Best Kalshi Trading Bots Ranked (2026)

Kalshi is the only CFTC-regulated prediction market exchange in the United States, and its REST API supports full automation from market scanning to order execution. This guide ranks every verified Kalshi trading bot available in 2026 — open-source repos with real GitHub activity, confirmed SaaS tools, and honest assessments of what each one actually does.

Breaking: two Kalshi API changes in 2026. First, the fixed-point migration (completed March 2026): Kalshi deprecated legacy integer-cent price fields and moved to fixed-point dollar strings (e.g., "0.6500" for $0.65), with quantities using the _fp suffix. See the Kalshi Fixed-Point Migration documentation. Second, and bigger for bots: the legacy order-mutation endpoints (POST /portfolio/orders with side: yes/no, action: buy/sell) were removed in June 2026, replaced by CreateOrder V2 at POST /portfolio/events/orders (changelog). Any bot that hasn’t shipped a commit adapting to the V2 order model since June 2026 cannot place orders — check each repo’s recent history before deploying.

Operational change (August 2026): exchange sharding. Kalshi began splitting the exchange into dedicated instances — Crypto, Tennis, and Baseball moved to their own shards, with auto-routing enabled by default on August 27 and a new exchange_index field on markets, orders, fills, and balances (sharding docs, changelog). This isn’t breaking the way the June order-endpoint removal was, but bots that preallocate collateral or track balances need to handle per-shard balances — see problem 11 in our Kalshi API problems guide. New cancel-all-orders endpoints (August 27) also simplify kill switches. What your bot must handle:

• Per-shard balances — collateral and balance queries return per-shard figures; aggregate before sizing.
• exchange_index on fills and orders — persist it, or reconciliation breaks once a market moves shards.
• Auto-routing — on by default since August 27; opt out explicitly if you route orders yourself.
• Cancel-all per shard — kill switches must fire the cancel-all endpoint on every shard, not just the primary.

New category (April 21, 2026): Kalshi, Benzinga, and Fiscal.ai announced a collaboration to expand prediction markets with company KPI contracts — markets tied to corporate performance indicators (Tesla production, DoorDash deliveries, Netflix subscribers, earnings outcomes). KPI-specialized bots are a new archetype that pairs Fiscal.ai data feeds and the Benzinga earnings calendar with Kalshi’s settlement engine. As of August 31, 2026, no purpose-built open-source KPI bot has surfaced on GitHub — the archetype is still open.

What Makes Kalshi Different for Bots

Before evaluating specific bots, these platform differences from Polymarket directly affect bot design:

These differences mean most bots need platform-specific adapters. Cross-platform tools like pmxt handle this internally.

Kalshi API vs OG.com API

OG.com exposes a public API for event-contract trading aimed at the offshore audience, but the two are not interchangeable for bot builders. Kalshi is CFTC-regulated with RSA-signed REST/WebSocket/FIX access, a full demo environment, and the V2 order model documented above. OG.com’s API carries no US regulatory standing and no equivalent sandbox. For US-legal automated trading, Kalshi is the only option; treat OG.com as an offshore odds source, not an execution venue for regulated strategies.

Quick Rankings

AI Trading Bots & Toolkits

The highest-activity Kalshi bots on GitHub use frontier LLMs to estimate event probabilities, then trade when the model’s estimate diverges from the market price. Approaches range from single-model toolkits to agentic multi-venue frameworks.

ryanfrigo/kalshi-ai-trading-bot

GitHub: ryanfrigo/kalshi-ai-trading-bot | 426 stars, 161 forks (August 31, 2026) | MIT License | Python 3.12+

The most complete open-source Kalshi toolkit — signed REST + WebSocket client, market ingestion via the Events API, position tracking with stop-loss/take-profit exits, SQLite telemetry, and a Streamlit dashboard. The README (revised again in late August 2026) now ships three example strategies and is explicit that none of them is “the right answer”: AI Directional (a single LLM per decision via OpenRouter with an error-fallback chain — the README states outright that “it is not a ‘5-model ensemble’ despite earlier README claims”), Safe Compounder (pure edge-based NO-side math, no LLM required, resting maker orders), and Beast Mode (aggressive, no category guardrails, explicitly not recommended for live trading).

Architecture:

INGEST          →  DECIDE               →  EXECUTE  →  TRACK
Kalshi REST API    LLM probability est.    Kalshi     SQLite DB
WebSocket Stream   (primary_model)         Orders     Streamlit Dashboard

Key features:

• Pluggable LLM decision layer via OpenRouter (primary_model, anthropic/claude-sonnet-4.5 by default — swap models with one config line)
• Quarter-Kelly sizing, 15% max-drawdown circuit breaker, 3% max position, 30% sector-concentration cap (defaults tuned from documented live losses)
• SQLite telemetry for every trade, AI decision, and LLM cost; daily AI-spend limit enforced before each API call
• Paper trading mode (paper_trader.py) plus a close-all command that liquidates every open position via limit sells at best bid
• Streamlit dashboard for real-time portfolio, positions, and P&L
• Non-LLM Safe Compounder strategy for builders who want edge math without inference costs

Requirements: Python 3.12+, Kalshi API key with RSA private key, OpenRouter key (only for the AI Directional strategy).

Honest assessment: Still the strongest Kalshi-specific starting point, and its README is unusually honest — it walks back its own earlier five-model-ensemble marketing, states the examples “lose money on certain markets,” and documents lessons from real losses (quarter-Kelly over three-quarter Kelly, category discipline beating AI confidence). The troubleshooting section addresses current API realities including the KXMVE parlay-ticker behavior of GET /markets — evidence of maintenance well past the June 2026 legacy-endpoint removal. Note the star count reads lower than earlier in August (426 vs the ~570 we counted on August 12); we report the live number.

Best for: Developers comfortable with Python who want a well-documented starting point for building AI-driven Kalshi strategies.

OctagonAI/kalshi-trading-bot-cli

GitHub: OctagonAI/kalshi-trading-bot-cli | 378 stars (August 31, 2026) | last push August 19, 2026 — actively maintained (formerly kalshi-deep-trading-bot)

A CLI-first trading bot that uses Octagon Deep Research for fundamental market analysis and OpenAI for structured betting decisions. Unlike the multi-model ensemble approach above, this bot focuses on depth of research per market rather than model consensus.

Key features:

• Deep fundamental research via Octagon AI (generates independent probability estimates)
• OpenAI structured outputs for trade decisions
• 5-gate risk engine before any trade executes
• Kelly Criterion sizing with edge computation against live order books
• JSON-structured CLI output for scripting and automation
• Supports market search, analysis, and automated execution

Example workflow:

# 1. Search for markets
MARKETS=$(bun start search crypto --json | jq '.data')

# 2. Analyze a specific market
bun start analyze KXBTC-26APR-B95000 --json

# 3. Returns modelProb, marketProb, edge, confidence, and drivers

Honest assessment: The active commit history and ongoing development suggest real iteration. The CLI-first design makes it easy to integrate into larger systems. Octagon Deep Research is a paid API, which adds cost beyond just LLM inference. The 5-gate risk engine is a mature design pattern.

Best for: Traders who want deep, research-backed analysis on individual markets rather than broad scanning.

alsk1992/CloddsBot

GitHub: alsk1992/CloddsBot | 841 stars (August 31, 2026) | created January 2026 | last push August 29, 2026

A fast-growing new entrant: an open-source Claude-based trading agent that operates across Kalshi, Polymarket, Binance, and Hyperliquid. Rather than a fixed strategy loop, CloddsBot is agentic — Claude drives market research, probability estimation, and order decisions across venues from a single framework.

Honest assessment: At 841 stars roughly seven months after creation — up 170+ in the last three weeks alone, with commits as recent as August 29 — this is one of the fastest-growing and most actively maintained prediction market repos of 2026. Multi-venue support is its differentiator, but that breadth also means the Kalshi integration is one adapter among four — verify its V2 order support against your use case before going live. As with all agentic bots, LLM-driven decisions need hard risk limits around them.

Best for: Developers who want a Claude-based agent framework spanning prediction markets and crypto exchanges.

suislanchez/polymarket-kalshi-weather-bot

GitHub: suislanchez/polymarket-kalshi-weather-bot | 708 stars (August 31, 2026) | created February 2026 | last commit March 2, 2026 — dormant

A specialist quant bot targeting Kalshi’s KXHIGH daily high-temperature weather markets (and the Polymarket equivalents). Instead of an LLM, it uses GFS ensemble weather-model forecasts to estimate outcome probabilities, then sizes positions with the Kelly criterion.

Honest assessment: This is the clearest example of the “specialist edge” archetype: a physical forecasting model (GFS ensembles) with a defensible information edge over casual traders, applied to a narrow market family. But apply this guide’s own litmus test: the repo has had no commits since March 2, 2026, which predates the June 2026 removal of Kalshi’s legacy order endpoints — its Kalshi order-placement code will not work against the current API without modification. Stars keep climbing (586 → 708 in August alone), which says more about the appeal of the idea than the state of the code. Treat it as a reference design for forecast-driven trading, not a deployable bot.

Best for: Quant-minded builders who want to study a non-LLM, model-driven edge — budget for a V2 order-endpoint update before running it live on Kalshi.

Sentiment & Signal Bots

Kalshi’s event-driven markets — Fed rate decisions, hurricane landfalls, company IPOs — are ideal for sentiment-driven bots. News breaks, odds shift, and the fastest bot wins.

Kalshi News Bot (Free, Open-Source)

The Kalshi News Bot is an open-source Python bot that uses Claude AI to analyze breaking news and identify mispriced events on Kalshi. At approximately 300 lines of code, it is the simplest functional Kalshi trading bot available.

Strengths:

• Free and fully open-source
• Uses Claude for news analysis — high-quality reasoning about event probabilities
• Minimal setup — one of the fastest paths to a working Kalshi bot
• Clean, readable codebase ideal for learning and customization

Limitations:

• No built-in backtesting
• Single-strategy (news sentiment only)
• Requires your own Anthropic API key (Claude inference costs apply)
• No web dashboard — terminal-only

Best for: Developers who want a working, understandable Kalshi bot to study, customize, and extend. The small codebase makes it the best starting point for learning how prediction market bots work.

yllvar/Kalshi-Quant-TeleBot (no longer active)

GitHub: yllvar/Kalshi-Quant-TeleBot | 66 stars | last push January 9, 2026 — inactive

A more complex bot with a Telegram interface for remote monitoring and control. The architecture separates the trading engine (Python) from the user interface (JavaScript Telegram bot), connected via WebSocket.

Key features:

• Python trading engine with pandas, numpy, and scikit-learn integration
• Telegram bot UI with interactive commands, real-time notifications, and alerts
• News sentiment analysis with event correlation to specific Kalshi markets
• Multi-strategy support: sentiment-based, statistical arbitrage, and volatility patterns
• Position sizing based on confidence levels, market volatility, and available capital
• Multi-layered risk controls including stop-losses, position limits, and exposure management

Honest assessment: The repo has had no commits since January 9, 2026 — seven months of inactivity that predates the breaking June 2026 removal of the legacy order endpoints. Treat it as unmaintained: the order-placement code will not work against the current API without modification. The dual-layer architecture (Python engine + JavaScript Telegram bot) remains a useful reference design for phone-based bot control, but don’t deploy it as-is.

Best for: Studying a Telegram-controlled bot architecture — not for live trading without a significant update pass.

ajwann/kalshi-genai-trading-bot

GitHub: ajwann/kalshi-genai-trading-bot | last push April 6, 2026 — predates the June 2026 V2 order migration

A minimal, “vibe-coded” trading bot that connects Grok (xAI) to Kalshi. The codebase was largely generated by LLMs, with the iteration prompts included in the /prompts directory.

Structure:

• main.py — Entry point calling Kalshi and Grok clients
• kalshi_client.py — API wrapper for Kalshi
• grok_client.py — API wrapper for xAI’s Grok
• Role prompting tells the LLM it is a professional prediction market trader

Honest assessment: This is a learning tool, not production infrastructure. The included prompt iteration history is genuinely useful for understanding how to instruct LLMs for trading decisions. The simple architecture makes the entire system easy to understand in an afternoon.

Best for: Beginners who want to understand the minimal viable architecture of an LLM-powered trading bot.

AI Research & Intelligence Tools

These tools provide the intelligence layer without handling execution directly. They complement any of the bots above.

Alphascope

Website: alphascope.app | SaaS | Free tier available

Alphascope is an AI-powered research platform built for prediction market traders, now covering Polymarket, Kalshi, Manifold, and Opinion. It provides probability estimates, news impact analysis, and cross-platform arbitrage detection — the intelligence layer that most trading bots lack.

Key features:

• AI probability estimates for individual markets across Polymarket, Kalshi, Manifold, and Opinion
• News impact scoring — ranks breaking stories by which markets they affect
• Cross-platform price comparison (Kalshi vs Polymarket) for arbitrage detection
• Interactive price charts with news event markers
• Market analysis with confidence scores and reasoning transparency

How traders use it: Submit a Kalshi market URL, receive an AI probability estimate with rationale, and compare against the live market price. When Alphascope’s estimate diverges significantly from the market price, that is a potential trading signal.

Honest assessment: Alphascope is a research tool, not an execution bot. It does not place trades for you. The value is in its cross-platform market views and AI probability estimates — useful as an input signal for any of the open-source bots above. The founder has publicly documented trading $10 into Kalshi using only Alphascope signals, providing transparency about real-world performance.

Best for: Traders who want AI-powered market research without building their own analysis pipeline. Pairs well with any execution bot.

pmxt — Unified Cross-Platform SDK

AgentBets guide: pmxt Python Library Tutorial

pmxt is the CCXT for prediction markets — one Python SDK (2,110 GitHub stars as of August 31, 2026) to read data and trade across Polymarket, Kalshi, Limitless, Probable, Baozi, Myriad, and Opinion. If you are building a custom bot that needs to scan or trade across multiple prediction market exchanges, pmxt eliminates the need to write platform-specific adapters. Multi-venue bots like this are also the natural bridge between prediction market exchanges and sportsbook odds feeds (via The Odds API) — the architecture the agent betting stack formalizes.

Why it matters for Kalshi bots: pmxt handles the authentication differences (RSA for Kalshi vs HMAC for Polymarket), price format normalization (fixed-point dollar strings vs USDC decimals), and order book normalization automatically. A bot built on pmxt can scan for arbitrage across platforms with a single API call pattern.

Important Kalshi caveat: pmxt’s hosted mode supports trading writes on Polymarket, Opinion, and Limitless only. Trading on Kalshi requires self-hosting pmxt with your own Kalshi RSA credentials — hosted mode can read Kalshi data but cannot place Kalshi orders on your behalf.

Best for: Developers building custom cross-platform bots, especially for arbitrage between Kalshi and Polymarket — self-hosted if Kalshi execution is required.

Real-World Kalshi Bot Lessons

Building a Kalshi bot is straightforward. Making it profitable is hard. A few documented lessons from the community:

Latency matters in live sports markets. A developer built a real-time Kalshi NFL trading bot that monitored ESPN win probability, planning to buy before the market adjusted. The result: professional sports bettors and market makers had access to data feeds at least 30 seconds faster than the unofficial ESPN API. The bot could not compete on speed in live sports markets. The full writeup is on Hacking the Markets.

Guardrails come from real losses. The ryanfrigo/kalshi-ai-trading-bot README explicitly states that category enforcement and risk guardrails were added after live trading without them led to significant losses. This is a common pattern — the first version of every trading bot loses money.

LLM probability estimates are noisy. Multiple bot authors note that LLM probability estimates diverge widely between models and across runs. Multi-model consensus (only trade when models agree) is one response to this noise, not a guarantee of accuracy — and notably, the most prominent Kalshi AI repo’s README now states outright that “it is not a ‘5-model ensemble’ despite earlier README claims,” calling one LLM per decision with a fallback chain that only triggers on errors.

The June 2026 order API change is a maintenance litmus test. Kalshi removed the legacy POST /portfolio/orders mutation endpoints in June 2026 in favor of CreateOrder V2 (POST /portfolio/events/orders). Any bot built on the legacy order model broke at that point. Before deploying any repo from this list — or anywhere else — check whether it has commits after June 2026 (or an explicit V2 migration); a repo with no activity since then almost certainly cannot place orders against the current API.

Building Your Own Kalshi Bot

Kalshi’s developer tools for building custom bots:

• Kalshi API Guide — Complete REST API documentation with RSA-PSS authentication, endpoints, and examples
• How to Set Up a Trading Bot on Kalshi — Step-by-step from account creation to first automated trade
• Prediction Market API Reference — Side-by-side Kalshi vs Polymarket comparison
• Kalshi API Tool Entry — Quick reference and links
• The Odds API — Kalshi markets are also exposed via The Odds API under the us_ex region (key kalshi), useful for sportsbook-vs-exchange price comparison without a native Kalshi integration. For order execution you still need the native API.
• Demo environment: demo-api.kalshi.co/trade-api/v2 for risk-free testing

Kalshi Authentication Quick Start

# pip install kalshi_python_sync  (the old kalshi-python package is deprecated)
from kalshi_python_sync import Configuration, KalshiClient

config = Configuration(
    host="https://api.elections.kalshi.com/trade-api/v2"
)

with open('path/to/private_key.pem', 'r') as f:
    config.private_key_pem = f.read()

config.api_key_id = "your-api-key-id"
client = KalshiClient(config)

# Check balance
balance = client.get_balance()
print(f"Available: ${balance.balance_dollars}")

KPI-Market Support: Open Archetype

Four months after Kalshi’s company KPI contracts launched (April 21, 2026, via the Benzinga + Fiscal.ai partnership), none of the bots ranked above ship pre-built integrations for them, and as of August 31, 2026 our GitHub searches turn up no purpose-built open-source KPI bot at all. These markets settle against structured corporate data rather than sports outcomes or binary events, so a KPI-capable bot needs a distinct data pipeline — Benzinga’s earnings calendar for event detection, Fiscal.ai for real-time KPI feeds, and settlement-aware models that match market definitions to reported metrics. The archetype remains genuinely open: a fork of a toolkit like ryanfrigo/kalshi-ai-trading-bot with a KPI data pipeline would have the category to itself.

Marketplace Reviews

For deeper dives into specific bot strategies on Kalshi, see the individual marketplace reviews:

• Best Sentiment Bot for Kalshi — News and social sentiment analysis bots
• Best Arbitrage Bot for Kalshi — Cross-platform spread trading tools
• Best Market-Making Bot for Kalshi — Liquidity provision and spread capture
• Best Copy-Trading Bot for Kalshi — Follow profitable Kalshi traders
• Best Momentum Bot for Kalshi — Trending event market strategies

See Also

• Best Prediction Market Bots 2026 — Overall rankings across all platforms
• Best Polymarket Bots 2026 — Polymarket-specific rankings
• Best Open-Source Prediction Market Bots — Free tools ranked
• Polymarket vs Kalshi for Bot Trading — Head-to-head platform comparison
• Best Prediction Market Arbitrage Bots — Cross-platform arbitrage rankings
• Kalshi API Guide — Complete Kalshi API documentation
• pmxt Python Library Tutorial — Unified SDK for cross-platform trading
• Agent Betting Glossary — 130+ prediction market terms defined
• Agent Marketplace — Browse and compare all agents

Rankings updated August 2026 (featured repo star counts and last-push dates re-verified against live GitHub). Every bot listed has been verified against a live GitHub repository or production website. Not financial advice. Built for builders.