"Scrum does not change. The complexity of the system changes." — This is not a cliché. Every role on your team has two lenses: a Scrum responsibility and a system intelligence responsibility. Both must be fulfilled.
Diagram 4 — Role-System Mapping
Scrum Roles → System Responsibilities
Each standard Scrum role carries additional system intelligence responsibilities in CP3405. The Dev Team roles below are internal specialisations, not separate entities.
SCRUM ROLES
Product Owner
Guards sprint value. Confirms Definition of Done. Owns the prediction call narrative.
↔ Sprint Goal Owner
Scrum Master
Removes blockers. Protects team flow. Facilitates retrospective. Does NOT command the team.
↔ Process Health Guardian
Development Team
Self-organising. Cross-functional. Builds and delivers the sprint increment. Internal specialisation below.
↔ Intelligence Pipeline Operators
↔
↔
↔
SYSTEM RESPONSIBILITIES
Sprint Intelligence Lead
System Layer: Output
Defines the prediction quality bar. Approves the final locked prediction. Narrates the system story on Monday.
↔ Final release sign-off
Pipeline Integrity Lead
System Layer: Process
Ensures the pipeline runs end-to-end. Identifies where the process broke down. Proposes system improvements in retro.
↔ Process → System feedback loop
Agent + LLM Operators
System Layer: Processing
Almanac & Macro, Technical, Data & Calibration, LLM & GitHub — each owns one stage of the pipeline. Human Score & UI sits with R5.
↔ Manual → Automated pipeline
Mandatory — Not Optional
Role Rotation
Staying in the same role every sprint is comfortable — you get good at it, so there's no reason to switch. That comfort is exactly the problem: it means one person never learns the other five roles. So rotation isn't left to teams to decide. Every member takes a different role each sprint. The Product Owner & Scrum Master keeps a week-by-week log of who held which role and includes it in the team's weekly report — same as the attendance record above.
R1's Weekly Commit — Fixed Format
Weekly Report Template
One file per sprint, committed to your repo at reports/vW<N>.md — same fixed-path convention as everything else this trimester. One block per member. This is what makes rotation and attendance across every team something that can actually be checked week to week, not just claimed.
Nothing was due for Sprint 1 — no presentation happened that week, so there was nothing to log yet. The first report, tagged vW2, must be committed and ready to present on Monday 21 Sep, in class — not sometime after.
SPRINT: 2 TEAM: Team 1 MEMBER: Aye Phyu Sin Soe ROLE: R1 PRESENTED: Y MINUTES: 3 --- MEMBER: Wai Linn Thaw ROLE: R3 PRESENTED: N MINUTES: 0 --- MEMBER: Hoai An Duong ROLE: R5 PRESENTED: Y MINUTES: 4 ---
Rotation is checked automatically by comparing ROLE for the same member across sprint files — no need to manually note "last week's role," it's already sitting in the repo's own history.
Split by Your Actual Team Size
How to Divide the Six Roles
Six roles, but teams aren't all the same size. Find your team size below.
8 Members
One person each on R1, R3, R4, R5. Two people together on R2 (Almanac & Macro) and two together on R6 (LLM & GitHub) — those two carry the heaviest weekly workload, so splitting them across a pair makes sense. On R6, this is the natural LLM-person / GitHub-person split.
7 Members
One person each on R1, R2, R3, R4, R5. Two people together on R6 (LLM & GitHub) — the role with the most hands-on operational work each week, and the natural place to split LLM duties from GitHub duties.
6 Members
One person, one role. No doubling up needed.
5 Members
One person covers R1 and R4 together (Product Owner & Scrum Master, plus Data & Calibration) — both are tracking and process roles, so it's a natural pairing. The other four roles (R2, R3, R5, R6) each get their own person.
All Six Roles — Responsibilities
Every Role Has a Weekend Job and a Monday Speaking Point
R1 — Product Owner & Scrum Master
Sprint Goal Guardian + Process Flow Protector
Weekend: Confirm this sprint's goal and validate the Definition of Done. Coordinate deadlines, remove blockers, ensure every role participates, run a brief retro check. Make the final call on which prediction represents the team's best work.
Monday: State the sprint goal, DoD status, and deliver the final prediction with confidence level. Present the retrospective: what failed, what improved, one specific process change.
Also Monday — every sprint, no exceptions: Record who actually presented, not just who's listed on the roster — name exactly who didn't, if anyone. Note the length of each person's speaking part if you can. This is what protects the rest of your team when someone did the work on paper but never showed up to speak for it. Also log which role each person held this sprint — role rotation is mandatory, and this log is how the team proves it's actually happening, not just claimed. Commit both using the Weekly Report Template below, not a separate format.
Monday: State the sprint goal, DoD status, and deliver the final prediction with confidence level. Present the retrospective: what failed, what improved, one specific process change.
Also Monday — every sprint, no exceptions: Record who actually presented, not just who's listed on the roster — name exactly who didn't, if anyone. Note the length of each person's speaking part if you can. This is what protects the rest of your team when someone did the work on paper but never showed up to speak for it. Also log which role each person held this sprint — role rotation is mandatory, and this log is how the team proves it's actually happening, not just claimed. Commit both using the Weekly Report Template below, not a separate format.
R2 — Almanac & Macro Analyst
Historical Pattern + Macro-Economic Analyst
Weekend: Pull seasonal patterns, historical analogues, and calendar signals (Fed meetings, options expiry, earnings season). Analyse the macro environment: inflation, Fed language, bond yields, dollar strength, sector rotation. Produce written output before LLMs are queried.
Monday: Present the key seasonal insight and the macro thesis supporting or undermining the prediction.
Monday: Present the key seasonal insight and the macro thesis supporting or undermining the prediction.
R3 — Technical Analyst
Chart & Pattern Analyst
Weekend: Analyse SPX, NDX, IWM charts. Identify key levels, trend structure, momentum signals, and volume patterns. Write independent technical output before LLMs.
Monday: Present chart evidence and explain why the technical picture supports (or contradicts) the prediction.
Monday: Present chart evidence and explain why the technical picture supports (or contradicts) the prediction.
R4 — Data & Calibration Lead
Evidence, Accuracy & Learning Tracker
Weekend: Record actual outcomes against predictions and calculate accuracy. Maintain the calibration suite, tracking directional accuracy per model and per agent over time. Identify systematic biases and feed findings into R5's Human Score input.
Monday: Present the outcome vs. prediction, the calibration trend, and which model has the best track record so far.
Monday: Present the outcome vs. prediction, the calibration trend, and which model has the best track record so far.
R5 — Human Score Lead & UI Operator ⭐ Most Important
Wild Card & Override Lead + Interface Owner
Weekend: After all agents and LLMs have reported, review the synthesis. Apply the Human Score. Decide whether to confirm, adjust, or override the AI consensus. Justify the Wild Card observation.
Monday: Present the Human Score reasoning. This is the primary marks differentiator. Weak R5 = weak team score. Then give a live demo of the team's UI/dashboard and show what improved this week.
Starting Sprint 3 — every sprint, not just the last one: R5 also owns the UI, in addition to the Human Score. Spend real time each week making the team's UI/dashboard genuinely better, not just functional. This is not a Demo Day scramble — visible, incremental UI progress is part of the weekly deliverable, same as the pipeline itself. The Human Score responsibilities above are not reduced.
Monday: Present the Human Score reasoning. This is the primary marks differentiator. Weak R5 = weak team score. Then give a live demo of the team's UI/dashboard and show what improved this week.
Starting Sprint 3 — every sprint, not just the last one: R5 also owns the UI, in addition to the Human Score. Spend real time each week making the team's UI/dashboard genuinely better, not just functional. This is not a Demo Day scramble — visible, incremental UI progress is part of the weekly deliverable, same as the pipeline itself. The Human Score responsibilities above are not reduced.
R6 — LLM & GitHub Operator
Multi-Model Comparison + Evidence & Release Owner
Weekend: Query the three LLMs (ChatGPT, Gemini, DeepSeek) with identical prompts — Claude optional if you have access. Build a comparison table. Commit all evidence files, keep the folder structure clean, create the correct release tag for the sprint, verify the release link is accessible before the deadline.
Monday: Present model agreement, divergence, and what that reveals about uncertainty. Open GitHub live, show the release tag, walk through the evidence folder.
Monday: Present model agreement, divergence, and what that reveals about uncertainty. Open GitHub live, show the release tag, walk through the evidence folder.
Monday — Who Says What
| Role | Speaking Point | Evidence to Show |
|---|---|---|
| R1 Product Owner & Scrum Master | Sprint goal, DoD status, final call, retrospective, presentation attendance, role rotation | Prediction file, DoD checklist, retro notes, attendance record, rotation log |
| R2 Almanac & Macro | Seasonal pattern + macro thesis supporting the call | Written Almanac/Macro output file |
| R3 Technical | Chart evidence and key levels | Chart screenshots + technical file |
| R4 Data & Calibration | Outcome vs. prediction, accuracy trend, model biases | Actuals comparison + calibration suite file |
| R5 Human Score & UI ⭐ | Human Score reasoning + Wild Card + UI progress | Human Score file with justification + live UI demo |
| R6 LLM & GitHub | Model agreement/divergence + live repo walk-through | LLM comparison table + GitHub open on screen |