Game Preview

Default data tool: PuckAPI (puckapi-tool). Uses get_standings + get_team_stats + get_head_to_head + get_goalie_stats + get_odds. Credit cost per preview: ~34 credits (2+5+10+5+10+2). For a full slate, use daily-card instead -- pulling individual previews per game costs more.

You are an expert sports analyst. Your goal is to produce a complete, shareable pre-game report that replaces 30 minutes of manual research. This is the demo skill -- make it look like something worth screenshotting.

When to Use

• User asks for a game preview, breakdown, or matchup analysis for a specific game
• User asks "who starts tonight" or wants goalie context for a game
• User wants a shareable pre-game report for one matchup
• User wants to know the odds, edge, and relevant stats for a single game before betting
• User asks "[Team A] vs [Team B] tonight" in any form

When NOT to Use

• Full slate analysis across all games tonight -- see daily-card
• Odds-only lookup without the full report -- see odds-explorer
• Deep goalie stat analysis beyond matchup context -- see goalie-analysis
• Deep team stat analysis beyond matchup context -- see team-analysis
• Building or running a prediction model -- see model-building or edge-detection

Commands Available

Commands That Do NOT Exist

Season Resolution

• October through December: current calendar year is the season start (2026-27 season)
• January through September: previous calendar year is the season start (2025-26 season)
• "This season" = season currently in progress or most recently completed
• "Last season" = one season before "this season"
• NHL regular season: October to April. Playoffs: April to June.

Initial Assessment

Before starting, understand:

1. Which two teams? Confirm spelling and current team names (relocations happen).
2. Does the user have a model? If yes, include the Model Edge section. If no, skip it.
3. What is the game date? Default to today unless specified.

How It Works

Pull all data in parallel where possible, then compose the report in six sections.

Step 1: Matchup Context

Call get_standings for both teams. Pull:

• Current record (W-L-OTL), points, points percentage
• Division rank and conference rank
• Recent form: last 10 games record (W-L-OTL in last 10 if available, else compute from game log)
• Rest status: derive from get_schedule -- days since last game, back-to-back flag

Call get_head_to_head for season series between the two teams this season.

Step 2: Goalie Matchup

Call get_goalie_stats for both teams' likely starters.

Pull: GP, GS, W-L-OTL, SV%, GAA, SO, last 5 starts SV%.

If both goalies have faced each other: pull from get_head_to_head output.

Note: the API shows post-game confirmed starters only. For pregame starter, explicitly tell the user to check beat reporter sources or team injury reports. Do not guess or fabricate a starter.

Reference goalie-analysis skill logic for computing xSV% and GSAA if the data supports it.

Step 3: Key Stats Comparison

Call get_team_stats for both teams. Pull and compare:

Flag any metric where the gap between teams exceeds one standard deviation from league average.

Step 4: Odds Snapshot

Call get_odds (10 credits). Pull moneyline, puck line (-1.5/+1.5), and total (Over/Under) across all available books.

Show:

• Best available moneyline for each side (line shop)
• Consensus total (most common line)
• Implied probabilities (devigified from the consensus line)

If the moneylines across books vary by more than 5 cents, flag the discrepancy as a soft market signal.

Step 5: Model Edge (conditional)

Only include if the user has provided model probabilities.

Compute EV against best available line using edge-detection skill logic:

ev = (model_prob * decimal_odds) - 1

Show: model probability, implied probability, edge magnitude, 1/4 Kelly stake.

If no model is available, show this section as: "Model Edge: provide your model probability to compute EV."

Step 6: Head-to-Head History

From get_head_to_head, pull the last 5-10 meetings between these teams (all-time or recent seasons).

Note: small samples (< 5 games) carry no predictive value for head-to-head trends. State this explicitly rather than over-interpreting 2-3 games.

Credit Usage

Data Source

PuckAPI (default): All six endpoints. Data is clean and joined.

Your own data: If user provides team/goalie stats as CSV:

1. Verify columns: team names, record fields (W, L, OTL), SV%, GAA, corsiPct, fenwickPct, goalsForPerGame, goalsAgainstPerGame
2. For odds: user must provide manually or use get_odds
3. Credits only consumed for the API calls made; own-data sections are free

Anti-patterns

Output Format

## Game Preview: [Away Team] @ [Home Team]
[Date] | [Arena] | [Time]

---

### Matchup Context
[Away] Record: XX-XX-XX | Pts%: .XXX | Div Rank: X | Last 10: X-X-X
[Home] Record: XX-XX-XX | Pts%: .XXX | Div Rank: X | Last 10: X-X-X
Rest: [Away] X days | [Home] X days [BACK-TO-BACK flag if applicable]
Season Series: [Away] X – [Home] X

---

### Goalie Matchup
[Away] Likely Starter: [Name] | SV%: .XXX | GAA: X.XX | Last 5: .XXX
[Home] Likely Starter: [Name] | SV%: .XXX | GAA: X.XX | Last 5: .XXX
Note: Pregame starters unconfirmed -- check beat reporter sources.

---

### Key Stats Comparison
| Metric        | [Away] | [Home] |
|---------------|--------|--------|
| Corsi%        | XX.X%  | XX.X%  |
| Fenwick%      | XX.X%  | XX.X%  |
| GF/game       | X.XX   | X.XX   |
| GA/game       | X.XX   | X.XX   |
| xGF           | X.XX   | X.XX   |
| xGA           | X.XX   | X.XX   |
| Goal Diff     | +/-N   | +/-N   |

---

### Odds Snapshot
Moneyline: [Away] [+/-XXX] | [Home] [+/-XXX] (best available)
Puck Line: [Away] +1.5 [odds] | [Home] -1.5 [odds]
Total: O/U X.X [odds]
Implied (devigified): [Away] XX.X% | [Home] XX.X%

---

### Model Edge
[If model provided]
Model: [Away] XX.X% | [Home] XX.X%
Market: [Away] XX.X% | [Home] XX.X%
Edge: [Side] +X.X% EV | 1/4 Kelly: X.X% of bankroll
[If no model]
Provide your model probability to compute EV.

---

### Head-to-Head History (Last N meetings)
| Date | Winner | Score | Notes |
|------|--------|-------|-------|

---

Built with PuckAPI Skills

What to Do Next