We grade every tracked account against real price action. Follower count barely moves with accuracy. And the biggest accounts are the least accurate band.
Across 137 ranked accounts, the correlation between follower count and accuracy is −0.21. Audience size is not skill.
Accounts under 25k followers that sit in our top 25 by score: accuracy plus how much their winners actually moved.
Four reproducible steps. We grade the call against the tape, not the person.
An LLM reads every public post and pulls the ticker, the direction (buy or sell) and the timestamp. Jokes and commentary are filtered out.
Each call is checked against actual market price over fixed windows, from 24 hours to a year. Every window has its own bar to clear before it counts as a win.
Win rate is wins ÷ resolved calls, straight from the price data, including the calls a trader later deletes. We keep the receipt.
The leaderboard blends how often they're right with how much the winners moved, so a lucky one-off can't outrank a long, consistent record.
Click any call and you see the original post and the exact price it settled against. Nothing is hand-picked.
We ingest every X (Twitter), Reddit, and StockTwits trader, then score their calls with Claude, OpenAI, and Gemini. The receipts you see below all start here.
每个人都由算法根据准确率和回报排名。


Snitchboard 是一条实时信号流,显示经过验证的喊单者正在涌入哪里。我们根据每位交易者的真实战绩评分,所以当准确的喊单者聚焦于某只股票,你能在行情启动前就看到人群形成。
Ticker Pulse 按有多少追踪者在喊单为每只股票排名,并告诉你谁在这只股票上一直准。输入任意股票,看结论。


相同数据,短线视角:按回报率计算盈利。仅做多、按市价计算、等同真金白银。
Your watchlist and the traders you follow, on one board — who's buying, who's selling, what moved overnight, and the volume behind it. Read it in ten seconds, before the open.


Every ticker arrives fully worked up: the catalysts driving it, insiders buying and selling straight from SEC filings, the earnings and analyst flips, and a plain-English read on the latest 10-K. The homework, already done.
Product launches, contract wins, guidance, M&A, litigation — mined from filings, company news, and the traders themselves, rolled up per ticker, per day.
SEC Form 4 buys and sells, with planned 10b5-1 sales separated from the discretionary ones that actually mean something.
The earnings calendar with beat/miss receipts, consensus counts, and the upgrade/downgrade flips with price targets — so a "beat" has context.
The latest 10-K / 10-Q distilled to what actually changed — not a 200-page PDF.
Fourteen of the most respected screens in investing — Munger, Piotroski, Altman-Z, Beneish, Graham, Lynch, Minervini — recomputed across the market every day. The homework a desk pays an analyst to grind out.


同一张图把交易者的喊单作为标记叠加在真实价格走势上。一眼就能看出这次喊单是命中了,还是被次日的行情碾过。
每条帖子在我们看到的那一刻就被存档。如果交易者事后删掉一次糟糕的喊单,我们仍保留截图、当时的价格快照和胜负结论。终身与个股准确率都按数量加权,所以 1 中 1 的侥幸不会和 200 中 300 的战绩并列。
当高准确率的喊单者在你关注的股票上发帖时,实时 Tip Line 立即发送邮件和网页推送,让你在喊单新鲜时就行动,而不是在行情走完之后。
每个档位看到的都是同样未经粉饰的准确率数据。付费版增加更大覆盖范围、实时提醒和更深的历史记录。Elite 版则配备完整的研究工作台。