夏普Sharpe
1.42
流程很简单：发现、构建、回测、监控。但值得信任一次运行的门槛更严格：手续费和滑点都会建模，样本量会被点明，前视风险会被检查，FDR / DSR 与夏普并列呈现，每一个实时监控都始终绑定它所来自的那次回测。The flow is simple: discover, build, backtest, monitor. But the bar for trusting a run is higher — fees and slippage are modeled, sample size is called out, look-ahead risk is checked, FDR / DSR sit next to Sharpe, and every live monitor stays bound to the backtest it came from.
同一份策略数据贯穿每一个环节，但每一个有用的决策，都从附在它身上的证据开始：成本、样本、可复现性，以及实时漂移。The same strategy data runs through every stage, but every useful decision starts from the evidence attached to it: cost, sample, reproducibility, and live drift.
精选信号预设、研究对象、指标配方。只有在看到附带的证据之后，才 Fork。Curated signal presets, research subjects, and indicator recipes. Fork only after you've seen the evidence attached.
引导式编辑器用于快速迭代，高级逻辑图编辑器用于完全掌控——底层是同一份数据。A guided editor for fast iteration, an advanced logic-graph editor for full control — the same data underneath.
贴近真实的手续费、滑点、样本警告与试验校正后的证据。对比运行、复现快照、导出报告。Realistic fees, slippage, sample warnings, and trials-corrected evidence. Compare runs, reproduce snapshots, export reports.
即时提醒、免打扰时段，以及绑定回到当初证明可上线的那次运行的健康检查。Instant alerts, quiet hours, and health checks bound back to the run that proved it was ready.
带着可信度信息发布一个策略。别人 Fork 它；你在工作区里追踪这条血缘。Publish a strategy with its trust info. Others fork it; you track the lineage in your workspace.
先从引导式编辑器开始——选一个指标、设一个阈值、选一个方向。准备好之后，把同一套规则切换到逻辑图编辑器，处理嵌套条件、自定义指标和版本历史。Start in the guided editor — pick an indicator, set a threshold, choose a direction. When you're ready, switch the same rules into the logic-graph editor for nested conditions, custom indicators, and version history.
每个结果在展示收益之前，都会先点出它可能存在的问题——样本量、前视风险、成交量缺失。自定义手续费、滑点模型，以及任意两次运行的并排对比。Before showing returns, every result calls out what might be wrong with it — sample size, look-ahead risk, missing volume. Custom fees, slippage models, and a side-by-side compare of any two runs.
规则真正触发时才即时提醒。尊重免打扰时段。健康面板会呈现信号新鲜度、与回测的漂移以及回撤趋势，让你知道何时该暂停。Instant alerts only when a rule actually fires. Quiet hours respected. The health panel surfaces signal freshness, drift from the backtest, and drawdown trend, so you know when to pause.
Factor Lab 是已完成实验的归宿。app 主页讲解产品；归档则让你审视已完成的研究对象，并在证据足够好时，把数据 Fork 到 Build。Factor Lab is where finished experiments live. The homepage explains the product; the archive lets you review completed research subjects and fork the data into Build when the evidence is good enough.
打开 Factor LabOpen Factor Lab没有策略灵感式的宣传语。归档把每一条都当作带有机制和原假设的已完成实验。No strategy-inspo taglines. The archive treats each one as a finished experiment with a mechanism and a null hypothesis.
净夏普、硬指标得分、闸门、容量状态和成本真实性，被设计成可以一起扫读。Net Sharpe, hard-metric score, gate, capacity status, and cost realism are designed to be scanned together.
当你信任这个对象时，Fork 到 Build 就是主操作。详情和导出依然可用，但属次要。When you trust a subject, forking it into Build is the primary action. Details and export stay available, but secondary.
主站负责卖产品。Factor Lab 则像一个独立的研究界面，本地导航更安静。The main site sells the product. Factor Lab acts like a standalone research interface with quieter, local navigation.
免费版是一个完整的研究工具——不是演示版。Pro 抬高上限。Pro+ 加入重度研究功能。编辑器本身没有任何套餐门槛。Free is a complete research tool — not a demo. Pro raises the limits. Pro+ adds heavy research features. The editor itself has no plan gate.
发现、构建并验证你的第一个加密策略所需的一切。Everything you need to discover, build, and validate your first crypto strategy.
为认真的迭代而生。抬高运行上限、延长历史、自定义执行假设。Built for serious iteration. Higher run limits, longer history, custom execution assumptions.
为同时运行参数扫描、Walk-Forward 和 50+ 监控的研究者而生。Built for researchers running parameter sweeps, walk-forward, and 50+ monitors at once.
引导流程不会问你的收入。它只问你想做什么、关心哪些市场、想多亲力亲为——然后据此调配 Discover、Build 和 Monitor。The setup flow won't ask your income. It only asks what you want to do, which markets you care about, and how hands-on you want to be — then tunes Discover, Build, and Monitor to match.
随处可跳过。Skip anywhere.
你随时可以在设置里切换路线。You can switch paths in settings any time.
不会。我们是研究工具。监控只做仅观察或模拟交易。我们从不托管资金，也从不向交易所发送订单。No. We're a research tool. Monitors are watch-only or paper trading. We never custody funds and never send orders to an exchange.
样本量、前视偏差、成交量缺失、相对交易规模的滑点是否真实，以及执行风险提示——这些都是会悄悄让回测变得不可靠的东西。Sample size, look-ahead bias, missing volume, whether slippage is realistic for your trade size, and execution-risk notes — the things that quietly make a backtest unreliable.
Pro 可把回测结果导出为 CSV。Pro+ 可导出 PDF 和完整的可信度报告。无论哪种，你的策略数据都是可移植的 JSON。Pro exports backtest results as CSV. Pro+ exports PDF and a full trust report. Either way, your strategy data is portable JSON.
回测在我们的服务器上运行，但 app 会缓存最近的运行、编辑器和你的策略库，所以你在飞机上也能浏览和微调。Backtests run on our servers, but the app caches recent runs, the editor, and your strategy library, so you can browse and tweak on a plane.
可以。到期后你会降回免费版。你的策略都会保留，变的只是上限。我们从不删除你的成果。Yes. You drop back to Free when the term ends. Your strategies stay; only the limits change. We never delete your work.
有——Pro+ 订阅者会自动加入抢先体验通道，并可随时在设置里退出。Yes — Pro+ subscribers are automatically added to the early-access channel, and can opt out in settings any time.
永久免费。五个策略。每月二十次回测。无需信用卡，没有付费墙。Free forever. Five strategies. Twenty backtests a month. No credit card, no paywall.