What you teach

An original 16-space investment game. Roll 1d4 automatically, decide whether to buy or upgrade, collect rent and 80 on passing Start. Tax space costs 35. Finish at bankruptcy or 60 turns; higher net worth wins.

Public cash and holdings; future dice are hidden. Fixed costs and rents, at most two upgrades. No auctions, mortgages or trades. The NN imitates a cash-reserve/rental-yield scorer, not financial advice.

1. World: try the reference, then control a round yourself.
2. Train: fit a small action-scoring network away from the animation.
3. Results: compare paired seeds, inspect losses and activate checked weights.
4. World: test a new seed. Save project to keep your model.

A single linear neuron scores each action. It receives these engineered features, not a teacher’s chosen action:

• Pass
• Buy
• Upgrade
• Cash reserve after cost
• Rental yield
• Opponent cash pressure

Ridge least squares fits the weights from synthetic reference scores. Rules supply legal actions, feature extraction and opponent behaviour. This is imitation learning. It does not discover an optimal strategy through self-play.

Public cash and holdings; future dice are hidden. Fixed costs and rents, at most two upgrades. No auctions, mortgages or trades. The NN imitates a cash-reserve/rental-yield scorer, not financial advice.

64 fitting seeds, 12 validation seeds and 24 final seeds. Validation selects between two ridge strengths; final tests cannot change the selected weights. Maximum 25 seconds of computation, 30-second worker timeout and two attempts per training round.

Ready · no provider or API key needed.

Activation gate: ≥80% near-best reference-score agreement, finite compatible weights and final mean score at least the untrained baseline and a checked incumbent, when present. Passing this small test does not guarantee future wins.