defaults immediately. Do NOT ask before producing the first result. The user will revise if needed.

---

DEFAULTS — apply these unless the user states otherwise

Dimension	Default
Metrics	Revenue, Gross Margin, Net Income (3-row quick table)
Period	Latest single quarter (quarterlyReports[0])
Unit	$M — auto-upgrade to $B if all values exceed 1 000
Targets	Tickers named in the message; if "vs peers" → use peers list from <stock_context>
Section heading	## Peer Comparison — {Metrics} (Latest Quarter, $M)

Only deviate from a default when the user explicitly names something different. Never ask about defaults.

---

Step 1 — Frame (thought only, no tool call)

In your reasoning state:

1. Targets list (from message or peers in stock_context).
2. Metric(s) — apply default if not specified.
3. Period — apply default if not specified.
4. Which tool each metric requires (see Data Source Map in Step 2).

Only call request_user_input when targets cannot be determined at all (e.g. user says "compare" with no tickers and no peers in context).

---

Step 2 — Collect Data for Every Target

Fetch sequentially: one tool call per target. Check <stock_context> / <acquired_data> first — skip the call if the value is already there.

Critical: Use the exact tool specified in the data source map below for each metric family. The defaults (Revenue, Gross Margin, Net Income) ALL require INCOME_STATEMENT, not OVERVIEW.

Data Source Map

Metric family	Primary tool	Key field(s)
Revenue, Gross Profit, Operating Income, Net Income, EBITDA	INCOME_STATEMENT(symbol=…)	totalRevenue, grossProfit, operatingIncome, netIncome
Free Cash Flow, Operating Cash Flow, CapEx	CASH_FLOW(symbol=…)	operatingCashflow, capitalExpenditures
EPS (actual + estimate), beat/miss	EARNINGS(symbol=…)	reportedEPS, estimatedEPS, surprise
Total Debt, Cash, Book Value, Current Ratio	BALANCE_SHEET(symbol=…)	totalDebt, cashAndCashEquivalentsAtCarryingValue, totalShareholderEquity
Market price, P/E, Market Cap, Beta	OVERVIEW(symbol=…)	MarketCapitalization, PERatio, Beta, 52WeekHigh, 52WeekLow
Latest price / daily return	GLOBAL_QUOTE(symbol=…)	price, changePercent

Fallback chain

1. <stock_context> / <acquired_data> — use if present.
2. Primary tool from the map above.
3. On failure: CASH_FLOW for FCF-adjacent; OVERVIEW for market metrics.
4. All tools fail for a target → cell = N/A. Never abort the whole comparison.

Rules

• One tool call at a time, no preamble.
• quarterlyReports for quarterly; annualReports for annual.
• Normalise all values to the same unit before building the table.

---

Step 3 — Build the Summary Table

Construct a clean markdown comparison table.

Table format

| Metric | {Target A} | {Target B} | {Target C} | Notes |
|---|---|---|---|---|
| Revenue (TTM, $M) | 123.4 | 456.7 | 89.0 | |
| Gross Margin | 42% | 61% | 38% | |
| Net Income (TTM, $M) | (12.1) | 34.5 | (5.6) | Losses in () |
| EPS (last Q) | (0.08) | 0.22 | (0.15) | |

Formatting rules

• One row per metric, one column per target.
• Use the same period for every target in a given row. If a target has no data for that period, show N/A.
• Losses and negatives in parentheses: (12.1).
• Percentages to one decimal place: 42.3%.
• Dollar values in $M (or $B if all values > 1000) to one decimal place.
• Multiples to one decimal place: 18.4x.
• Add a Notes column only when a cell needs qualification (e.g. "FY estimate", "TTM").
• Do NOT add a prose paragraph restating the numbers — the table is self-contained.
• Add a single bold callout line below the table identifying the standout finding: **Standout:** {Target X} leads on {metric} by {magnitude} vs peers.

Section heading

Place the table under:

## Peer Comparison — {Metric(s)} ({Period})

Optional: add a multi-series line chart

If the comparison is time-series (multiple periods, not a single snapshot), append a chart-line multi-series block immediately after the table using the format from chart-writeup Step 5 → Multi-series. Use null for missing periods.

Example for quarterly revenue across 4 peers:

{
  "title": "Quarterly Revenue — Peer Comparison ($M)",
  "unit": "$M",
  "labels": ["2024-Q1", "2024-Q2", "2024-Q3", "2024-Q4"],
  "series": [
    { "label": "LAZR", "values": [20.97, 16.45, 15.49, 22.48] },
    { "label": "INVZ", "values": [null,  null,  null,  6.03]  },
    { "label": "OUST", "values": [null,  null,  null,  30.09] }
  ]
}

---

Step 4 — Return Results

Populate the output JSON:

• thesis: include the full updated thesis with the new ## Peer Comparison section appended. Preserve all prior content.
• chat: one sentence stating what was compared, how many targets, and the key takeaway.
• assessment: one sentence describing what was just done.
• skill: "compare"

When data is missing for a target

• Mark that cell N/A — do NOT omit the row or the target column.
• Mention missing targets briefly in chat only: "AEVA data unavailable from API."
• Never add disclaimers, caveats, or "Note:" blocks inside the thesis table.

---

Quick Reference — Common Comparison Patterns

User asks	Targets	Metric	Tool
"compare revenue"	named tickers	totalRevenue quarterly	INCOME_STATEMENT × N
"benchmark margins"	named tickers	grossProfit / totalRevenue	INCOME_STATEMENT × N
"how does EPS compare"	named tickers	reportedEPS	EARNINGS × N
"compare FCF"	named tickers	operatingCashflow − capitalExpenditures	CASH_FLOW × N
"vs peers"	ticker + FMP peers from <stock_context>	as specified	appropriate tool × N
"sector comparison"	tickers from <stock_context> peers list	as specified	appropriate tool × N