Prediction Market Research

Your AI trades
Kalshi while
you feast

A Telegram-native workspace for Kalshi prediction markets. Browse two open research feeds — LLM-driven market audits and what top podcasters are actually saying — and place trades yourself, on your terms.

Kalmari home view in Telegram — paper-mode dashboard with portfolio, trade history, schedules, and Ask AI controls.

Two open feeds.
Better information in.

Prediction markets only work if the people trading on them have done the homework. Kalmari pipes two public research feeds into your Telegram, so you can read what frontier LLMs and podcasters are saying before you click trade.

Kalshi Audits
LLM Research

Deep research generated by Claude (Anthropic) — frontier LLMs read primary sources and surface what the model thinks about open Kalshi markets. Trade with the best models, not against them.

Live feed
Loading feed…
View full source
Mentioned in Podcasts
Podcast Intel

Concrete claims made by top podcast hosts — politics, the economy, business, culture — mapped to live Kalshi markets with the side to take to copy the speaker. Trade on real conversations, not social noise.

Live picks
Loading picks…
View full source

Signals in.
You decide.

Kalmari is signals-based: it surfaces ideas and lets you place each trade yourself. No autopilot, no surprises.

01 / CONNECT
Open in Telegram
Tap @kalshi_frontrun_bot in Telegram. Browse markets and feeds — no account needed. Connect your Kalshi API key only when you’re ready to trade real money.
02 / READ
Browse the feeds
Pull the Kalshi Audits feed for LLM-driven market research. Pull the Podcasts feed to see what hosts are actually saying. Both update continuously.
03 / SIMULATE
Paper-trade first
Every account starts in paper mode — place trades, track P&L, and stress-test your reads without risking a dollar. Live mode requires an explicit opt-in later.
04 / TRADE
You click the button
Kalmari does not auto-trade. Every order — paper or live — is placed by you, in chat. Your keys, your trades, your call.

What you get

📢
Two open signal feeds
Kalshi Audits (Claude-driven research) and Mentioned-in-Podcasts (host transcripts mapped to markets). Both rendered inline in Telegram.
🧬
Paper mode included
Track positions, P&L, cash, and trade history in a simulated portfolio. Switch to live only after an explicit opt-in.
🔐
BYOK: your keys, your control
Connect your own Kalshi API key. Private keys are AES-256-GCM encrypted at rest. Revoke access from Kalshi any time.
💬
Telegram-native
No app to install. No dashboard to learn. Chat with the bot in plain English — or use slash commands when you want speed.
🧠
Claude-powered analysis
Ask the bot to explain a market, sanity-check a position, or summarise a feed. Every answer cites its sources.
👥
Isolated by user
Each user’s positions, history, and credentials are kept separate. Your research and your trades stay yours.

Your Telegram, your terminal

Type /home in chat and you get a portfolio view: cash, positions, realised and unrealised P&L — with one-tap access to feeds, trade history, and settings. Nothing else to install.

  • Plain-English questions or fast slash commands
  • Paper-mode portfolio with live Kalshi prices
  • Trade history, schedules, and settings inline
  • Ask AI any market question in the chat
Kalmari home view in Telegram showing paper-mode total value, P&L breakdown, and navigation tiles for Portfolio, Trade History, Schedules, Settings, Ask AI, and Refresh.

Make prediction markets a real information channel.

We think prediction markets can become a high-quality way to surface what’s happening in business, politics, and the economy — if the people trading them do real research. Kalmari is a small experiment in that direction: open signal feeds, transparent reasoning, and a community testing new research methods to find strategies that actually hold up.

Start exploring signals today

Free to browse. Paper mode by default. Bring your own Kalshi key when you’re ready to trade.

OPEN IN TELEGRAM
@kalshi_frontrun_bot · Browse free · Paper mode included