Ask Claude, ChatGPT, Cursor or any other AI assistant who you should draft, and it can now see our entire 2026 board. Not a preview. Not the top 25. All 330 players, with our projection, consensus ADP, the draft round, the floor and the ceiling, and our sleeper or bust call on every single one.
There is no key. There is no email. There is no account. You connect it once and your assistant can read the board for the rest of the season.
What you actually get
- The full 330-player board in standard, half-PPR or full PPR
- Every sleeper and every bust our model disagrees with the market on — 92 calls, all of them
- All 154 players whose draft cost swings between ESPN, Sleeper, Yahoo, CBS and Fantrax, so you know where a guy goes cheapest
- Best available at any pick, with the players already gone removed
- "Who do I draft" — a straight answer for your roster at your pick, with the alternatives
The honest part
We used to cap this. An anonymous caller got 25 rows and a note asking for an email to see more. We ran it for three weeks and watched what happened: a few thousand capped responses, three people clicked through to the site, and exactly one gave us an email.
One.
So the wall was not selling anything. What the server was doing was reaching people — over a thousand distinct assistants pulled from it in three weeks. We were protecting revenue that did not exist, and charging for it with the only thing we actually generate.
That is a bad trade, so we stopped making it. Here is the whole board.
Why a cap was worse than it looked
There is a second reason, and it is the one that actually decided it.
Our model's opinions do not live at the top of the board. Consensus is right about the first two rounds — of the top 50 players by ADP, only three carry a value flag at all. The disagreements are deep: 43 of the 94 players past ADP 250 carry one.
That is not a quirk, it is what a sleeper is — someone the market drafts later than it should. Josh Palmer comes off the board around pick 394. Noah Gray around 459. Keon Coleman 303.
Which means any cap at a sensible depth withholds almost exactly the part worth having, and hands out the part you could get from any free ADP list. A capped board was the least useful version of this product we could have shipped, and we shipped it for three weeks.
What is still behind the subscription
Pro adds the judgment on top of the board: the week-by-week boom and bust odds on every player, where the positional tiers actually break, and the reasoning behind each call. The board tells you who. Pro tells you what it means.
And the markets side — where our NFL model disagrees with live Kalshi prices, the mispricing scanner, the commodity edges — stays Pro. That is where the work is and that is where the subscription lives.
Connect it
claude mcp add --transport http predictionmarketspicks https://predictionmarketspicks.com/api/mcp/mcpFor Claude.ai, ChatGPT, Cursor or anything else that takes a custom connector, add that same URL. Nothing installs. It is a hosted server, so the tools just appear — and the next time you ask "who should I take at 27," your assistant answers off the real board.
If you want the longer walkthrough with screenshots, we wrote one: plug the board into Claude, ChatGPT or Grok.
Every player has a page now
One more thing shipped with this. Every player on the board has his own page — the projection, the ADP, the floor and ceiling, his team's full schedule, and any live Kalshi prop markets on him. When your assistant names a player, it has somewhere to point.
Why you can trust a free board
Because you can check it. Every call the model makes is graded in public at /track-record — wins, losses and net result, per engine, including the engines that lose money. We publish the misses.
That is the reason to trust this, and it is not that it costs nothing.
The board updates through the season. Fantasy projections are estimates, not predictions. Prediction markets carry risk; contracts can settle at $0.