Four labels answer four different questions
Published fact
A regulator, operator, or other named source establishes the field being shown—for example, a payout, seat count, buy-in range, or rake schedule. The source may support only that field, not the entire table profile.
Simulation default
DeciDojo chooses a value needed to make the table playable, such as a starting wallet, shoe penetration, smallest chip, or a rule not specified by the source. It is a product setting, not a claim about a venue.
Verified strategy
A Blackjack decision matrix is tied to a specific rule package and checked against identified strategy references or an independent calculation. Its identity changes when the rules that affect decisions change.
Learning baseline
A limited educational model provides broad, supportable guidance but does not claim exact optimization. This is the label used when the evidence supports useful principles without supporting a complete solver-grade policy.
The selected profile controls the game
A rule label is useful only if settlement follows it. Blackjack profiles control dealer procedure, natural payout, doubling, splitting, surrender, insurance, and shoe behaviour. Poker profiles control seats, blinds, buy-ins, stack continuity, straddles, rake eligibility, rate, cap, rounding, and ledger timing.
Example: the Continental Blackjack profile does not merely display “ENHC.” Its settlement exposes added split and double wagers to a later dealer Blackjack. The Macau-inspired OBBO profile protects qualifying added wagers under a different loss policy.
Persisted sessions store the profile identity and version. A saved cash table cannot silently resume under a different economic contract.
Advice is constrained by what the model knows
Blackjack can use a rule-specific basic-strategy authority for the visible player hand and dealer upcard. Even there, a beginner baseline is labelled when the available matrix does not justify an exact finite-shoe or special-settlement claim.
Poker feedback begins with state facts: position, price, pot, stack, legal actions, public cards, and action history. Broad learning guidance can use those facts, but the current public game does not present itself as a full GTO solver and does not infer hidden cards.
What a good explanation should do
It should identify the fact that supports the feedback, state any limitation that changes its authority, and help the player make the next decision—not rewrite the result after seeing the final card.
The method includes a correction path
Every game can open the DeciDojo report form from its settings. A report can include the visible page, viewport, and current non-secret game state, but it never attaches files or hidden cards. Reports receive a reference and are stored for review.
Rule sources and their review dates live with the profile definitions. When a source changes or a limitation becomes clearer, the appropriate fix is to update the source record, profile version, tests, explanation, or all four—not to leave a stronger claim in the interface.
DeciDojo’s methodology is intentionally conservative about the strength of a claim, but not about product scope: the games can still model varied casino-style environments as long as the sourced and simulated parts remain visible.