Mahjong Coach

01 — Welcome & Rule Pack Selection

9:41

●●●

Mahjong Coach

AI-powered hand review for every game you play

🀄

Select Your Rule Set

🀄 Guangdong Standard

🀅 Sichuan Blood Battle

🀆 Hunan Red Dragon

🀇 More Variants Coming

Guangdong

Sichuan

Hunan

International

🀄

Guangdong Standard

Widest national coverage · Simplest rules · Ideal starting point

Home

Review

Train

Profile

Settings

02 — 10-Game Initial Snapshot Preview

9:41

●●●

Your First 10-Game Snapshot

10

Games Played

34%

Win Rate

22%

Deal-In Rate

64%

Tile Efficiency

Win Rate

3 wins / 10 games · Below avg

Win / Loss per Game (G1–G10)

G1

G2

G3

G4

G5

G6

G7

G8

G9

G10

Top Mistake: Over-held 發 (Green Dragon) across 4 games

🔒

Unlock Full 30-Game Profile

Basic Training — see the why behind every hand

Home

Review

Train

Profile

Settings

01 — Recent Games History

9:41

●●●

Recent Games · Guangdong Standard

28

Games Logged

39%

Win Rate

13%

Self-Draw

18%

Deal-In Rate

Today

14:32

Game #28 · Won · Peng Peng He

Self-draw · +32 pts · 18 rounds · 3 flags

Today

11:05

Game #27 · Lost · Dealt into Dealer

Over-held 發 · −16 pts · 22 rounds · 1 flag

Yesterday

20:48

Game #26 · Won · Qi Dui (Seven Pairs)

Self-draw · +64 pts · 15 rounds · Clean

Yesterday

18:20

Game #25 · Lost · Opponent Self-Draw

Concealed Kong · −8 pts · 31 rounds · 2 flags

AI Review Ready: 3 critical moments flagged in G#28

Home

Review

Train

Profile

Settings

02 — Game #28 Replay Overview

9:41

●●●

Game #28 · Peng Peng He · Won

+32

Points Earned This Game

↑ Above your avg +19

18

Total Rounds Played

↑ Fast finish · efficient

Tile Efficiency

Above your 30-game avg of 64%

Cumulative Score by Round

R4

R8

R11

R15

R18

3 AI Coaching Flags Detected · 1 excellent play

Home

Review

Train

Profile

Settings

01 — Critical Discard Breakdown

9:41

●●●

Round 11 · Critical Discard Moment

🀇 1-Bam

🀈 2-Bam

🀉 3-Bam

🀙 1-Dot

🀛 3-Dot

🀄 中 Zhong

87%

Discard Decision Score

↑ +22% vs your avg

Decision Quality

You discarded 中 (Red Dragon) — correct

🀄

Discard: Red Dragon (中)

Danger Tile Judgment

Correct

SETS

Tenpai +2 tiles

REPS

2.3 sec

MIN

87% optimal

CAL

AI: Excellent — clears danger tile, opens bamboo 2-sided wait

Home

Review

Train

Profile

Settings

02 — AI Coach Full Commentary

9:41

●●●

AI Coach · Game #28 · Full Review

⚠️

Mistake · Round 8 · Held 發 3 Extra Turns

Value Tile Management

High Risk

SETS

3 rounds held

REPS

Over-retained

MIN

−22% win rate impact

CAL

✅

Good Play · Round 11 · Pivot to Peng Path

Adaptive Hand Shaping

Optimal

SETS

Saved 4 tiles

REPS

1.8 sec

MIN

+31% win rate impact

CAL

💡

Missed · Round 14 · Hidden Qi Dui Path

Pair Recognition

Overlooked

SETS

2 tiles short

REPS

Not taken

MIN

Seven-pairs route available

CAL

Overall Game Score

Good · Top 30% of your 28 games

Home

Review

Train

Profile

Settings

01 — Daily Training Card Deck

9:41

●●●

Today's Training · 5 Cards Due

Daily Deck Progress

3 of 5 cards reviewed today

🀙

Tenpai Recognition · 3-Sided Bamboo Wait

Wait Pattern Reading

Due

SETS

3 hands

REPS

Active drill

MIN

Flagged from G#26

CAL

🀄

Danger Tile Judgment · Honor Tiles Under Threat

Defensive Awareness

New Card

SETS

5 scenarios

REPS

5 min

MIN

From G#27 mistake

CAL

🀇

Efficiency Drill · Bamboo Sequence Building

Hand Construction

Review

SETS

2 rounds

REPS

3 min

MIN

Streak: 4 days

CAL

Advanced Tier: Unlocks full spaced-repetition drill history

Home

Review

Train

Profile

Settings

02 — Mistake Drill Library

9:41

●●●

Your Mistake Library · 18 Saved Drills

All

Discard

Tenpai

Defense

Peng/Kong

🀅

Over-held 發 (Green Dragon)

Value Tile Management

Repeated

SETS

4 games

REPS

Top Mistake

MIN

#1 deal-in cause

CAL

🀙

Chose 1-Sided vs Available 2-Sided Wait

Wait Optimization

Recurring

SETS

3 games

REPS

Intermediate

MIN

−18% tile efficiency

CAL

🀄

Premature Peng Blocking Self-Draw Path

Meld Timing

Common

SETS

5 games

REPS

High impact

MIN

Win rate −11%

CAL

18

Total Drills in Library

↑ −3 resolved this week

Home

Review

Train

Profile

Settings

01 — 30-Game Complete Style Profile

9:41

●●●

Your 30-Game Complete Profile

38%

Win Rate

14%

Self-Draw %

19%

Deal-In Rate

67%

Tile Efficiency

Aggression Index

Attack 76% of all tenpai opportunities

Defense Score

Below avg — primary deal-in source

⚔️ Aggressive

🀄 Honor Hunter

💨 Fast Finisher

Aggressive

Balanced

Defensive

Opportunist

Style: Aggressive Honor Hunter · Risk-Reward Seeker

Home

Review

Train

Profile

Settings

02 — Win Rate Curve & Growth Metrics

9:41

●●●

Growth Dashboard · 30 Games

Win Rate by 10-Game Block

G1–10

G11–20

G21–30

Deal-In Rate Trend (Improving)

G1–10

G11–20

G21–30

+15%

Win Rate Growth Across 30 Games

↑ Training is working

−11%

Deal-In Rate Reduction

↑ Defense improving

Ready for Advanced Training: Consistent improvement detected

Home

Review

Train

Profile

Settings

01 — Platform KPI Overview

app.example.com/coach-analytics-dashboard

Mahjong Coach · Platform Analytics

12,847

Active Users

284,392

Games Reviewed

1.2M

Hands Analyzed

¥980K

MRR

Daily Active Users — Last 30 Days

Wk 1

Wk 2

Wk 3

Wk 4

Revenue Split by Subscription Tier

Trial ¥9

Basic ¥29

Advanced ¥68

Pro ¥198

68%

Trial-to-Paid Conversion Rate

↑ +8% vs last month

4.2

Avg Games Reviewed per User per Week

↑ Strong engagement signal

02 — Hand Efficiency & Regional Sessions

app.example.com/coach-analytics-dashboard

Hand Efficiency Insights · All Users

Most Common Mistake Types (Platform-Wide)

Value Tile Over-hold

Suboptimal Wait

Premature Peng

Missed Defense

Kong Timing Error

67.3%

Platform Avg Tile Efficiency

↑ +3.1% vs last quarter

21.4%

Platform Avg Deal-In Rate

↑ −2.8% training effect visible

Live

Now

Guangdong · 8,241 Active Sessions

Peak hours 19:00–22:00 · 3,400 concurrent avg

Live

Now

Sichuan Blood Battle · 3,102 Sessions

Fastest growing variant · Blood Battle rules active

Live

Now

Hunan Red Dragon · 1,504 Sessions

High deal-in cohort — training impact most visible here

01 — Active Rule Pack Registry

app.example.com/rule-pack-manager

Rule Pack Manager · 3 Active Variants

🀄

Guangdong Standard

National Coverage · Core Pack

8,241

SETS

active users

REPS

v1.2.0 · Stable

MIN

24 hand types

CAL

🀅

Sichuan Blood Battle

Southwest China · Major Variant

3,102

SETS

active users

REPS

v1.1.4 · Stable

MIN

19 hand types

CAL

🀆

Hunan Red Dragon

Hunan Province · Regional Pack

1,504

SETS

active users

REPS

v1.0.8 · Beta

MIN

21 hand types

CAL

Market Variant Coverage

65% of Chinese regional markets served

In Development: Japanese Riichi · Taiwanese Mahjong · EMA Standard

02 — Guangdong Rule Configuration

app.example.com/rule-pack-manager

Guangdong Standard · Rule Configuration

Scoring

Special Hands

Kong Rules

Flower Tiles

🀄

Zi Mo (Self-Draw Bonus)

All players pay double when winner self-draws

Enabled

🀅

Guo Zhong Jie — Dealer Retention on Red Dragon

Dealer retains seat if they win with Zhong (中)

Enabled

🀆

Flower Tiles Scoring Multiplier

Regional flower tiles add scoring weight per variant

Optional

Enable Flower Tile Scoring

Allow Concealed Kong Redraw

Qi Dui (Seven Pairs) Special Hand

24

Recognized Hand Types Configured

↑ Guangdong Standard

Feature Stack & Deliverables

Complete overview of confirmed features, deliverable items, and technical architecture for Mahjong Coach.

🏗️

Tech Stack

Swift / SwiftUIPython / FastAPITensorFlow LitePostgreSQLRedisReact / Next.js

⚡

Core Technologies

🍎

Swift / SwiftUI — Native iOS tile renderer, replay animator, paywall, and training card UI

🐍

Python / FastAPI — Game log ingestion API, hand parsing engine, coaching report generator

🧠

TensorFlow Lite — On-device quantized model for discard scoring and mistake classification

🐘

PostgreSQL — Game log storage, hand history, player profiles, and rule pack registry

⚡

Redis — Real-time hand analysis queue, session cache, leaderboard snapshots

⚛️

React / Next.js — Web admin dashboard for platform analytics and rule pack management

📦

V1 Deliverables Checklist

iOS SwiftUI app with full Guangdong, Sichuan, and Hunan Mahjong rule support
AI hand analysis engine with per-discard decision scoring and flag generation
10-game initial profile with style tag inference and top-mistake detection
30-game complete profile with win rate curve and five-dimension radar chart
Training Card system with spaced-repetition scheduling from flagged game moments
Growth Dashboard with deal-in trend tracking and improvement velocity metrics
Four-tier subscription system: Trial / Basic Training / Advanced Training / Personal Trainer
Web admin panel with platform KPI dashboard and hand efficiency analytics
Rule Pack SDK and configuration UI for regional variant management
App Store submission with StoreKit 2 in-app purchase and TestFlight beta pipeline

🔧

Architecture Layers

Mobile Client

Swift / SwiftUI

Tile board renderer, game log importer, hand replay viewer, training card deck, subscription paywall, growth dashboard

AI Coaching Engine

Python + TensorFlow Lite

Discard decision evaluator, tenpai pattern recognizer, mistake classifier, style tag inferencer, drill card recommender

API & Business Logic

FastAPI + PostgreSQL

Game log parser, 10/30-game profile builder, subscription validator, rule pack registry, coaching report scheduler

Data & Caching Layer

PostgreSQL + Redis

Hand history store, player statistics aggregation, session cache, real-time analysis queue, tile efficiency snapshots

Web Admin Platform

Next.js + Recharts

Platform KPI dashboard, hand efficiency analytics, rule pack CRUD editor, A/B test manager, revenue and conversion reporting

Mahjong Coach

Onboarding & Rule Selection

Game Review Feed

Hand Analysis Detail

Training Cards & Drills

Player Growth Dashboard

Coach Analytics Dashboard

Rule Pack Manager

Mahjong Coach

Feature Stack & Deliverables

Tech Stack

Core Technologies

V1 Deliverables Checklist

Architecture Layers