How the Intelligence Engines Work
This page describes the current implementation at a high level. Each engine depends on available source data and its current configuration; implementation details and provider coverage can change.
1. Both Teams to Score (BTTS)
The existing BTTS model evaluates recent goals scored and conceded, home/away scoring profiles, historical BTTS rates, sample size, and competition context. A Poisson-style expected-goals component is blended with historical BTTS rates. The current Worker applies minimum history and model-quality thresholds, and the history sample is cut off before the target date to avoid using same-day or future results as prediction inputs. If there are not enough qualifying fixtures or source data, the engine may return no selection.
The displayed AI-reference odds are derived from model probabilities; they are not live bookmaker odds. They should not be treated as prices available at a sportsbook.
2. VIP Daily Winning Ticket
This is a separate football engine that considers configured football markets and ranks eligible fixtures using its current probability and history settings. Its access is controlled by the existing account and VIP rules. The number of selections and market mix can vary with the day's eligible fixtures. No ticket or outcome is guaranteed.
3. Cross-exchange arbitrage screening
The scanner compares available public spot-market quotes from supported exchanges. It evaluates buy/sell direction, estimated gross spread, configured trading-fee assumptions, reported market depth or liquidity, and the existing eligibility rules. A positive calculation is only a screening result. Quotes from separate exchanges are not necessarily synchronized, and the scanner does not prove that the required balances, transfer paths, order sizes, or fills are available to a specific user.
4. Triangular arbitrage screening
The triangular engine evaluates three-pair conversion routes within an exchange using the available market relationships and price direction for each leg. It applies modeled fees and route checks. Missing prices, inadequate displayed quantity, precision limits, fees, and sequential execution can make a theoretical route unavailable or unprofitable in practice.
5. Spot signal intelligence
The current spot scanner uses Gate.io USDT spot market data and recent candles. It considers the supported market list, recent movement, configured fee and safety assumptions, the selected target window, and visible order-size information when constructing possible manual setups. It does not place trades, and the estimated time to reach a target can be wrong. Tracking updates while the page remains open; browser throttling or a closed page can delay updates.
6. Smart Trading Assistant
The assistant is implemented through the Cloudflare Workers AI binding and is gated by the existing account/VIP logic and a daily usage limit. It receives the user's question and a selected market snapshot to explain the current setup. It cannot see private exchange balances, orders, or fills, and its answers can be wrong. It is not a substitute for independently checking live market data.
Data freshness and unavailable results
Exchange and football feeds can fail, return incomplete data, or change between requests. If the interface reports an error, stale quote, missing candle, insufficient history, or no qualifying opportunity, treat that as a limitation rather than assuming a signal exists. Do not interpret a blank or unavailable result as a recommendation.
Read the Risk Disclaimer and Results and Performance Methodology before relying on a displayed estimate.