Pick of the Day publishes up to twenty selections on Polymarket. Multiple selections can share a game and share exposure. Each published card records the side and price. We post each result when the market settles.
The selection marked Free. Card numbers show publication order, not quality. Sign in, and at release it shows the side, the frozen price, and the wallets behind it: the same card Pro sees. Pro opens five picks in total, including the designated free selection. Max opens every published pick, up to twenty. Without access, game details stay hidden until resolution.
We review Polymarket market and trader information before publishing a pick. The selection method is private. Each published card shows the side, recorded price and available trader facts.
We publish a pick after reviewing the available market information. Our decision rules remain private. The published market, side, price and eventual result are available for you to check.
Trader positions and recent activity are available as market context. We do not disclose how individual inputs affect selection.
The cards show Polymarket wallets and their recorded grades, positions and available sport statistics. A grade describes a trader’s settled record; it does not guarantee a pick’s result.
The optional specialist block shows a trader’s recorded position and available statistics in the named sport. Those facts describe the trader, not the pick’s probability of winning. Positions can change after the recorded snapshot.
We cover Polymarket sports and esports markets, including Moneyline, Spread, Total and props. Available picks depend on the markets covered that day.
The day of its kickoff in US Eastern time. A day runs from midnight to midnight Eastern, and the archive, the permalinks, and the email all key off that date. This page shows drop and kickoff times in your device's timezone.
New picks can arrive throughout the game day. For a scheduled pick, use the release time shown on the page. The published record shows when a pick became public alongside its kickoff time.
Today's scheduled selections that have not yet been released. An entitled viewer sees the release time and kickoff. Other viewers see a locked slot and an upgrade action. Our server withholds the game, side, price, and identifying clocks until you have access or the pick resolves.
A scheduled slot does not guarantee publication. Use the latest page status and release time. The published record carries the recorded market, side and price.
We do not publish a pick every day. A day without published picks does not enter the record.
The frozen one. We capture the backed price when the pick publishes, and it is written once and never rewritten. The market keeps moving after that, so your entry can differ from the price on the card.
Any move before publish is already in the price on the card, because the record prices every pick when it publishes, not when the wallets bought. The recorded publication price is separate from any trader's earlier entry. Where the capture succeeds, the market's last price before kickoff sits next to the frozen price, so each measured pick shows whether the price moved after publish, and the stats page shows the average and how many picks it covers. In our timing study, the median quoted spread on large sports buys was 1¢. Your own fill and Polymarket's taker fee come out of any edge, and your entry can differ from the price on the card.
The market decides. We read the outcome the market resolved to and grade the pick against the side we backed. A market that settles with no winner, such as a cricket match with no result, is a void: the pick stays on the record and counts as neither a win nor a loss. A scheduled job re-reads every published pick and posts the result.
In two cases: its game is explicitly cancelled, or the pick was published after its game had already started. A pick published after the start, including one on a match that was suspended and resumed later, was never a pre-game pick, so it does not count. A withdrawn pick leaves this page, the archive and every figure in its statistics. The stats page lists each withdrawn pick by name, with its reason and result. The public API ledger keeps its original publication record and any existing proof. Withdrawal does not settle the market: Polymarket's recorded resolution still decides win, loss or void. A postponed game's result stays pending until the market resolves. We never guess a result.
Only when the market's recorded resolution changes. A scheduled job re-reads every published pick, and a pick whose market was corrected to a different outcome follows that correction. A pick can never disagree with the market's settled result.
Wins divided by decided picks included in the track record, where decided means wins plus losses. Withdrawn recommendations, voids and picks still pending do not count, so an open pick does not move the rate before it settles.
The figures model a flat $1,000 stake on every win and loss in the track record that carries a frozen price, filled at that price, before Polymarket's fees. Voids and withdrawn recommendations are excluded. One unit is one $1,000 stake, so +0.61 units means a $610 gain. ROI divides the profit by the total stake, and profit is the sum of those returns minus the stakes. These figures do not establish real fills or subscriber profit.
Closing line value: how the frozen price compares with the market's last price before kickoff, or before Polymarket's live feed first showed the game in progress. A positive CLV means our side's price rose between publish and that close. That is the standard test of whether a call had edge before the result was known. Each measured pick shows its CLV and the numbers behind it. The average on the record is the mean move in percentage points, the close minus the frozen price, so a cheap pick and a favorite count on the same scale. A pick whose close could not be captured shows CLV as unavailable, and the stats page shows how many picks are measured. A pick published after kickoff is marked not applicable.
On the stats page. It breaks the record down by sport, day, month, and entry price, with units, ROI and profit before fees, win rate, and CLV. Withdrawn recommendations are excluded from every breakdown and count. Public proof coverage is shown separately from the product track record.
Yes. The archive links to /pick-of-the-day/YYYY-MM-DD pages for days with included picks. A past day shows those picks, the frozen price, and the result. Withdrawn picks remain in the public API ledger.
A loss is not a reason to remove a pick. The archive keeps losses and voids for recommendations that have not been withdrawn. Withdrawn picks leave its rows and statistics, but their original publication records and any existing proofs stay in the public API ledger. The 0xinsider/picks Git mirror records the picks it receives, including losses. The offline verifier cannot establish that a pick missing from both records it checks was ever published.
Open the public 0xinsider/picks Git mirror. For a pick whose hash reached it before kickoff, recompute the hash from the revealed pick and nonce, then inspect the first commit that carried it. Older picks, mirror outages, and late public hashes have no pregame proof in that mirror. Their publication records remain in the public API ledger. The verifier compares git author dates with kickoff, but those dates are writer controlled. New Seal commits get signed receipts for their exact SHA; only a verified witness time before kickoff strengthens timing evidence. An independently retained pregame copy can do the same. The pick email goes out at release with the matchup in its subject on a mail server we do not run.
No pick is sold on its own, and no pick carries a fee on a win. Pick of the Day is one part of a Pro subscription that you can cancel at any time. A scheduled job reads every result from the market's resolution; we do not enter it. The archive shows losses and voids for recommendations that have not been withdrawn. The public API ledger keeps withdrawn publication records. Settled results and public proofs are open without paying, so the check costs nothing.
Sign in to open the selection marked Free: the side, frozen price and wallets behind it. Every resolved pick and result is public. Other unresolved picks keep their game and backed side hidden. Pro opens five picks in total, including the free selection; Max opens every published pick, up to twenty.
Three ways: an email when an entitled pick goes live, a bell notification in the app, and this page. A signed-in account without Pro gets the email when the designated free selection goes live. Pro receives its five entitled picks; Max receives every published pick up to twenty. The page opens entitled slots at the drop. The toggle above the pick turns the email and the bell on or off together, and so does your account page. The email names the matchup only, and our Discord posts the date and a link to the pick; neither shows the side.
Yes. GET /api/v1/pick-of-the-day returns today's frozen picks and GET /api/v1/pick-of-the-day/archive returns the record, both to a Pro API key. Before release the first returns a 404 with a retry_at time, so schedule the next call instead of polling.
Yes. The MCP server carries get_pick_of_the_day and get_pick_of_the_day_archive, the same two reads as the API. Run npx -y @0xinsider/mcp for Claude Code, Cursor, or Codex, or point a hosted agent at the remote endpoint. Both need a Pro API key.
No. Pick of the Day shows where profitable wallets are positioned, and a position is not a guarantee. Prediction markets carry risk. Do your own research, and never risk more than you can afford to lose.
