# SignalStorm Rules

SignalStorm is a six-round strategic forecasting game. Teams advise a Japan-based B2B software company deciding whether to launch a Southeast Asia expansion.

## State

The game tracks four Strategic State meters. Each starts at 50 and is clamped to 0-100:

- Demand
- Pipeline
- Execution
- Regulatory

Round 6 is the launch decision.

## GO Rule

At Round 6, final status is GO if:

- at least three of the four meters are 60 or higher
- no meter is below 40
- no facilitator-defined governance veto is active

Otherwise final status is NO-GO.

## Teams

Team Gut forecasts from the public board only. It has no AI, no card database, no spreadsheet, limited notes, and a short discussion window.

AI Analyst uses the public table state, revealed history, previous forecast tickets, public priors, and possible-cause modelling. It must not see the secret truth manifest, exact hidden cause IDs for unrevealed cards, or unrevealed cause effects.

Ask ChatGPT copies the public board state into ChatGPT, asks what to do, accepts the recommendation, and enters it as the team's forecast.

## Round Forecast Ticket

Each active team submits one ticket per forecast round:

| Field | Type | Score |
| --- | --- | --- |
| Expected next-round Demand | Numeric | Absolute error |
| Expected next-round Pipeline | Numeric | Absolute error |
| Expected next-round Execution | Numeric | Absolute error |
| Expected next-round Regulatory | Numeric | Absolute error |
| P(GO by Round 6) | Binary probability | Brier |
| Prelaunch spend this round | Payoff allocation | Payoff only |

Round 6 is the launch decision, so there is no next-round forecast and no Round 6 prelaunch spend.

## Prelaunch Investment

After a team locks P(GO), it may spend from its prelaunch budget. Spending affects payoff only. It does not change card draws, hidden causes, meters, dynamic effects, or synergies.

SignalStorm defaults to a budget of 1000 per team. Teams may spend any whole number up to their remaining balance in Rounds 1-5.

If final status is GO, each round's spend compounds at 10% per simulated month until launch:

| Round | Multiplier |
| --- | --- |
| 1 | 1.61 |
| 2 | 1.46 |
| 3 | 1.33 |
| 4 | 1.21 |
| 5 | 1.10 |

If final status is NO-GO, prelaunch spend is lost.

## Scoring

Binary forecasts use Brier scoring:

`Brier = (p - outcome)^2`

Here `p` is the decimal probability and `outcome` is 1 if GO happens and 0 if it does not. Lower is better.

Strategic State forecasts use absolute error for each meter:

`Error = |forecast - actual|`

Prelaunch payoff is tracked separately from forecast accuracy.

## Round Flow

1. The facilitator calls the phase.
2. Existing signal cards are flipped to cause sides.
3. New signal cards are dealt, except in Round 6.
4. The table is photographed or scanned.
5. The Game Master applies signal effects, revealed cause effects, dynamic effects, and tag-based synergies.
6. Active teams submit forecast tickets and prelaunch spend while the bracket is open.
7. The Game Master closes the bracket to lock the forecasts and advance.

Do not close the bracket before forecasts are submitted. Do not use calculation as a hidden round advance.