Quick Start

1. Click any market in Markets -- all 7 panels update: order book, chart, 9 quant models, 3D surface, AI verdict, news, holders, whale flow
2. Click the BRIEF tab in Depth for a one-page trading briefing: BUY/LEAN/PASS verdict, edge metrics, signal consensus, liquidity check, smart money positions, risk flags, and action cards
3. Open ALPHA for AI-generated prediction cards with quant signals, edge detection, and conviction scoring
4. Click FORESIGHT in the ALPHA panel -- AI scenario engine simulates two parallel futures per market with research, timelines, actors, and trading edges
5. Click Deep Analysis in the Analysis panel to upgrade any market to Claude Haiku 3.5
6. Use the VWAP calculator in Book tab to estimate fill price before trading
7. Check Autopsy in the Flow panel to forensically analyze any trader's wallet
8. Switch modes: General (2,500+ markets), Elon (tweet brackets), War/Peace (geopolitical risk), Nexus (cross-panel intelligence)
9. Press / to search, ? for tour, r to refresh, keys 1-7 to jump panels
10. Use CAL tab to tune all model parameters, or click presets (Conservative / Balanced / Aggressive)

The 7 Panels

Markets

Primary market list. Two modes: Hot (sorted by 24h volume) and New (sorted by start date). Sortable by any column. Click a market to load everything. The bottom ticker scrolls the most active markets, deduplicated and diversified across sectors.

Depth (7 tabs)

Shows the currently selected market name below the tabs. Seven progressively deeper views:

• Book -- Mirror order book. YES bids on the left (green bars, right-to-left), NO bids on the right (red bars, left-to-right). Liquidity stats, spread, and cost-to-move-5% at the top. Below the book: a VWAP Fill Estimate calculator -- type any dollar amount (or use the $50/$100/$500/$1K presets) to see your exact fill price, shares, and slippage from walking the real order book levels. Shows both BUY YES and SELL YES sides with level-by-level breakdown
• Chart -- Canvas-rendered price history. Intervals: 1H, 1D, 1W, MAX. OHLCV stats overlay with high/low/current markers
• Models -- Nine quantitative models run on real data (see Quantitative Models section below). Includes composite probability from 11 signal sources, Bayesian posterior chain, LMSR price impact, base rate priors, and cognitive bias detection
• 3D -- Interactive 3D liquidity surface. Drag to rotate, double-click for auto-rotate. Built entirely from real order book data with Gaussian kernel interpolation. Green = bid-heavy, red = ask-heavy, amber = neutral, blue = deep liquidity. Pillars show liquidity walls, holder conviction, and cost-to-move markers. HUD overlays show market structure (left) and signals (right)
• CAL -- Calibration engine. Three presets plus manual slider tuning for every model parameter. Per-sector overrides. Import/export settings as JSON. When CAL is active, the AI Picks panel switches to the Live Prediction Modeler
• SIM -- Advanced simulations: Particle Filter (sequential MC with 200 particles tracking price history), Tail Events (importance sampling with variance reduction for crash/spike scenarios), Correlated Markets (t-copula joint probability for same-sector markets)
• BRIEF -- One-page trading briefing that consolidates all available data into an actionable view. Seven sections: Verdict (BUY/LEAN/PASS with composite 0-10 scoring from edge magnitude, Kelly, AI agreement, signal consensus, and KL divergence), Edge Metrics (EV, Kelly, KL, Bayesian, Monte Carlo, Theta in a 2x3 grid), Signal Consensus (all 10 enriched signals as color-coded traffic light rows sorted by strength), Liquidity Check (spread, VWAP at $100/$500, slippage estimates from real book walks), Smart Money (whale trades on this specific market), Risks and Biases (flags for extreme price, time risk, recency, anchoring, base rate neglect, liquidity), Bottom Line (side-by-side YES/NO action cards with entry targets, half-Kelly sizing, EV/ROI, plus key risk and catalyst). Zero new API calls -- reads entirely from existing terminal state

Intel (4 tabs)

Research and intelligence for the selected market:

• News -- AI-matched articles with colored sector tags. Click any tag to filter. Each article links to related markets
• Chat -- Community comments from Polymarket
• Holders -- Top 10 YES and NO holders in a mirror layout (like the order book). Bars show relative position size. Click any holder to open their Polymarket profile. Summary shows total shares and open interest
• Dates -- Upcoming macro events: FOMC decisions, jobs reports, CPI, earnings, crypto milestones, elections. Sorted by days remaining with urgency badges

AI Analysis

AI-powered analysis of the selected market. The intelligence brief appears at the top with edge assessment, timeline, momentum, smart money, data quality, and contrarian view. Below that: verdict (STRONG YES through STRONG NO), confidence score with bar, summary, key factors, and news impact with directional arrows. A Deep Analysis button upgrades to the premium model for higher-quality reasoning.

Sectors (4 tabs)

• Grid -- 8 sector tiles with market count, average YES%, and total volume. Click to filter the Markets panel
• War -- Geopolitical tension barometer. 5 zones: Ukraine/Russia, Israel/Gaza/Iran, China/Taiwan, North Korea, Other. Score levels: CRITICAL, HIGH, ELEVATED, MODERATE
• Sports -- League Heat Index. 6 zones: NFL, NBA, Soccer, NHL, UFC/Boxing, Other. Levels: HOT, WARM, ACTIVE, QUIET
• Ent -- Entertainment Buzz Index. 6 zones: Film/Oscars, Music/Grammy, Streaming/TV, Social Media, Celebrity, Other. Levels: VIRAL, TRENDING, ACTIVE, QUIET

All barometers use the same methodology: filter markets by sector, group into zones via keyword matching, calculate volume-weighted YES% as the score, and display top 4 markets per zone. Click any market row to select it.

Flow (3 tabs)

• Trades -- Live whale trades across all markets. Filter by size ($250+, $1K+, $5K+, $10K+) and side (All, Buy, Sell). Click any trade to select that market
• Leaders -- Top traders by PnL. Periods: Day, Week, Month, All Time. Each trader shows sector tags from Polymarket category leaderboards (Sports, Politics, Crypto, Culture, Science, Business). Click to open their Polymarket profile
• Autopsy -- Forensic wallet analysis. Two layers: an instant overview (populated from data already loaded) and a deep forensic scan (server-side analysis of all market trades). See the Autopsy Methodology section below for full details

ALPHA (3 modes)

AI-powered market intelligence, scenario analysis, and recommendation engine. Three modes:

• Packs (default) -- AI-generated prediction cards ranked by edge detection, Kelly signal, and conviction. Each card shows market question, side (YES/NO), entry price, conviction (1-5), rarity tier, quant data (price, fair value, edge, Kelly, MC path stability %), signal badges, and volume/liquidity stats. Rarity is deterministic -- higher quality picks (more signals, larger edge, stronger Kelly) earn higher tiers
• Foresight -- AI scenario engine. Every 6 hours, selects the top 3 tradeable markets, gathers live research via Firecrawl, and runs Gemini Flash to generate two parallel futures per market: what happens if YES vs NO. Each scenario includes trigger event, timeline with dates, key actors, cascading effects across domains (oil, defense, currencies), signals to watch, risk factors, and confidence rating. The Edge tab shows trading thesis, Kelly sizing, and signal strength composite. Click any card to open the full Foresight overlay with 4 deep-dive tabs: Timelines, Research, Data, Edge
• Live Prediction Modeler (when CAL tab is active) -- Interactive what-if tool. Four sliders: estimated probability, daily volatility, days to resolution, and order size ($). Adjusting any slider recomputes Kelly, EV, Monte Carlo, and price impact in real-time with a live histogram. Shows post-impact price, post-impact EV, and a Trade Viability Score (0-100) that combines edge strength, impact ratio, bias penalty, and Kelly confidence. TRADEABLE > 60, MARGINAL 30-60, PASS < 30

Composite Probability Estimate

The Models tab derives an estimated true probability by combining 11 independent signal sources. Each source shifts the estimate away from the market price when evidence supports it:

Confidence scales with the number of available sources. 8+ sources = full confidence. Fewer sources = proportionally reduced shift. Final estimate is clamped between 1% and 99%.

Quantitative Models

Kelly Criterion

Optimal bet sizing: f* = (bp - q) / b. Four levels: Full Kelly (max growth), Half Kelly (recommended), Quarter Kelly (conservative), and Empirical Kelly (uncertainty-adjusted). Signal classification: STRONG MODERATE MARGINAL NO EDGE

Empirical Kelly applies a CV (coefficient of variation) haircut to Half Kelly: f_empirical = f_half * (1 - CV_edge). More confirming signals = lower CV = sizing closer to theoretical. Fewer signals = higher uncertainty = more conservative sizing. Based on institutional methodology from Becker (2025) prediction market microstructure research.

Bias Zone Detection

Based on empirical analysis of 72.1M+ trades (Becker 2025). Markets under 10% implied probability are flagged as Longshot Bias Zone: 1-cent contracts win only 0.43% vs 1% implied (-57% mispricing). Takers buying YES at extreme low probabilities lose at systematically elevated rates. Markets 10-20% get a moderate warning. Markets above 90% get a favorite bias flag. This is the calibration bias that hedge funds exploit.

Taker Cost Warning

Every trade has a maker (limit order) and taker (market order). Research proves takers exhibit negative excess returns at 80 of 99 price levels. The order book now shows taker cost classification (LOW/MODERATE/HIGH) based on spread width. Wide spreads mean market orders cost more edge. Use limit orders when possible.

Monte Carlo Simulation

1,000-5,000 simulations (tunable) of bounded random walk price paths in [1%, 99%]. Box-Muller normal distribution with antithetic variates and stratified sampling for variance reduction (~4x effective quality). Outputs a 20-bucket histogram, percentile ranges (5th/25th/50th/75th/95th), paths above 50%, mean and standard deviation. Note: this simulates price noise, not resolution outcomes -- real-world events determine if a market resolves YES or NO. Volatility sourced from enriched data when available, falls back to calibration default.

Expected Value

EV per dollar wagered for both YES and NO sides. ROI calculation, breakeven probability, and edge detection (fair value minus market price).

Implied Theta Decay

Daily price erosion toward resolution. Daily theta (cents/day), implied annualized volatility, time value remaining, and convergence rate.

KL Divergence

Information-theoretic edge metric shown in the EV card. KL = p*log(p/q) + (1-p)*log((1-p)/(1-q)) where p = model estimate and q = market price. Unlike flat edge (model - market), KL auto-adjusts for extreme prices: moving from 5% to 15% is a bigger information shift than 50% to 60%. Signal levels: STRONG KL > 0.05, MODERATE KL > 0.02, WEAK KL < 0.02.

VWAP Fill Estimate

Live pre-trade calculator in the Book tab. Type any dollar amount and instantly see your real volume-weighted average fill price by walking actual order book levels. Shows BUY YES (crossing asks) and SELL YES (hitting bids) side by side. Each side shows: VWAP vs best price, total shares acquired, slippage percentage (LOW < 2%, MOD 2-4%, HIGH > 4%), and a level-by-level breakdown showing how your order eats through the book. Flags PARTIAL when the book can't fill your full order. Essential for estimating real execution cost before entering a position.

LMSR Price Impact

Fits an effective LMSR (Logarithmic Market Scoring Rule) liquidity parameter from the live order book. Shows your price impact at $50, $100, $500, and $1,000 order sizes. Liquidity classification: DEEP (b > 5000, institutional-grade), MODERATE (b 1000-5000), THIN (b 200-1000, significant impact), ILLIQUID (b < 200, your trade IS the price movement). If your order size exceeds 30% of effective-b, your edge is partially illusory because you move the price against yourself.

Bayesian Posterior Chain

Sequential Bayesian updating. Starts from a prior (market price or sector base rate) and chains through each signal as a likelihood ratio update via Bayes' rule. Signals: whale flow (LR 1.8 accumulating / 0.55 distributing), momentum (1.5 / 0.67), smart money (2.0 / 0.4), buy pressure (1.6 / 0.5), volume-price divergence (1.7 / 0.55), holder conviction (1.3 / 0.8). The card shows each step of the update chain with colored arrows. Combined LR > 3.0 = strong multi-signal confirmation. LR < 0.3 = strong negative convergence.

Base Rate Priors

Historical frequency anchors for event categories. Used as the Bayesian prior when a sector pattern matches the market question. Examples: incumbent re-election 60%, FDA Phase 3 approval 85%, recession probability 15%, fed rate cut 40%, ceasefire/peace deal 25%, beat economic estimate 52%. When the market price deviates more than 20% from the base rate, a BASE_RATE_NEGLECT warning fires -- the crowd may be ignoring statistical history.

Cognitive Bias Detection

Flags 5 psychological traps that distort market pricing. Each bias gets a severity badge (HIGH / MOD / LOW) and actionable advice:

• BASE_RATE_NEGLECT -- Market deviates >20% from historical base rate. The crowd may have valid reasons, or may be ignoring statistics
• SUNK_COST -- Position thesis has reversed but the market hasn't corrected. Evaluate exit regardless of current P&L
• ANCHORING -- Price sitting at a round number (25%, 50%, 75%) with low volume. May not reflect true probability, just a psychological anchor
• AVAILABILITY -- Volume spike without a news catalyst. Could be social media noise or herd behavior, not information
• RECENCY -- Large recent price move exceeds typical daily range by 2x+. Mean reversion is likely; the move may be an overreaction

When 2 or more HIGH-severity flags appear on a market, treat it as a strong caution signal.

AI Prediction Cards

AI-generated market intelligence cards. Click the Packs tab in the ALPHA panel to generate a pack of 5 cards on demand.

How Packs Are Built

1. Top 40 markets scored by volume, liquidity, price uncertainty, time-to-resolution, and edge potential
2. 5 sector-diversified markets selected (max 2 per sector, no duplicates from existing collection)
3. Each card gets Bayesian fair value estimation (shrinkage toward 50% based on volume evidence), half-Kelly sizing, Monte Carlo price stability, and 8 signal types detected from market characteristics
4. Rarity assigned deterministically from pick quality -- not random. Quality score = signal count + edge magnitude + Kelly strength

Rarity Tiers (Deterministic)

Quality score = (signal count * 2) + min(10, |edge|) + min(5, kelly/2). Better AI insight = higher rarity. No luck involved.

Card Anatomy

Each card shows: market question, sector tag, side (YES/NO), entry price, payout multiplier, conviction (1-5 stars), rarity tier, 5-column quant grid (price, fair value, edge %, Kelly %, MC path %), signal badges (MOMENTUM, SMART $, BUY PRESSURE, HOLDERS, VOLUME, WHALE, EDGE, EXPIRING), volume and liquidity stats. Zero AI API cost -- all scoring is deterministic math from Polymarket data.

Foresight -- AI Scenario Engine

Foresight simulates parallel futures for the most tradeable prediction markets. Access via the FORESIGHT tab in the ALPHA panel.

How It Works

1. Market Selection -- Every 6 hours, scores the top 100 markets by volume, price uncertainty (peak at 50%), liquidity depth, time-to-resolution, and 24h price momentum. Selects the top 3 with sector diversity (max 2 from same sector). Filters out bracket markets and crypto
2. Research Gathering -- For each market, runs parallel Firecrawl news searches (8 articles, relevance-filtered by 2+ key terms), pulls cached enriched signals (whale flow, momentum, smart money, holder conviction), and checks cached breaking news
3. Scenario Generation -- Gemini Flash 2.0 receives the full research dossier and generates two parallel futures: what happens if YES resolves vs NO. Each scenario includes a trigger event, dated timeline with significance notes, named actors with motivations, cascading effects across concrete domains (oil, defense stocks, currencies), signals to watch, risk factors, and a confidence rating
4. Trading Edge -- Each market gets an AI-estimated probability, edge percentage vs market price, directional thesis, catalyst, and primary risk. The Edge tab computes Kelly sizing and a 10-point signal strength composite

4 Tabs

• Timelines -- Side-by-side YES and NO scenario columns. Each column shows: scenario title, trigger event, dated timeline entries, key actors with actions and motivations, cascading effects with magnitude badges (minor/moderate/major), signals to watch, risk factors, and a confidence meter
• Research -- 4-card grid: Research Facts (sourced claims with dates and tone), Sources (clickable article links), Market Signals (whale flow, momentum, smart money, holder conviction from enriched data), Market Info (volume, liquidity, sector, price)
• Data -- 4-card grid: Probability Gauge (AI vs market price visual), Signal Dashboard (all enriched signals as traffic-light indicators), Scenario Comparison (canvas chart comparing YES/NO probabilities and confidence), Research Sentiment (article tone distribution)
• Edge -- 4-card grid: Edge Meter (large percentage display with classification badge), Kelly Sizing (full/half/quarter/tenth recommendations), Signal Strength (10-point composite from price, volume, momentum, whale, smart money, holder, lifecycle, buy pressure, divergence, rotation), Trade Thesis (direction, thesis text, catalyst, risk, with COPY button for sharing)

Live Deltas

When terminal market data is available, Foresight compares its snapshot prices to live prices. Each market card shows price change (cents), and a CONFIRMING or DIVERGING badge indicating whether the live price movement aligns with the AI's directional call.

Data Freshness

Scenarios are cached for 8 hours. Updated automatically by cron at 00:00, 06:00, 12:00, 18:00 UTC. A stale banner appears when data exceeds 8 hours. Manual refresh available with cron key authentication.

Autopsy -- Forensic Wallet Analysis

The Autopsy tab reverse-engineers any trader's strategy on a specific market. It answers: who is trading this market, what are they doing, and should you care?

How it works

When you select a market, two things happen automatically:

1. Instant layer (immediate) -- Cross-references three data sources already loaded in the terminal: whale trades filtered to this market, holders (YES/NO token positions), and leaderboard rankings. Scores each wallet by "interestingness" and shows the top 8 as situation cards
2. Forensic scan (2-3 seconds) -- The server fetches all recent trades for the market, groups them by wallet, computes 6 forensic metrics for the top 8 active wallets, and classifies each trader. Results are cached for 3 minutes

Click any card or row to deep dive into a wallet's full forensic profile.

Situation Cards (Instant Layer)

Each card gets a tag based on what data sources it appears in and what behavior stands out:

Forensic Scan Classifications

The server analyzes trade patterns and assigns one classification per wallet:

Deep Dive: 6 Forensic Metrics

When you click into a wallet, the full forensic profile shows:

Position Summary

At the top: the market being analyzed, current market price, the wallet's dominant side (YES or NO), net share count, VWAP, and unrealized P&L in both percentage and dollar terms.

VWAP (Volume-Weighted Average Price) is the average price this wallet paid for their shares, weighted by trade size. If someone bought 100 shares at 40c and 200 shares at 50c, their VWAP is (100*0.40 + 200*0.50) / 300 = 46.7c. Compare VWAP to current price to see if the trader is winning or losing.

Time Profitable

What percentage of time this wallet has been "in the money" since their first trade. Calculated by checking the price history at every data point after their first trade and comparing to their average entry. Higher = better timing. Below 30% = consistently underwater.

Hedge Check

Detects if the wallet traded both YES and NO sides. If the combined average price (YES avg + NO avg) is below $1.00, the wallet is arbitraging -- they locked in a guaranteed profit regardless of outcome. The "edge" is 100% minus the combined average.

Bot Detection

Analyzes the time gaps between consecutive trades. Sub-30-second average = BOT. Under 2 minutes = LIKELY BOT. Over 5 minutes = HUMAN. Between = MIXED. Shows both average and median gaps (median resists outliers).

Accumulation Pattern

Compares average trade size in the first half vs second half of the wallet's history. INCREASING = scaling up (building conviction or doubling down). DECREASING = scaling down (taking profit or losing confidence). FLAT = consistent sizing.

Price Trend

How the wallet's entry prices changed over time. CHASING HIGHER = buying at progressively worse prices (FOMO signal). DOLLAR COST AVG = very tight price variance across all trades (systematic). DISCIPLINED = varied prices but not trending (opportunistic entries).

Current Status

Net position direction (YES or NO bias based on buy vs sell volume), VWAP, and unrealized P&L. Positive = currently winning. Negative = currently underwater.

AI Verdict

After the rule-based classification, you can request a deeper AI analysis. This sends all 6 metrics to Gemini Flash, which produces a classification, confidence score (1-10), one-line summary, and risk factors. The AI can catch nuances the rules miss, like a sophisticated strategy disguised as retail behavior.

Enriched Market Signals

When a market is selected, the enrichment API cross-references price history, trades, holders, and open interest to produce 9 signal categories:

Calibration Engine

The CAL tab controls every quantitative parameter. Three presets:

• Conservative -- 5,000 MC sims, tight edge thresholds (15%/8%/3%), low AI temperature (0.2)
• Balanced (default) -- 2,000 MC sims, standard thresholds (10%/5%/2%), moderate temperature (0.3)
• Aggressive -- 1,000 MC sims, loose thresholds (7%/3%/1%), high temperature (0.7)

Parameter groups: Monte Carlo (sim count, max days, vol fallback, Brownian multiplier), Kelly (strong/moderate/marginal thresholds), Momentum (slope, vol bounds), Volume (spike multiplier, ratios), Whales (min trade size, accumulation/distribution), Holders (HHI levels), AI (temperature, max tokens, model).

Per-sector overrides let you tune parameters differently for crypto vs politics vs sports. All settings persist to localStorage. Import/export as JSON.

3D Wave Methodology

The 3D surface visualizes the full order book as a terrain map. No synthetic noise. Every peak and valley comes from real order data.

• X-axis: Price (0% to 100%)
• Z-axis: Cumulative book depth (front = best prices near market, back = full book)
• Y-axis: Order size (height = liquidity at that price/depth level)

Construction: 32 price bins x 16 depth layers. Each order contributes to nearby bins via Gaussian kernel (sigma=0.002, radius=12%). Cumulative slicing: Z=0 shows top 6% of book, Z=max shows everything.

Color encoding: Green = bid-dominant, Red = ask-dominant, Amber = neutral, Blue = deep liquidity layers. Height modulated by holder conviction (HHI) and whale pressure (net flow). Spread gap carved at market price. Liquidity wall pillars at top 3 concentration levels. Holder conviction pillars (YES/NO shares). Cost-to-move-5% markers.

Controls: Drag to rotate. Double-click to toggle auto-rotation. Touch supported.

Elon Mode

A dedicated dashboard for tracking Elon Musk's X (Twitter) posting activity against Polymarket tweet-count bracket markets. Access via the ELON button in the mode bar.

What It Tracks

Polymarket runs weekly bracket markets on how many times Elon Musk will post on X during a 7-day period (e.g., "Elon Musk # tweets Feb 27 - Mar 6"). Each event has 15-30 brackets (140-159, 160-179, ... 580+). Elon Mode monitors his live tweet count and compares it to these bracket prices in real time.

6 Panels

• Heatmap -- Hour-by-day tweet activity grid for the active bracket period. Shows daily totals, hourly averages, and trend arrows. All data from live XTracker API
• Markets -- All Polymarket bracket events (past, current, future). Event tabs let you switch between periods. Each bracket shows YES/NO prices, tweets remaining, required daily rate, and volume. Projected bracket highlighted. Fully dynamic from Polymarket API: new events appear automatically
• Intel -- Five tabs: Posts (recent tweets with AI mood tags), Mood (sentiment breakdown), Events (brand impact analysis: Tesla, SpaceX, xAI, DOGE, Politics), News (Brave Search results), xAI (Grok-powered analysis)
• Charts -- 11 chart tabs: Moving Averages, Hourly Distribution, Cumulative Count, Day of Week, Distribution, Volume, Mood Timeline, Bias, Flight Patterns, Event Volume, Winners and Losers (edge vs market price)
• Activity -- Stats grid (rolling 24h/48h/7d counts, daily rate, fractional average), activity gauge, and daily posting chart
• Signals -- Three tabs: Bracket cards with modeler, 3D Bracket Probability Landscape (interactive surface showing market price vs fair value vs edge), Monte Carlo histogram (5000 sims with percentile lines)

Analysis Engine

• AI Analysis -- Two tiers: Analyst (Gemini Flash, fast) and Strategist (Claude Sonnet, deeper). Provides recommended bracket, confidence score, Kelly sizing, Monte Carlo projection, edge type, news impact, and risk factors
• Fair Value -- Normal CDF model using current tweet count, daily rate volatility, and days remaining to compute probability distribution across brackets. Compared to market prices to find mispriced brackets
• Monte Carlo -- 5,000 simulations using real historical daily count volatility. Outputs histogram with P10/Median/P90 percentile lines and bracket overlay

Data Sources

• Tweet counts: XTracker API (live), stored in Upstash Redis, updated by cron every 5 minutes
• Bracket markets: Polymarket Gamma API (live). Both open and closed events fetched dynamically
• Mood/sentiment: Keyword-based classification with AI enhancement via OpenRouter
• News: Brave Search API for recent Elon Musk posting activity context

War/Peace Mode

A geopolitical risk intelligence dashboard. Access via the WAR/PEACE button in the mode bar.

Dashboard Panels

• War-Peace Index -- Composite risk gauge computed from conflict-related market prices. Tracks overall geopolitical tension level
• Commodities -- Commodity market signals affected by geopolitical risk (oil, gold, defense stocks)
• Regional Breakdown -- Conflict zones: Ukraine/Russia, Israel/Gaza/Iran, China/Taiwan, North Korea, and other hotspots. Each zone scored by market price aggregation
• AI Analysis -- AI-powered geopolitical risk assessment with model attribution
• Conflict Markets -- All prediction markets tagged as geopolitical, with prices and volume
• Selected Market -- Deep dive on selected conflict market
• Market Signals -- Enriched signal categories for the selected market
• Geopolitics News -- Brave Search-powered news filtered to geopolitical events

Updated every 12 hours via cron. CRT scanline overlay + animated background for atmosphere.

AI Picks -- Main Site

The main page at yesno.events shows AI-generated prediction cards with full quant data. Same card engine as the ALPHA panel Packs, presented as a scrollable page with pack opening, collection view, and resolved cards.

Card Intelligence

• Quant grid: Price, fair value (Bayesian), edge %, half-Kelly %, MC path stability %
• Signals: MOMENTUM, SMART $, BUY PRESSURE, HOLDERS, VOLUME, WHALE, EDGE, EXPIRING
• Side assignment: yesPrice <= 0.40 = YES, >= 0.60 = NO, middle = weighted random
• Rarity: Deterministic from quality score. Common (<8, 1x), Uncommon (8-13, 1.5x), Rare (14-19, 2x), Legendary (20+, 3x)
• Card value (active): (10 / entryPrice) * currentPrice
• Card value (resolved correct): (10 / entryPrice) * rarityMultiplier

Player ID stored in localStorage. State persisted in Upstash Redis. Markets resolved by cron every 30 minutes.

MCP Server -- Machine-Readable Intelligence

A 16-tool Model Context Protocol server at yesno.events/api/mcp. Connect Claude Desktop, Cursor, or any MCP client to live prediction market intelligence.

Tools (15)

Resources (2)

Prompts (2)

Credits required for 6 tools. Connect via Claude Desktop, Cursor, or any MCP client at yesno.events/api/mcp

Keyboard Shortcuts

Command Bar

Press / to open. Commands:

Sector Classification

Every market is auto-classified into one of 8 sectors via keyword matching:

Data and Caching

• Markets: Loaded on startup, auto-refreshed every 30 seconds
• Enriched signals: Fetched per market on selection, cached in IndexedDB (3-minute TTL)
• Whale trades: Refreshed every 60 seconds
• Price history: Cached in IndexedDB (5-minute TTL)
• AI Picks: On-demand generation, 15-minute refresh cycle
• Elon tweets: Cron every 5 minutes via XTracker API
• Foresight scenarios: Cron every 6 hours (market selection, research, AI scenario generation)
• AI Picks resolve: Cron every 30 minutes (resolve ended markets, update card values)
• War/Peace index: Cron every 12 hours
• Calibration: localStorage, survives reloads
• IndexedDB cleanup: 24-hour TTL with automatic expiry on startup

Data Sources

The platform aggregates from Polymarket APIs (Gamma, CLOB, Data) and a custom enrichment backend across all 5 terminals:

AI models via OpenRouter: Gemini Flash 2.0 (Foresight scenarios, agent analysis), Claude Haiku 3.5 (market analysis), Claude Sonnet 4 (deep analysis, conviction picks). State stored in Upstash Redis.

Developer API

Programmatic access to YESNO intelligence via REST API and MCP server. Team subscribers get API keys for 10,000 calls/day across 8 endpoints -- markets, whales, leaderboard, news, holders, trending, sectors. MCP server provides tools for AI assistants.

Developer Portal

Nexus Mode

Consolidated intelligence dashboard. Cross-panel data synthesis with system pulse scoring, AI digest, and market momentum overview. Access via the NEXUS button in the mode bar.

Open Nexus

Pricing

Free: 3 AI analyses/day, 2 news searches/day, 5 enriched signals/day with full terminal access. Credits (pay-as-you-go, $10 min): unlimited access, REST API, MCP server, up to 5 API keys.

View Plans