Couponwatch

自動比對出場、配息、下限價只是基本功能。選標的的階段也能用：輸入市場和篩選條件，AI會列出候選標的，但年化波動度、最大回撤、目前價位區間這些數字，是系統用真實歷史股價算出來的，不是AI編的。AI負責給理由，數字由系統負責。條款設計也是同樣的邏輯：一組假設的出場價和轉換價，會在過去每一個可能的進場日跑一次回測，勝率、提前出場率、平均持有月數立刻就能看到，不必等商品真的發行才知道條件設得合不合理。 Automatically matching exit/coupon/knock-in triggers is just the starting point. Finding the underlying is covered too: describe the market and filters, AI lists candidates, but the annualized volatility, max drawdown, and current price-range position are numbers the system computes from real historical prices — not numbers the AI made up. AI supplies the reasoning, the system supplies the numbers. Structuring a product's terms works the same way: run a hypothetical exit/conversion price against every possible entry date in history, and the win rate, early-exit rate, and average holding period show up immediately, instead of waiting for the product to actually run before finding out whether the terms make sense.

除了標準型商品，也支援雙贏型（Twin-Win）：只要下限價從未被觸及，標的到期時無論漲跌都算正報酬。商品到期時系統會自動算出估算報酬，理專不必每天手動核對。通知可以走Email或LINE，兩者各自獨立，只開一個也可以。每位理專只看得到自己名下的商品和客戶，彼此的資料互不相通。 Beyond the standard payoff, there's a Twin-Win structure — if the knock-in level is never touched, the product pays out positively regardless of whether the underlying ends up or down. Maturity notifications calculate the estimated payout automatically, so the advisor doesn't have to check by hand every day. Notifications go out over email and LINE, either channel independently. Each advisor only ever sees their own products and clients — never anyone else's.