Privacy policy

• Your account: your email address, your language, how you sign in (email link or Google), when you last signed in, and your email choices. With Google sign-in we receive only your email address and that Google has verified it: no name, photo, contacts or other Google data.
• What you do on the site: the rules you run and their results, the rules you track forward, the holdings you save for the stock check, the sentences you type into the rule box and the messages you type in the rule chat, and the questions you ask the event search.
• Payments: if you buy Pro, your plan, what you paid for (the months, currency and amount, and any discount code), the date it runs to, and our payment provider's reference numbers for you, your subscription and each invoice or refund. Your card or wallet details stay with the payment provider: we never see or store them.
• Emails: which emails we sent you, and whether you opened them (a small image in the email). If you join the waitlist, your email address and the plan you chose.
• Pages and features you use: which pages you open and which features you use (for example the stock check, running or tracking a rule, sign-in), counted on our own server to decide what to improve. Each record also keeps the site you came from (its domain only, never the full address), whether you used a phone, tablet or computer, and the language of the page. Each visitor is a code that changes every day, made from your IP address and browser and impossible to turn back into either; we keep neither. When you are signed in, these records carry your account and whether it was on Free or Pro at the time, and on the day you sign in, that day's other records can be linked to it through the day's code. On the plans page we also note which of our pages linked to it. To learn what leads people to Pro, we look at an account's records together with whether and when it paid (never card or wallet details). Campaign tags in a link (utm source, medium and campaign) are kept only when they are plain tags, and if the link came from an ad we note that it did (yes or no, never the ad network's click id). From these and the site you came from we group visits into channels, such as search, social or ads. For the stock check we keep the stocks it found, and the words of a search only when it found nothing we cover. Crawlers and AI agents are counted separately by name. We also count: which link in the daily email was clicked (a tag on our own links), a thumbs up or down you give a report or a backtest, the reason you pick if you turn your renewal off (optional, from a fixed list), and a research report closed before it finished (how many seconds it took). On the plans page and at checkout we also note whether a discount code came with it (yes or no, never the code).
• AI clients and scripts using our MCP server or API: for each call, which tool was used, whether it worked and how long it took; our own event, index and market ids and the stock codes our tables matched (never the words or the conversation sent to the AI); the client program's name and version as it reports them; the percentage and number of days a study asked for; and a random session number we issue when it connects, so one session's calls can be counted together (a client that keeps reusing it links its calls across days). Like other records these carry the day's visitor code, never an IP address, and follow the same 395-day limit.
• Technical data: your IP address, used to limit how fast anyone can call the site (kept only in the server's memory, never written to disk, and gone at the latest when the server restarts) and stored with each sign-in link until a day after it expires; when you run Fundamental Research before signing in, a random code in your browser's cookie and a keyed hash of your network (not the address itself), kept to count the free report; and the site's error logs, read only to fix problems.
• Cookies: two. One keeps you signed in, protects the site's forms, completes a sign-in and holds the free-checkup code above; one remembers your language. No advertising, analytics or tracking cookies, and no third-party scripts on our pages.

A screenshot you upload to the stock check is read in memory to find the stock codes in it, then discarded: it is never stored.

• Google Cloud hosts the site and its database in the United States, and runs Google sign-in.
• Resend sends our emails.
• Airwallex takes payments for Pro on its own checkout page. When you start a checkout we send it your email address, your account's id with us, the plan, currency and language you chose and any discount code. It collects your card or wallet details itself, tells us which invoices were paid or refunded, and keeps its own records of your payments under its own privacy policy.
• Anthropic (Claude) reads the sentences and rule-chat messages you type (with the rule they change, in words), the screenshots you upload and your event questions, to turn them into rules, stock codes and research. For Fundamental Research it gets only the stock and our scores for it, and for the daily email only a market or stock symbol, a date and that day's move, never anything about you. Anthropic's commercial terms do not allow it to train its models on them.
• Cloudflare runs our domain's DNS and forwards mail sent to privacy@plainquant.com.

Each processes the data only to provide its service to us.

• Your account and what you saved: until you ask us to delete them.
• Payment records: kept for 7 years after the payment, as Hong Kong tax rules require, then deleted. If your account is deleted, the records of invoices paid and refunds stay, no longer linked to it.
• Records of sign-in emails: 30 days. Sign-in links: a day after they expire.
• Sentences you typed (rule-chat messages included) and event questions: 365 days.
• Records of pages and features used: 395 days, then only daily totals with no visitor or account.
• Forward tracking pauses after 90 days without a sign-in.
• Backups: deleted after 30 days.