#Book of Odin — Hold & Win Math Change Request

Status: open. This is a requirements handoff, not a Math Spec, PAR sheet, simulation result or certification claim.

#1. Objective

Create a new canonical math/config version for book-of-odin that preserves the recognizable BOOK/Expanding Symbol Free Spins game while adding a 5×3 Hold & Win triggerable from both base game and Free Spins.

Do not mutate book-of-odin-v1 in place. Use a new version identifier, provisionally book-of-odin-v2 until the backend naming is approved.

#2. Existing source identity

The source comment states that the existing seed has an approximately 9.7% Free Spins trigger rate against an approximately 1% target. Therefore current frequencies, RTP allocation and feature-buy prices must not be treated as an accepted baseline merely because they are present in the migration.

#3. Fixed product rules

These are approved design inputs:

1. Grid remains 5 reels × 3 rows with 10 fixed paylines.
2. BOOK remains Wild + Scatter.
3. 3+ BOOK award 10 Free Spins.
4. One regular symbol A–I is selected for expansion for the entire FS series.
5. FS retrigger is disabled in this version.
6. Add COIN; 6+ visible COIN trigger Hold & Win in base game or Free Spins.
7. Triggering COIN carry into the same cells as initial locked values.
8. Hold & Win begins with 3 respins; any new COIN resets the counter to 3; a miss decrements it.
9. Ordinary termination occurs at 0 respins; full-grid termination occurs at 15 locked cells.
10. Ordinary award is the sum of locked total-bet multipliers.
11. Full grid adds one GRAND total-bet multiplier on top of the COIN sum.
12. No MINI/MINOR/MAJOR, direct Buy Hold & Win or additional wager.
13. Hold & Win may trigger repeatedly from separate Free Spins in one series.
14. Simultaneous base triggers resolve as line win → Hold & Win → Free Spins.
15. A Hold & Win inside FS pauses and then resumes the same FS series without consuming an extra Free Spin.

#4. Variables the math owner must determine

No value in this section may be silently inferred from Magic Vault or another game.

#5. Required mathematical work

#5.1 Preserve and revalidate the base system

• Recalculate base line RTP after introducing COIN to the strips.
• Confirm BOOK substitution/scatter behavior against the actual evaluator.
• Replace the known high FS-trigger draft distribution with an explicitly approved target.
• Prove that adding COIN does not create unintended line awards or substitutions.

#5.2 Free Spins and expanding symbol

• Calculate each FS_BOOK_A through FS_BOOK_I variant independently.
• Measure the selection-weighted aggregate and the contribution of expansion.
• Include COIN frequency and H&W probability within every variant.
• Report the distribution of zero, one and multiple H&W launches per FS series.
• Confirm that disabling FS retrigger is reflected in both config and model.

#5.3 Hold & Win

• Model entry with every reachable initial locked count and spatial distribution.
• Include different entry distributions from base game versus each FS variant.
• Calculate conditional mean, variance, duration and full-grid probability.
• Calculate ordinary COIN-sum awards separately from GRAND contribution.
• Verify termination, reset-on-hit and last-respin-hit behavior exactly.
• Quantify tail exposure when multiple H&W features occur in one FS series.

#5.4 Whole-game reconciliation

At minimum, reconcile:

whole-game RTP
  = base line contribution
  + Free Spins line/expansion contribution
  + base-origin Hold & Win contribution
  + Free-Spins-origin Hold & Win contribution
  + GRAND contribution

Avoid double-counting an FS-origin Hold & Win: its award belongs to the same purchased/base game cycle but must be reported as a distinct component for auditability.

#6. Required exact checks and simulations

Keep exact and sampled results in separate tables.

#Exact where tractable

• payline evaluator parity for representative grids;
• BOOK trigger rule and selected-symbol variant weights;
• COIN trigger count and carry-over cell mapping;
• award summation, GRAND addition and global-cap order;
• feature state transitions and all termination boundaries;
• theoretical maximum/reachability proof or a documented reason enumeration is infeasible.

#Simulation

• whole-game RTP with confidence interval;
• hit rate, win distribution and volatility metrics;
• base and FS feature-trigger frequencies;
• H&W conditional distribution by source and initial locked count;
• H&W duration distribution;
• full-grid/GRAND probability;
• maximum observed exposure and convergence evidence;
• Buy Free Spins EV and RTP for every offered buy option.

Simulation output must go under games/book-of-odin/evidence/ with seed, sample size, backend commit, canonical config hash and runner identity. A simulation is not certification.

#7. Engine and protocol capability checks

Before tuning, engineering must confirm the runtime can represent:

• COIN-triggered durable Hold & Win from a paid base spin;
• the same durable feature nested inside a Free Spins sequence;
• suspension and exact resumption of remaining FS state;
• repeated H&W launches from separate FS outcomes;
• deterministic replay of the selected expanding symbol, locked cells and coin multipliers;
• a full-grid award that adds to the COIN sum rather than replacing it;
• one game-level cap across line, FS, H&W and GRAND awards;
• idempotent feature steps and recovery without a second wager.

Any missing capability is a STOP for finalizing the Math Spec; do not tune around a protocol or state model that cannot express the approved rules.

#8. Acceptance criteria

The change request is complete only when:

• a new canonical config version exists and book-of-odin-v1 is unchanged;
• its backend Git identity and SHA-256 are recorded here and in the game README;
• GDD, Math Spec and config agree on all triggers, ordering and termination rules;
• exact checks and simulations are stored as separate evidence;
• whole-game RTP reconciles within the approved tolerance;
• volatility, max win and full-grid reachability are evidenced;
• every Buy Free Spins option has newly calculated price/EV or is removed;
• nested/repeated feature recovery has integration tests;
• QA test vectors cover base H&W, FS H&W, simultaneous BOOK+COIN, repeated FS H&W, last-respin landing, full grid and reconnect;
• product, math, engineering, QA and compliance approvals are recorded.

#9. Explicit non-claims

This document does not claim that the current backend supports nested H&W, that the candidate coin multipliers are balanced, that any RTP target has been achieved, or that the game is ready for stage, production or certification.