從市場資料到策略驗證，建立有脈絡、可回溯的量化研究流程。A considered path from Taiwan market data to systematic evaluation and traceable research.
成鳳科技專注台股量化研究軟體。整合資料、規則與研究紀錄，是我們持續推進的產品方向。Windforce develops quantitative research software for Taiwan equities, bringing data, explicit rules, and research records into a connected workflow.
如何檢驗一項台股策略假設？How should a Taiwan equity strategy be evaluated?
我們為台股量化研究情境開發軟體，著重資料準備、規則驗證及研究紀錄。讓研究者能回到每個假設的起點，理解結果如何產生。We develop software for quantitative research on Taiwan equities, focusing on data preparation, rule-based evaluation, and research records—so researchers can revisit each assumption and understand how a result was produced.
以下為研發範圍與預期輸出；實際功能以內部測試版本為準。Development scope and intended outputs below; available features depend on the internal test version.整理合法取得的股價、成交量與公司資料，依股票代碼與交易日期對齊。從來源、缺漏值到版本紀錄，讓後續研究建立在清楚的資料脈絡上。Organize lawfully obtained prices, volumes, and company data by stock code and trading date. Source references, missing-value checks, and version records give each study a clear data foundation.
以明確的選股、進出場、持有期間與部位規則進行回測，納入成本與滑價假設，並區分樣本內外期間，檢視策略對參數及市場情境的敏感程度。Evaluate explicit screening, entry, exit, holding-period, and position rules. Include cost and slippage assumptions, separate in-sample and out-of-sample periods, and examine sensitivity to parameters and market conditions.
將研究問題、資料版本、參數與結果串成完整紀錄。規劃以 AI 協助草擬摘要及整理待驗證問題，保留原始數值與人工核對的依據。Connect research questions, data versions, parameters, and findings in one record. Planned AI assistance drafts summaries and follow-up questions while retaining numerical evidence for human review.
以下以示意研究說明產品的工作流程方向。每一步都保留要檢查的條件，讓數值、假設與人的判斷各有依據。This illustrative study explains the intended workflow. Each step carries its own checks, keeping calculations, assumptions, and human interpretation grounded in evidence.
示例：成交量條件是否改變策略的樣本外結果？先定義比較方式，再看結果。Example: does a volume condition change out-of-sample results? Define the comparison before inspecting the outcome.
在建立回測前，先檢視資料授權、交易日期及欄位一致性。研究記錄應能連回使用的資料版本。Before evaluation, review data rights, trading dates, and field consistency. A study should link back to the version of the data it uses.
用程式化計算檢驗既定規則，記錄交易成本及滑價，區分開發策略的期間與獨立驗證期間。Test explicit rules with programmatic calculations. Record transaction costs and slippage and separate the strategy-development period from independent validation.
AI 輔助方向為整理研究脈絡與草擬結果說明。計算流程負責數值，研究者負責核對引用、假設與解讀。Planned AI assistance organizes research context and drafts explanations. Calculation workflows provide the figures; researchers verify references, assumptions, and interpretation.
我們正在評估將 Claude API 納入研究助理模組，協助整理文件、解釋回測結果與草擬研究摘要。整合仍在規劃中。We are evaluating Claude API for a research assistant that organizes documents, explains backtest findings, and drafts research summaries. Integration is still in the planning stage.
經授權文件、研究問題、彙總計算結果Authorized documents, research questions, and aggregated computed results
摘要草擬、方法說明、待驗證問題Draft summaries, method explanations, and follow-up questions
核對引用、數值、假設與結論限制Verify references, figures, assumptions, and limitations
回測數值由計算流程提供，模型協助整理與表達。Backtest figures come from calculation workflows; the model assists with organization and explanation.
規劃保留資料日期、版本與引用，供研究者核實。The intended design retains dates, versions, and references for researcher verification.
規劃使用已授權資料，不向模型提供交易帳戶憑證。The planned workflow uses authorized data and excludes trading-account credentials.
成鳳科技有限公司成立於 2023 年，由創辦人主導產品研發。我們專注於台股量化研究軟體，持續透過內部測試回饋，調整資料整理、策略驗證與研究文件的工作流程。Founded in 2023, Windforce is a founder-led company developing quantitative research software for Taiwan equities. Internal testing feedback informs ongoing work on data preparation, strategy evaluation, and research documentation.
產品供內部投資人測試使用，尚未對外公開提供服務。開放時程依測試結果與開發進度確認。The product is being tested by internal investors and is not yet a public service. Public availability will depend on testing results and development progress.
目前產品供內部投資人測試使用，尚未提供公開註冊。如希望了解產品研發、研究需求交流或技術合作，歡迎透過商務信箱聯絡。The product is currently in testing with internal investors and does not offer public registration. Contact us for product development, research discussions, or technical collaboration.
Claude API 是研究助理的規劃整合方向，主要協助研究文件、方法說明與結果摘要。數值仍由計算流程產生，研究者負責核實與判讀；目前官網本身沒有串接 Claude API。Claude API is a planned integration for assistance with research documents, method explanations, and summaries. Numerical results come from calculation workflows and remain subject to human review. This company website does not integrate the Claude API.
本網站介紹公司與研究軟體研發方向，不提供個別投資建議、代客操作或報酬保證。回測為歷史模擬，不能代表未來績效。This site describes the company and its research software development. It does not provide personalized investment advice, discretionary account management, or guaranteed returns. Backtests are historical simulations and do not imply future performance.
產品研發、台股研究與技術合作，歡迎與我們聯繫。For product development, Taiwan equity research, and technical collaboration.
更新日期：2026 年 10 月 12 日。適用於本公司介紹網站；未來產品服務另以正式服務契約及資料處理說明為準。Updated October 12, 2026. These notices cover this company information website. Future product services will be governed by their own service agreements and data-processing notices.
本網站由成鳳科技有限公司提供，用於介紹公司及產品。產品目前供內部投資人測試使用，未對外公開提供服務。標示為規劃或預期的功能，不構成可立即使用、交付日期或服務水準的承諾；測試、合作範圍、費用與交付內容依另行約定辦理。This website is provided by 成鳳科技有限公司 to describe the company and its product. The product is in testing with internal investors and is not offered as a public service. Planned features do not constitute commitments to immediate availability, delivery dates, or service levels. Testing, collaboration scope, fees, and deliverables are subject to separate arrangements.
網站原創文字與設計之權利由本公司或合法權利人保有。引用請註明來源；重製、商業使用或使用第三方市場資料，須取得必要授權。不得利用本網站冒用公司名義、散布不實資訊、侵害他人權利或干擾網站運作。Rights in original site content and design remain with the company or their lawful owners. Attribute quotations and obtain necessary permission for reproduction, commercial use, or third-party market data. Do not impersonate the company, spread false claims, infringe rights, or interfere with the site.
研發內容可能隨實際進度調整。外部網站由各自營運者負責，其條款與隱私政策另行適用；提及第三方產品不代表合作、認證或背書。疑問或錯誤請透過本頁商務信箱聯絡。Development information may change with progress. External websites are operated independently under their own terms and privacy policies. Mentioning a third-party product does not imply a partnership, certification, or endorsement. Contact our business email with questions or corrections.
本網站未提供會員註冊、付款或文件上傳功能，也未嵌入廣告追蹤或分析程式。切換語言僅作用於目前頁面。提供網站的代管服務可能因連線、安全與維運需要處理 IP 位址、瀏覽器資訊、存取時間及技術紀錄；實際處理依代管服務適用政策。This website has no registration, payment, or document-upload feature and embeds no advertising trackers or analytics scripts. Language selection affects the current page only. Hosting providers may process IP addresses, browser information, access times, and technical logs for delivery, security, and operation, under their applicable policies.
點選信箱連結將開啟您的電子郵件程式，不會自動送出資料。您主動來信時，我們會取得信箱、姓名及您提供的內容，用於回覆、合作洽談及必要的業務紀錄。請勿寄送密碼、券商憑證或非必要的敏感個資。電子郵件服務商可能為傳送與儲存郵件處理相關資料。Selecting an email link opens your email application and does not automatically send data. If you contact us, we receive your email address, name, and message to respond, discuss collaboration, and maintain necessary business records. Do not send passwords, brokerage credentials, or unnecessary sensitive personal information. Email providers may process data to transmit and store messages.
聯絡資料僅於處理往來、必要業務紀錄及法令要求的期間內保留。您可透過本頁信箱請求查詢、更正或刪除所提供的個人資料；我們將在核實身分及適用法令範圍內處理。本網站未串接 Claude API，不會因瀏覽而將您的內容傳送給 Claude；未來產品如採用 AI 資料處理，將另行說明處理目的、範圍與服務供應商。Contact information is retained for correspondence, necessary business records, and legal requirements. You may email us to request access, correction, or deletion of information you provided, subject to identity verification and applicable law. This website does not integrate the Claude API or send your content to Claude through browsing. Future AI processing in a product will be described separately, including its purpose, scope, and service providers.
本網站不提供個別投資建議、投資招攬、代客操作或報酬保證。研究方法與示意流程不構成買賣任何證券的建議。回測屬歷史模擬，可能受到資料品質、存活者偏誤、前視偏誤、過度擬合、交易成本與流動性影響；歷史結果不代表未來績效，實際投資可能發生損失。This website provides no personalized investment advice, investment solicitation, discretionary account management, or guaranteed returns. Research methods and conceptual workflows are not recommendations to trade securities. Backtests are historical simulations affected by data quality, survivorship and look-ahead bias, overfitting, transaction costs, and liquidity. Past results do not imply future performance, and real investments may lose value.