Kalshi — A Complete Reference for Building Trading Models

0The one-paragraph mental model

1What Kalshi is and how it's regulated

1.1 What is Kalshi?

The first regulated exchange for event contracts — instruments whose value comes from whether a real-world event happens, not from owning a company (stocks) or a commodity. Every contract is one binary decision: Yes (it happens) or No (it doesn't).

1.2 What is a prediction market?

A marketplace where contract prices are the crowd's probability estimate, and each contract pays $1 if you're right. A stock exchange trades shares of ownership; a prediction market trades contracts on whether specific events happen — e.g., "Will interest rates rise next quarter?" (Price mechanics in §3.)

1.3 Regulatory status

Kalshi is a Designated Contract Market (DCM) regulated by the CFTC (the U.S. derivatives regulator since 1974), authorized to list futures, swaps, and options. Rulebook: kalshi.com/regulatory/rulebook. Full regulatory detail in §13.

1.4 Kalshi Klear (the clearinghouse)

Kalshi Klear LLC is Kalshi's CFTC-approved clearinghouse (it replaced MIAXdx). In-house clearing enables more market categories, faster listings on breaking news, faster settlement, and APY on idle cash (advertised 3.25%, variable).

1.5 How Kalshi makes money

A neutral exchange — Kalshi holds no position on outcomes. Revenue is a transaction fee on each contract's expected earnings (§6). No settlement fee, no membership fee.

Sources: What is Kalshi? · What are prediction markets? · How is Kalshi regulated? · What is Kalshi Klear? · How does Kalshi make money?

2The core invariants (rules that never change)

These are the load-bearing facts. A correct model must encode all of them.

Sources: How are prices determined? · Buying Yes vs Selling No · Who are you trading with? · Combos

3How prices work (price = probability)

Kalshi uses a dynamic pricing model: the contract price tracks the market's perceived probability, 0%→100%. As perceived likelihood of Yes rises, the Yes price rises and No falls — always summing to $1.

Source: How are prices determined?

4Market structure & terminology (the data model)

Kalshi organizes markets in a strict hierarchy. Knowing it is essential for spreadsheet layout (a tab per series, rows per event/market, columns per field).

• Category — high-level grouping for related series; a series belongs to one.
• Subcategory — narrower grouping; a series can belong to several; tags in API filters.
• Series — a recurring template defining structure, settlement sources, metadata. Events in a series look at similar data over disjoint time periods with no outcome dependency between them, and share a ticker prefix.
• Event — "a collection of markets and the basic unit that members should interact with."
• Market — "a single binary market," often a single strike. Type is binary or scalar.

4.1 Strikes

Within an event, markets are often strikes — thresholds of an underlying quantity (temp ≥ 60/62/64°F; "S&P closes above X"). The rules summary is filterable by strike; you can request additional strikes.

4.2 Mutually exclusive vs. directional groups

• Mutually exclusive group — exactly one market can resolve Yes (e.g., "Who will be confirmed?"). Terminal states: exactly one Yes, or all No.
• Directional (nested threshold) group — logically ordered: a harder threshold being Yes implies easier ones are Yes (e.g., ">3M check-ins" Yes ⇒ ">2M" Yes).

Both shapes enable collateral return (§8).

4.3 Key per-market fields (spreadsheet columns)

From the API Market schema: ticker, event_ticker, market_type, yes_sub_title/no_sub_title, open_time, close_time, status, yes_bid_dollars, yes_ask_dollars, no_bid_dollars, no_ask_dollars, last_price_dollars, volume_fp, volume_24h_fp, liquidity_dollars, open_interest_fp, result, rules_primary, rules_secondary.

Sources: API Glossary · Get Markets (API) · Rules Summary · Collateral Return · Finding Markets

5The order book, order types, and execution

5.1 The order book

The order book shows all resting orders (offers not matched immediately) with quantity and price. Bid = max a buyer will pay; Ask/offer = min a seller will accept. Kalshi's example: 5,010 to buy Yes at $0.74 and 75,000 to sell Yes at $0.71; viewed as No: 75,000 to buy No at $0.29 and 5,010 to sell No at $0.26.

5.2 Maker vs. taker

Every trade is between a maker (posts resting order, provides liquidity) and a taker (matches existing order, removes liquidity). Three archetypes: directional traders (price vs. true probability), hedgers (offset external risk; often pay a premium), market makers (earn the spread; opinion-neutral). Most participants today are retail directional traders and hedgers.

5.3 Order types

Quick-order example: 500 Yes at 10¢ and more at 12¢; a 1,000 order buys 500 @ 10¢ + 500 @ 12¢ = 11¢ average (panel shows the blend).

5.4 Flip selling

Flip selling lets you sell more than you hold by closing your position and opening the opposite side at once (hold 5 No, sell 10 → sells 5 No and buys 5 Yes). A settings toggle; useful for fast reversals.

5.5 Block trades (institutional)

A block trade is a large privately-negotiated trade. Both parties must be Eligible Contract Participants (ECPs); current minimum 25,000 contracts (no aggregation); "fair and reasonable" price; reported within 15 minutes; published separately.

5.6 Trading hours

Kalshi trades 24/7, except maintenance Thursdays 3:00–5:00 AM ET (~22h Thursdays).

Sources: The Orderbook · Quick Start: Market Data (API) · Who are you trading with? · Quick Orders · Limit Orders · Flip Selling · Block trades · Trading hours

6Fees (the exact math)

Kalshi charges a trading fee only on orders immediately matched (takers). Resting orders filled later are not charged the standard fee — unless the market is flagged for maker fees.

6.1 General trading-fee formula

fee = roundup( 0.07 × C × P × (1 − P) )

  P = price of the contract in DOLLARS (50¢ → 0.50)
  C = number of contracts
  roundup = round up to the next whole cent

Logic: the fee is 7% of the contract's expected earnings, where expected earnings = $1 × P × (1−P) per contract.

6.2 Maker-fee formula (maker-flagged markets only)

maker fee = roundup( 0.0175 × C × P × (1 − P) )

Charged only when the resting order executes; canceling is always free. Rounding over-payments above $10/month are reimbursed the next month.

6.3 Discounted tier — S&P 500 (INX) & Nasdaq-100 (NASDAQ100)

fee = roundup( 0.035 × C × P × (1 − P) )    # 3.5%, half the general rate

Applies to tickers beginning INX (INXD/W/M/Y/U…) or NASDAQ100 (NASDAQ100D/W/M/Y/U…).

6.4 The fee curve (general 7% tier)

6.5 Non-trading fees

• Settlement / membership: none.
• ACH deposit/withdrawal: free. Wire: Kalshi adds none (bank fees vary; wire withdrawals only ≥ $500,000).
• Debit card deposit: up to 2%. Crypto: third-party processor fees, disclosed first.

Sources: Fees · Kalshi Fee Schedule PDF (eff. Feb 5, 2026)

7Settlement, expiration & outcome determination

7.1 Timeline & payout

Each market's "Timeline and Payout" shows opened / closes / expected payout. Two closure types: closure when the outcome occurs (event-driven) and closure at a listed date/time (time-driven). Always check which applies.

7.2 How outcomes are determined

The rules, the data used, and its source are in each contract's terms. After expiration, Kalshi determines the outcome from rules + source data. Determination typically takes ~1 to >12 hours, governed by when the source agency publishes.

7.3 "Settled" & timing

A market settles when the official outcome is confirmed and finalized; winners pay out to cash. Most settle within a few hours (~3h). Keep distinct: close time (trading stops) vs determination time (official result confirmed). A market remaining open is not a signal about resolution — Kalshi often waits for the source agency.

7.4 Reading market rules (do this first)

• Rules summary — value measured, timeline, verification source; filterable by strike. It's a summary; full rules under "View full rules."
• Additional details — clarifications the markets team may add (not obligated to).

Sources: Timeline and Payout · Market Outcomes · Market FAQs · Market Rules · Additional Details

8Collateral return (capital efficiency for hedged baskets)

Collateral return (API flag netting_enabled) returns cash early when you hold hedged positions in mutually-exclusive or directional groups, lowering capital-at-risk. Off by default. Trade-off: once collateral is returned, you may be unable to sell that position before settlement. Lock-in quirk: the flag is set at your first order in an event (even an unfilled one) and is then locked for that event.

Source: Collateral Return

9Combos (multi-leg parlays)

A combo trades a custom combination of events as one position with its own order book. It resolves to the PRODUCT of its legs and pays a max of $1.00/contract.

• Pricing via RFQ (Request For Quote): you request, others quote live. Quotes can lag and change before execution; fills not guaranteed. Once filled, all combo trades are final.
• Settlement: after all legs resolve (typically 1–12h after the last). All legs $1 → $1; any leg $0 → whole combo $0.
• Scalar / DNP legs: a partial leg (e.g., DNP → $0.70) is not refunded — payout = product including the partial: $0.70 × $1 × $1 = $0.70. A blue arrow flags a scalar settlement.

Sources: Combos · Market FAQs

10Portfolio, cash & account data

• Portfolio = cash + market value of open positions + resting orders. Shows # contracts, average purchase price (cost basis), current value (what you'd get selling now, liquidity-permitting), and total payout if it resolves your way. Value fluctuates with prices even when idle. May be unable to "cash out" if no buyer at your price.
• Cash = unallocated, withdrawable funds. Deposits land here; withdrawals require cash.
• Trading history / documents = cumulative P&L statements for performance tracking.
• APY — idle balances can earn interest (Klear: ~3.25%, variable, eligibility).

Sources: Portfolio · Cash · Trading History · Kalshi Klear

11Data & graph tools (model inputs)

• Price graph — historical traded price (the probability path).
• Forecast graph — visualization of traders' current expectation of where the market expires (e.g., ~6.8 inches of snow). Not real-time; derived from trades.
• Live data graphs — real-time external feeds to assess value intraday. Caveat: a delay can exist vs the official verification source — outcomes are set by the source, not the live feed.

11.1 The API

Kalshi offers a REST API for your orders/trades/portfolio/history and public market data (all order books + stats). Public market-data endpoints need no authentication, at https://external-api.kalshi.com/trade-api/v2 (e.g. /series/{ticker}, /events, /markets?series_ticker=…, /markets/{ticker}/orderbook). Docs at docs.kalshi.com.

Sources: Price Graph · Forecast Graph · Live Data Graphs · Kalshi API · Quick Start: Market Data

12The trading-model layer

Interpretation built on the sourced facts above — the bridge from "how Kalshi works" to "how to build a model."

12.1 The fundamental edge equation

For a Yes taker buying at ask price a (dollars), with estimated true prob p̂:

  EV per contract = p̂ × ($1 − a) − (1 − p̂) × a − fee_per_contract
                  = p̂ − a − fee_per_contract
  where fee_per_contract uses the §6 formula on price a.

  • Trade Yes when  p̂ > a + fee   (underpriced)
  • Trade No  when  p̂ < bid_yes − fee   (overpriced; buy No at 1 − bid_yes)
  • No-trade band between thresholds = spread + fees
    (wider in illiquid / coin-flip markets)

12.2 Sizing

• Capital-at-risk per contract = your cost (Yes: a; No: 1 − a), or with collateral return, the basket's max loss.
• Consider fractional Kelly on the (p̂ − a) edge, heavily haircut — p̂ is uncertain and binary outcomes are high-variance. Never size as if p̂ were exact.

12.3 Strategy families (all supported by Kalshi mechanics)

12.4 Structural arbitrage checks a model should run

1. Yes/No coherence: if best_ask_yes + best_ask_no < 100¢ (net of fees), lock a profit.
2. Mutually-exclusive sum: sum of all Yes prices across the group should ≈ 100¢. Persistent < 100¢ (buy all Yes) or > 100¢ (No all) flags an arb.
3. Directional monotonicity: Yes price must be non-increasing as the threshold gets harder. A violation is an arb.

Always net out fees and check executable depth — apparent arbs often vanish after taker fees and slippage.

12.5 What the market price already "knows"

The market-implied probability bakes in the crowd's information, fees, and risk premia (hedgers overpay). You have edge only where you have better information, a better forecast, or you're providing a service (liquidity / risk-transfer) the crowd pays for. Be honest about which.

13Regulation & market integrity (deep dive)

13.1 The legal foundation

Prediction markets aren't new: early examples helped spur Congress to pass the Commodity Exchange Act (CEA), the federal statute governing commodities. The CEA covers swaps, futures, and Kalshi's main product, event contracts — which is why Kalshi is licensed and regulated by the Commodity Futures Trading Commission (CFTC), the U.S. commodities & derivatives regulator since 1974. The CFTC coordinates with the SEC, FTC, Treasury, and DOJ on oversight and enforcement.

13.2 What "DCM" actually obligates

Kalshi is regulated as a Designated Contract Market (DCM) — a CFTC-authorized marketplace for futures, swaps (incl. event contracts), and/or options. All trading is subject to the CEA plus 23 Core Principles (17 CFR Parts 38 & 39) that DCMs must adopt and enforce. The headline regulation categories:

13.3 Why it matters — the four benefits Kalshi cites

• Transparency — mandated publication of trading/market data, rule-change notices, and public disciplinary notices.
• Integrity — prohibitions on manipulative/deceptive/abusive trading + fair-and-orderly-market duties; universal KYC means Kalshi knows who trades and can detect fraud, laundering, manipulation, underage access, and insider trading.
• Security — funds held at a federally regulated clearinghouse under the same framework as other federal derivatives exchanges, with risk-management and anti-fraud safeguards.
• Trust — Kalshi submits trade data to the CFTC daily; the CFTC reviews it for manipulation/fraud, and Kalshi coordinates to avoid listing easily-manipulated markets.

Source: How are prediction markets regulated?

14Prohibited & unlawful trading

14.1 Federal law & CFTC regulations

CFTC 17 CFR 38.152 prohibits manipulative or fraudulent activity on exchange, misstatements/omission of material facts, defrauding another user, and price manipulation. The CFTC and DOJ have for years pursued insider-trading enforcement on CFTC-regulated markets; penalties have included multi-million-dollar fines and criminal enforcement. Where investigations suggest criminal intent, prosecutors commonly use the federal wire-fraud statute, 18 U.S.C. § 1343 — punishable by up to 20 years in prison, a $250,000 fine, and other penalties.

14.2 Kalshi's exchange-rule definition of insider trading

The Kalshi Rulebook defines insider trading as trading by:

1. Any person with access (or in a position to access) material non-public information before it's publicly available;
2. Any employee or affiliate of a source agency for any contract;
3. Any decision-maker (direct or indirect), or anyone with influence on the outcome of the underlying event.

Kalshi additionally prohibits any activity that is fraudulent, abusive, manipulative, or deceptive, or that operates as a fraud or deceit upon another person.

14.3 How Kalshi detects & enforces

• KYC at onboarding — every account submits identifying info before any trade; activity is linked to identity from the start (17 CFR 38.151–.159).
• Surveillance — the Compliance team monitors internal flagging systems (patterns, timing, positions, market history), 24/7, across every open market, analyzing each position against price movement, volume, and public info in real time.
• Third-party tooling — vendors including IC360 share alerts triggered on other sports-market platforms, warning Kalshi about activity live elsewhere.
• Investigation flow — an uncleared flag → account frozen (funds locked) → Compliance + Legal review & interview → if reasonable evidence, a notice of charges → the user can settle or face Kalshi's Disciplinary Panel; outcomes are reported to the CFTC, with informal law-enforcement referrals where appropriate.

Source: Prohibited and Unlawful Trading Activity

15Regulated vs. unregulated prediction markets

Onshore, CFTC-regulated markets (Kalshi) and offshore, unregulated markets differ profoundly despite surface similarity. Know which you're modeling — their data integrity and contract universe are not comparable.

• Licensing barrier — under 7 U.S.C. § 7, a U.S. prediction market must obtain a DCM license (hundreds of regs, 23 Core Principles, CFTC-approved rules). Offshore markets skip all of it.
• User identification — CFTC rules (17 CFR 38.153) require identifying each user; offshore venues often collect none → anonymous trading, weak policing, defrauded users.
• Illicit-activity magnet — research indicates offshore venues are targeted by wash trading (opposite sides of the same market to manipulate price — illegal here) and offshore insider trading U.S. rules would bar.
• Prohibited contracts — the CEA (17 CFR 40.11) lets the CFTC reject contracts on assassinations, war, or terrorism; offshore venues face no such limits.

Source: Regulated vs. Unregulated Prediction Markets

16Prediction Markets 101 — risk & strategy discipline

16.1 Treat trading like a business model

Before trading, answer: Which markets, and why? How much can I responsibly risk? When do I trade? What do I do when trades go well / poorly? When do I close, pause, or stop? These define your information edge, profit/exit goals, comfortable capital, financial strategy, and plans for both profits and losses.

16.2 Read BOTH rule sets (the highest-impact habit)

Each market has two rule sets: Market Rules (the short version, beneath the price display) and Contract Rules (the full linked PDF).

16.3 Market risk & price behavior

• The outcome can't be known in advance → unexpected losses are possible. A change in an event's likelihood may not move the contract price, blocking a profitable exit or a loss-cut.
• Event-contract prices depend on expected probability, which makes traditional derivative-pricing models inapplicable. Moves are often driven by news / new information and can be sudden.

16.4 Fully collateralized — and timing matters

• All trades are fully collateralized: you deposit the full max-loss up front, and it stays locked for the duration you hold the position — this is your capital-at-risk (see §8, §12.2).
• Resolution timelines vary (a Fed announcement next week vs. a "by year-end" market) — factor holding period into strategy.
• Spreads widen as the outcome gets more certain, making a losing position harder to exit. Watch volume and spreads (see §5).

16.5 Position sizing discipline

• The goal isn't to win every trade — every trader loses many; aim for more wins than losses and control loss size.
• Always decide how much you're willing to lose and your next step if you do.
• Don't deploy all capital on one trade — spread it; consider starting small and scaling as a working strategy emerges.

Source: Prediction Markets 101

17Glossary (Kalshi terms of art)

18Common modeling pitfalls

1. Treating market price as truth. It's fee/spread-laden and crowd-biased — your p̂ must be independent.
2. Ignoring the fee parabola. Breakeven edge is largest at 50¢; many coin-flip "edges" don't survive fees.
3. Pricing off top-of-book. Real fills walk the book; model slippage from depth.
4. Forgetting the reciprocal book. Yes ask = 100 − No bid; the API returns bids only.
5. Assuming combo legs are independent. Same-game legs are correlated; the product overstates "all-happen."
6. Underweighting resolution-source risk. Exact source/definition/timestamp can flip a "sure" market.
7. Confusing close vs determination time. A market open after the event ≠ outcome undecided.
8. EV without capital-at-risk / collateral return. ROI-per-dollar, not just EV, ranks strategies.
9. Counting Yes and No as separate exposures. Buying Yes ≡ selling No — one position.
10. Assuming binary settlement always. Scalar / DNP markets settle fractionally.
11. Modeling a trade the user is legally barred from. If the user is an insider on the resolution source or can influence the outcome, there is no tradable "edge" — only a bannable, potentially criminal act. Flag it (§14).

19Papers & further reading

The theory behind the mechanics, with direct PDFs. Start with the two Kalshi-specific papers.

Most directly relevant to Kalshi

• Makers or Takers: The Economics of the Kalshi Prediction Market — Bürgi & Whelan, GWU 2026. 300k+ contracts; documents the favorite-longshot bias on Kalshi (cheap contracts overpriced, expensive underpriced) and builds the fee formula into breakeven. Start here.
• Kalshi and the Rise of Macro Markets — Federal Reserve, 2026. Kalshi's CPI/FOMC/payrolls markets carry real predictive signal.

Foundations — price-as-probability and its limits

• Prediction Markets — Wolfers & Zitzewitz, 2004. The canonical survey.
• Interpreting the Predictions of Prediction Markets — Manski, 2004. The skeptic's case: price ≠ mean belief.
• Do Prediction Markets Produce Well-Calibrated Forecasts? — Page & Clemen, 2013. Calibration is good near expiry, biased far out.

Sizing math — Kelly

• Application of the Kelly Criterion to Prediction Markets — arXiv. Optimal Kelly when price ≠ your belief.
• Fortune's Formula, or the Road to Ruin? The Generalized Kelly Criterion — Whelan, UCD 2023. Clean derivation + the multi-outcome (generalized) Kelly extension.
• Modified Kelly Criteria — Swartz. How much to shrink Kelly when your probability estimate is uncertain (with R code).

Arbitrage & AMM pricing

• Unravelling the Probabilistic Forest: Arbitrage in Prediction Markets — AFT 2025. Formalizes the two arb types; ~$40M realized on Polymarket.
• Logarithmic Market Scoring Rules — Hanson. The original LMSR/AMM paper (Polymarket's AMM, not Kalshi's order book).

Practitioner whitepapers

• A Quantitative Framework for Polymarket — Horacle Capital. Most quant-rigorous; Brier-tiered Kelly + Avellaneda-Stoikov market-making adapted to [0,1]. Maps closely to Kalshi.
• Forecast Efficiency in Kalshi Weather Markets — 40k contracts; a 96.6% win-rate strategy nets only 0.62% ROI after fees. The "edge dies at the fee line" case study.

Official Kalshi docs — Fee Schedule PDF, API docs, and the LLM index: see the Source index (§20).

20Source index