learn

Name: learn
Author: withqwerty

1installs4forks

How to install learn?

withqwerty/nutmeg/learn

$npx -y skills add withqwerty/nutmeg --skill learn

Installs into the current project.

Summary

Learn about football analytics concepts and explore provider documentation. Use when the user asks what a metric means (xG, PPDA, expected threat, xT), wants learning resources, papers, or courses, is new to football analytics, or wants a learning path. Also use when the user ask

Files · 1

View on GitHub

SKILL.md

1	# Learn
2
3	Teach football analytics concepts, recommend resources, provide a learning path, and answer questions about data provider documentation — all adapted to the user's level.
4
5	## Accuracy
6
7	Read and follow `docs/accuracy-guardrail.md` before answering any question about provider-specific facts (IDs, endpoints, schemas, coordinates, rate limits). Always use `search_docs` — never guess from training data.
8
9	## First: check profile
10
11	Read `.nutmeg.user.md`. If it doesn't exist, tell the user to run `/nutmeg` first.
12
13	## Glossary of core concepts
14
15	### Chance quality metrics
16
17	\| Metric \| What it means \| Intuition \|
18	\|--------\|--------------\|-----------\|
19	\| xG (Expected Goals) \| Probability a shot results in a goal (0-1) \| "How good was the chance?" \|
20	\| xGOT (xG on Target) \| xG adjusted for shot placement in the goal \| "How good was the finish?" \|
21	\| xA (Expected Assists) \| xG of the shot that resulted from a pass \| "How good was the chance created?" \|
22	\| xT (Expected Threat) \| Value added by moving the ball to a more dangerous area \| "How much did this pass/carry increase goal threat?" \|
23	\| PSxG (Post-Shot xG) \| Same as xGOT. StatsBomb terminology. \|
24
25	### Possession and pressing
26
27	\| Metric \| What it means \|
28	\|--------\|--------------\|
29	\| PPDA \| Passes allowed per defensive action. Lower = more pressing \|
30	\| High press \| Defensive actions in the opponent's defensive third \|
31	\| Counterpressure \| Immediate defensive reaction after losing the ball \|
32	\| Build-up \| How a team progresses the ball from defence to attack \|
33	\| Possession value \| How much each action contributes to scoring probability \|
34
35	### Passing
36
37	\| Metric \| What it means \|
38	\|--------\|--------------\|
39	\| Progressive pass \| Pass that moves the ball significantly toward the opponent's goal \|
40	\| Key pass \| Pass directly leading to a shot \|
41	\| Assist \| Pass directly leading to a goal \|
42	\| Through ball \| Pass played into space behind the defence \|
43	\| Switch of play \| Long pass crossing the centre of the pitch \|
44	\| Pass completion % \| Successful passes / total passes (misleading in isolation) \|
45
46	### Shooting
47
48	\| Metric \| What it means \|
49	\|--------\|--------------\|
50	\| Shots per 90 \| Shot volume normalised by playing time \|
51	\| Conversion rate \| Goals / shots (noisy, small sample issues) \|
52	\| Big chance \| High-xG opportunity (typically xG > 0.3) \|
53	\| Shot on target % \| Shots on target / total shots \|
54
55	### Defensive
56
57	\| Metric \| What it means \|
58	\|--------\|--------------\|
59	\| Tackles won \| Successful tackle attempts \|
60	\| Interceptions \| Reading and intercepting opponent passes \|
61	\| Clearances \| Defensive clearances (often under pressure) \|
62	\| Blocks \| Blocking shots or passes \|
63	\| Aerial duels won \| Headers contested and won \|
64
65	### Per-90 normalisation
66
67	Always normalise player stats per 90 minutes, not per match:
68	```
69	per_90 = (raw_stat / minutes_played) * 90
70	```
71
72	Why: a player with 2 goals in 180 minutes (per 90: 1.0) is performing the same as one with 1 goal in 90 minutes. Per-match stats penalise part-time players.
73
74	Minimum sample: ~900 minutes (10 full matches) before per-90 stats are meaningful.
75
76	## Learning path
77
78	### Stage 1: Getting started
79
80	1. Read: "The Numbers Game" by Chris Anderson and David Sally. Accessible introduction to football analytics.
81	2. Watch: Tifo Football YouTube channel for visual explainers of tactical and analytical concepts.
82	3. Do: Load StatsBomb open data and make a shot map. Just plot the x,y coordinates of shots, colour by goal/no goal.
83	4. Understand: What xG is and isn't. Read StatsBomb's public xG methodology.
84
85	### Stage 2: Building skills
86
87	1. Read: "Soccermatics" by David Sumpter. Mathematical modelling applied to football.
88	2. Learn: How to make pass networks and xG timelines.
89	3. Practice: Analyse a full match. Write up what happened and what the data shows.
90	4. Explore: FBref for season-level stats. Compare teams across multiple dimensions.
91	5. Tool up: Learn pandas/polars (Python), tidyverse (R), or D3.js (JavaScript) for data manipulation and visualisation.
92
93	### Stage 3: Going deeper
94
95	1. Read key papers:
96	- Decroos et al. (2019) "Actions Speak Louder than Goals" (VAEP model)
97	- Fernandez & Bornn (2018) "Wide Op

People who liked this also viewed

6 picks

Type

learn

1installs4forks

View on GitHub

How to install learn?

withqwerty/nutmeg/learn

$npx -y skills add withqwerty/nutmeg --skill learn

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Learn
2
3	Teach football analytics concepts, recommend resources, provide a learning path, and answer questions about data provider documentation — all adapted to the user's level.
4
5	## Accuracy
6
7	Read and follow `docs/accuracy-guardrail.md` before answering any question about provider-specific facts (IDs, endpoints, schemas, coordinates, rate limits). Always use `search_docs` — never guess from training data.
8
9	## First: check profile
10
11	Read `.nutmeg.user.md`. If it doesn't exist, tell the user to run `/nutmeg` first.
12
13	## Glossary of core concepts
14
15	### Chance quality metrics
16
17	\| Metric \| What it means \| Intuition \|
18	\|--------\|--------------\|-----------\|
19	\| xG (Expected Goals) \| Probability a shot results in a goal (0-1) \| "How good was the chance?" \|
20	\| xGOT (xG on Target) \| xG adjusted for shot placement in the goal \| "How good was the finish?" \|
21	\| xA (Expected Assists) \| xG of the shot that resulted from a pass \| "How good was the chance created?" \|
22	\| xT (Expected Threat) \| Value added by moving the ball to a more dangerous area \| "How much did this pass/carry increase goal threat?" \|
23	\| PSxG (Post-Shot xG) \| Same as xGOT. StatsBomb terminology. \|
24
25	### Possession and pressing
26
27	\| Metric \| What it means \|
28	\|--------\|--------------\|
29	\| PPDA \| Passes allowed per defensive action. Lower = more pressing \|
30	\| High press \| Defensive actions in the opponent's defensive third \|
31	\| Counterpressure \| Immediate defensive reaction after losing the ball \|
32	\| Build-up \| How a team progresses the ball from defence to attack \|
33	\| Possession value \| How much each action contributes to scoring probability \|
34
35	### Passing
36
37	\| Metric \| What it means \|
38	\|--------\|--------------\|
39	\| Progressive pass \| Pass that moves the ball significantly toward the opponent's goal \|
40	\| Key pass \| Pass directly leading to a shot \|
41	\| Assist \| Pass directly leading to a goal \|
42	\| Through ball \| Pass played into space behind the defence \|
43	\| Switch of play \| Long pass crossing the centre of the pitch \|
44	\| Pass completion % \| Successful passes / total passes (misleading in isolation) \|
45
46	### Shooting
47
48	\| Metric \| What it means \|
49	\|--------\|--------------\|
50	\| Shots per 90 \| Shot volume normalised by playing time \|
51	\| Conversion rate \| Goals / shots (noisy, small sample issues) \|
52	\| Big chance \| High-xG opportunity (typically xG > 0.3) \|
53	\| Shot on target % \| Shots on target / total shots \|
54
55	### Defensive
56
57	\| Metric \| What it means \|
58	\|--------\|--------------\|
59	\| Tackles won \| Successful tackle attempts \|
60	\| Interceptions \| Reading and intercepting opponent passes \|
61	\| Clearances \| Defensive clearances (often under pressure) \|
62	\| Blocks \| Blocking shots or passes \|
63	\| Aerial duels won \| Headers contested and won \|
64
65	### Per-90 normalisation
66
67	Always normalise player stats per 90 minutes, not per match:
68	```
69	per_90 = (raw_stat / minutes_played) * 90
70	```
71
72	Why: a player with 2 goals in 180 minutes (per 90: 1.0) is performing the same as one with 1 goal in 90 minutes. Per-match stats penalise part-time players.
73
74	Minimum sample: ~900 minutes (10 full matches) before per-90 stats are meaningful.
75
76	## Learning path
77
78	### Stage 1: Getting started
79
80	1. Read: "The Numbers Game" by Chris Anderson and David Sally. Accessible introduction to football analytics.
81	2. Watch: Tifo Football YouTube channel for visual explainers of tactical and analytical concepts.
82	3. Do: Load StatsBomb open data and make a shot map. Just plot the x,y coordinates of shots, colour by goal/no goal.
83	4. Understand: What xG is and isn't. Read StatsBomb's public xG methodology.
84
85	### Stage 2: Building skills
86
87	1. Read: "Soccermatics" by David Sumpter. Mathematical modelling applied to football.
88	2. Learn: How to make pass networks and xG timelines.
89	3. Practice: Analyse a full match. Write up what happened and what the data shows.
90	4. Explore: FBref for season-level stats. Compare teams across multiple dimensions.
91	5. Tool up: Learn pandas/polars (Python), tidyverse (R), or D3.js (JavaScript) for data manipulation and visualisation.
92
93	### Stage 3: Going deeper
94
95	1. Read key papers:
96	- Decroos et al. (2019) "Actions Speak Louder than Goals" (VAEP model)
97	- Fernandez & Bornn (2018) "Wide Op

People who liked this also viewed

6 picks

Type