Strategies – formulas, evidence and AI review
A CoinRobot strategy describes the decision rules used by one or more Plans. Strategies are versioned and may be private, shared with selected users or public. Plan results remain separate, while strategy-level analysis aggregates compatible evidence across the Plans using the same strategy family and formula revision.
1. Identity and visibility
Name and description should explain the intended market behaviour. Private strategies are visible only to their owner, Limited strategies can be shared with selected referrals, and Public strategies may be selected by the community. Updating a finalized strategy creates a new version in the same strategy family so earlier results remain attributable to the formula that produced them.
2. BUY, SELL and StopLoss formulas
The BUY formula defines entry evidence; SELL defines ordinary profit-taking or risk exits; StopLoss is evaluated as a separate protective rule. CONVERT Plans also preserve the evaluated BUY and SELL formula outcomes, but their execution objective is asset-unit growth through rotation rather than only cash profit. Calculator names, descriptions, intervals and available fields are shown in the strategy editor and can be copied with the formula catalog.
3. Evidence used for evaluation
CoinRobot combines short and longer windows, daily KPIs such as trade count, win rate, profit factor, realized result and CONVERT growth, execution evidence, StopLoss follow-up behaviour, active formula revision history and a compact market-regime context. Only the active revision can receive a new change; older revisions are background evidence for judging whether an earlier tightening or relaxation worked.
4. Requesting a dedicated AI review
The owner can request an AI review from the strategy list when at least one Plan supplies evidence. The request costs 100 credits and a strategy family may be submitted once per UTC day. No credit is deducted while the request is merely queued. The charge is recorded only after a successful AI analysis and remains applicable when the defensible conclusion is no change. The AI receives fresh local market statistics plus recent cached research context without initiating additional external market-data downloads. See Credits for billing details.
5. Validation, activation and feedback
A proposed formula must pass parser and multi-scenario checks before activation. The evaluator checks normal, boundary, missing-data, zero-denominator and different market-regime inputs and blocks formulas that collapse into constant true or false behaviour. The owner receives a localized email describing the reviewed evidence, KPIs, AI conclusion, validation outcome and a direct link back to the strategy.
6. Using and finalizing the adaptive version
Plans that use the adaptive option automatically receive the active validated formula override for the strategy family. The original version remains traceable. After sufficient evidence has accumulated, an administrator may promote the active adaptive formulas into a finalized new strategy version. Automated analysis reduces manual work but does not remove market risk; read the Risk disclaimer before trading.