PHX
Away
MODEL
61.2%
DEN
Home
Model predictions
get_sport_predictions
Win probabilities, fair lines, model edges, and league-specific analytics for every loaded slate.

Model Context Protocol
Bring your prediction agents, pick history, model estimates, and market context into compatible AI chats and coding agents through one secure connection.
You’ll paste this URL into Claude in the next step.
Open Customize → Connectors → Add custom connector. Name it WagerProof, paste the URL, and leave Advanced OAuth fields blank.
Open Claude ConnectorsApprove WagerProof access, then ask Claude about a game, model estimate, market price, or one of your agents.
Start in ClaudeRemote Streamable HTTP · OAuth 2.1 sign-in · 13 tools
Setup guides for Claude, ChatGPT, Cursor, Claude Code, OpenClaw, and Hermes. Availability depends on client version, plan, and workspace permissions.Check the live connector docs
I’ll load today’s NBA model estimates, compare them with prediction-market prices, and sort the largest probability gaps first.
The model and market data are ready. I found three matchups with a gap above four percentage points.
WagerProofget_sport_predictions
Scan today's NBA slate and show me where WagerProof's model differs most from the market.
The largest model-to-market differences are below. These are analytic estimates, not guarantees or betting advice.
Illustrative MCP response
3 matchups

+6.8 pp
largest gap
Phoenix @ Denver
Model 61.2% · Market 54.4%
Miami @ New York
Model 57.9% · Market 52.8%
Dallas @ Minnesota
Model 55.6% · Market 51.3%
Illustrative UI using the live WagerProof connector.
Give a compatible assistant secure access to WagerProof model research and your own account data.
7 tools
Live connector catalogPHX
Away
MODEL
61.2%
DEN
Home
get_sport_predictions
Win probabilities, fair lines, model edges, and league-specific analytics for every loaded slate.
58.4%
Model
−1.7
Vs. market
2
Injuries
get_game_detail
A complete model, matchup, and prediction-market breakdown for one specific game.
Polymarket
PHX @ DEN
54.4¢
get_market_odds
Compare WagerProof estimates with Polymarket-implied probabilities, grouped by game.
124
sample games
57.3%
historical rate
query_sports_database
Test bespoke historical theories with one bounded SQL query and return aggregate findings plus illustrative games.
Agent performance
+10.2u
get_agent_performance
Compare win rate, units, streaks, and sport or bet-type performance for your agents.
Following
Quant Lab
62–44
Tempo Reader
41–30
Data Edge
55–49
Split Pro
29–24
list_my_follows
List followed public agents with their archetype and current win-loss record.
55.8%
WIN RATE
86 graded picks · informational tracking
get_my_record
Summarize wins, losses, pushes, win rate, sport splits, and shared wins.
WagerProof uses MCP (Model Context Protocol), an open standard that lets compatible AI clients call external tools. Add the remote connector URL, sign in with WagerProof, approve access, and the AI can request WagerProof analytics inside the conversation.
This tutorial includes custom-connection guides for Claude, ChatGPT, Cursor, Claude Code, OpenClaw, and Hermes. WagerProof uses remote Streamable HTTP MCP with OAuth 2.1; availability, menu labels, and workspace permissions can vary by client, plan, and app version.
There are 13 tools in the current connector catalog: public model predictions, game detail, game search, prediction-market prices, and editor analyses; signed-in sports schema and custom database research; plus private tools for your agents, their performance and picks, followed agents, community activity, and tracked record.
You can scan a league’s slate, investigate a specific matchup, compare model probabilities with market prices, test bespoke historical theories against the sports database, review published analyses, and summarize your own agents, followed agents, community activity, or tracked history.
No API key is required. The connector uses OAuth: your AI sends you to WagerProof, you sign in, and you approve access. Your password is never shared with the AI client.
Private tools run as the signed-in user and are scoped by WagerProof’s row-level security. Public sports tools use global data, while personal agent and record tools can only read data available to the authorized account.
No. It cannot create or edit agents, publish picks, place wagers, move money, or interact with a sportsbook.
No. WagerProof reports model estimates, historical performance, and market information for analysis and research. Sports outcomes remain uncertain, and the connector does not provide guarantees or betting advice.
Yes. Any compatible remote-MCP client can connect to https://wagerproof-mcp.habib225.workers.dev/mcp. Name the connector WagerProof, leave advanced settings empty unless your client requires them, and complete the WagerProof OAuth flow when prompted.