AI-assisted trading for Kalshi prediction markets
Available AI models estimate probabilities; coverage varies by task and service availability. Fractional Kelly sizing, settlement monitoring, and account-level limits shape each order. Tradewink monitors supported Kalshi markets with paper trading for public subscriptions. Live orders require separately approved private beta access.
Requires a Kalshi account, even in paper mode. Kalshi is not available in every country.
CFTC-regulated exchange · Public subscriptions are paper-only · Informational research
Want stock trading signals instead?
Why Tradewink Predictions
Multiple checks, not one model
Review AI estimates alongside fractional Kelly sizing, trade results, and liquidity context. Event accuracy is unavailable.
AI probability estimates
Available models estimate probabilities for the task. When multiple estimates are available, review their combined forecast alongside the evidence and market price. Accuracy is not guaranteed.
Fractional Kelly position sizing
Kelly criterion with fractional scaling, bankroll limits, and confidence gating. Sizes each simulated bet from a formula rather than gut feel. Formula-based sizing does not guarantee profitable outcomes.
Settlement and outcome tracking
Open positions are checked for resolution on a schedule, and trade results are recorded. Event accuracy is unavailable. Tradewink does not currently trade the post-event repricing window.
Order-book and liquidity context
The estimator reads Kalshi order-book depth and spread, and the betting strategies require minimum liquidity before a market qualifies. Order-book data does not predict direction on its own.
Trade results, with accuracy limits
Review trade P&L and win/loss history. Event accuracy is unavailable because trade results alone do not verify event outcomes. Existing automatic adjustments and risk rules remain in use.
User-set risk limits
Category concentration limits, max bet limits, daily loss limits, and bankroll protection. Designed so no single market wipes the account — though cumulative losses across correlated markets remain possible.
Market Categories
Markets the AI can trade
From Fed rate decisions to weather records. The AI researches each category with available data sources and uncertain probability estimates.
Economics
Fed rate decisions, CPI, jobs reports, GDP growth. AI pulls in macro data and news and estimates probabilities across possible outcomes.
Politics
Elections, legislation, Supreme Court rulings. AI research cross-references polls, news, and historical patterns.
Sports
Game outcomes, player milestones, championship futures. AI researches recent news and public information to estimate probabilities.
Climate
Temperature records, hurricane landfalls, precipitation thresholds. Open-Meteo weather forecasts feed directly into probability estimates.
Crypto
Bitcoin price thresholds, ETF approvals, regulatory actions. AI tracks news and sentiment across multiple sources.
Financials
S&P 500 levels, earnings surprises, IPO outcomes. The same data pipeline that powers stock trading, adapted for binary markets.
Culture
Awards shows, viral moments, media milestones. Evaluated only when you select this category.
Tech & Science
Product launches, scientific milestones, regulatory approvals. Evaluated only when you select this category.
6 Strategies
How the AI analyzes prediction-market edge
Not one trick — probability estimation, position sizing, and several betting approaches work together. Each targets a different kind of mispricing, and none guarantees an edge.
Ensemble Forecasting
Available models estimate probabilities in separate runs. When multiple estimates are available, the system combines them using the geometric mean of odds.
Kelly Criterion Sizing
Bet sizing using fractional Kelly, with YES and NO sides, bankroll limits, and model confidence taken into account.
Value Bets
When the ensemble estimate differs from the market price by more than your minimum edge, the market is scored as a value opportunity and sized.
Near-Certainty Bets
Contracts priced near certainty that resolve soon, taken only with high model confidence and enough liquidity. Small edges, with real tail risk if the unlikely outcome hits.
Contrarian
Flags markets priced at an extreme that the estimate does not support, then evaluates a position against the market price.
Longshot Fade
Sells low-priced tail outcomes when the estimate sits meaningfully below the market price, aiming at the long-shot bias.
The Pipeline
Data in. Probabilities out. Paper positions tracked.
Available AI estimates and data feeds inform the estimation engine. When an estimate differs enough from the market price, simulated positions are sized within your limits. Public subscriptions are paper-only.
Data Sources
Kalshi
Live market prices
Research models
Probability estimators
Web search
Polls and discussion
Open-Meteo
Weather forecasts
News Feeds
Recent headlines
Prediction Engine
Available AI estimates
Execution
Kelly Sizing
Fractional bet size
Paper Fills
Simulated with slippage
Settlement Tracking
Outcomes recorded
Trade Results
P&L and win/loss
How It Works
Set your limits. Practice in paper mode.
Configure your bankroll, risk tolerance, and market preferences. Background loops handle scanning, estimating, and sizing around the clock. Public subscriptions use simulated positions; live orders require separately approved private beta access.
AI scans supported markets
Kalshi markets in the categories you choose are scanned continuously. Candidates get available AI research — news, data, forecasts, and probability estimates.
Estimate gap found, paper bet sized
When the AI estimate diverges from market price by more than your minimum, fractional Kelly sizes a paper bet. A gap is not proof of mispricing. Category limits keep exposure diversified.
Review simulated positions
Public subscriptions create paper positions within the limits you set. Review the estimate, exposure, and recorded outcome in your dashboard.
Getting Started
Pricing
Start free. Scale as you grow.
The Free plan needs no credit card. Prediction-market research is shared across plans. Paid plans add hosted models for supported research requests, signal delivery options and API access by tier. Paid plans include a 7-day trial (payment method required). Public subscriptions are paper-only on every plan.
Common questions
Currently Kalshi, a CFTC-regulated exchange for event contracts; no other venue is connected. The system evaluates binary event contracts in categories such as economics, politics, sports, climate, crypto and financials; culture and tech & science markets only when you select them.
The system uses market research, available order-book context and AI probability estimates. When multiple estimates are available, their outputs are combined using the geometric mean of odds. Background prediction research is shared across plans; paid model tiers apply to supported user-initiated research requests. The number of estimates depends on the task and service availability. Estimates are uncertain and do not guarantee an outcome.
A bet sizing formula designed to maximize long-term bankroll growth when the probability estimate is accurate. Because estimates are uncertain, the system uses fractional Kelly — half-Kelly (0.5x) by default, with a 0.25x conservative preset — and adjusts for both YES and NO side bets.
No. Tradewink checks settlement and records trade results, but it does not currently trade the post-event repricing window. Event accuracy is unavailable. Only Kalshi is connected, so there is no cross-exchange arbitrage either.
Yes, as context. The estimator reads Kalshi order-book depth and spread for each candidate market, and the betting strategies require minimum liquidity before a market qualifies. Order-book data does not predict direction on its own.
You can enable automated paper research and simulated positions within the limits you set. Public subscriptions are paper-only. Live Kalshi orders require separately approved private beta access; connecting an account does not grant that access.
Real-money prediction market trading involves substantial risk. Public subscriptions are paper-only, and paper results do not show how real orders would have filled or performed. Past performance does not guarantee future results. The system has risk controls (fractional Kelly sizing, bankroll limits, category limits), but losses are possible — including significant losses. Only trade with money you can afford to lose. Prediction market trading is not suitable for everyone.
You need a Kalshi account to use Predictions, even for paper bets. Kalshi sets its own eligibility rules, including location, age and identity-verification requirements. Kalshi charges trading fees on each contract — see kalshi.com for current fee schedules. Tradewink does not control Kalshi's fees, eligibility rules, account minimums, or market availability.
Yes. Public Predictions subscriptions run the AI pipeline in paper mode without placing real money bets. Adjust your simulated bankroll and risk limits in Predictions settings.
AI models have real limitations. Black swan events, unprecedented geopolitical shocks, and sudden market microstructure changes can cause significant losses. Multiple models can share the same errors. Event accuracy is currently unavailable, and trade win rates do not establish forecast accuracy. The system cannot guarantee any outcome.
Tradewink Autopilot
Interested in stocks and options?
Predictions works with event contracts on Kalshi. The core Tradewink workspace covers stocks and options research and paper trading, with account connections through direct broker integrations and SnapTrade.
Tradewink Signals
Prefer trade ideas you review yourself?
Signals delivers AI-grade stock research to Discord or email — currently enabled signal categories with full explanations. You make the call.
Test the AI's odds in paper mode
AI probability estimates, fractional Kelly sizing, and settlement tracking on Kalshi, a CFTC-regulated exchange. Public subscriptions are paper-only; live orders require separately approved private beta access.
Important disclosures
Informational purposes only
Tradewink is published by Tradewink LLC, which is not a registered investment adviser, broker-dealer, commodity trading advisor, or financial planner. All data, signals, and analytics on this page are general, impersonal, and for informational purposes only. They do not constitute investment advice, financial advice, or a recommendation to buy or sell any security or other instrument.
Trading risk
Past performance does not guarantee future results. Trading involves substantial risk of loss, including the possibility of losing more than your initial investment. You are solely responsible for your own trading decisions.
Hypothetical & backtested results
These results are based on simulated or hypothetical performance results that have certain inherent limitations. Unlike the results shown in an actual performance record, these results do not represent actual trading. Also, because these trades have not actually been executed, these results may have under- or over-compensated for the impact, if any, of certain market factors, such as lack of liquidity. Simulated or hypothetical trading programs in general are also subject to the fact that they are designed with the benefit of hindsight. No representation is being made that any account will or is likely to achieve profits or losses similar to these being shown.