Kalshi paper trading simulator

Kalshi Paper Trading on Live Displayed Books

Forward-test a Kalshi strategy with virtual cash while pricing its simulated fills from displayed production-market books. Run a lab rule or send signals from your own bot, then inspect the equity curve, positions, trade log and automatic $0/$1 settlement in one workspace.

Start paper trading free See the paper-trading workflow

Kalshi paper trading simulates a strategy without risking real funds. DepthFeed differs from Kalshi's official demo environment: it records virtual strategy fills from displayed live-market order books, while Kalshi states that prices and behavior in its separate demo environment may not reflect real markets.

What the Kalshi paper-trading engine actually supports

DepthFeed treats paper trading as a controlled strategy experiment. These current parameters distinguish the forward-testing engine from a generic practice balance.

crypto assets

BTC, ETH, SOL, XRP, DOGE, BNB and HYPE are selectable when eligible Kalshi series are available.

market windows

Lab rules target supported 15-minute, 1-hour and 24-hour Kalshi crypto windows.

sizing methods

Compare fixed-dollar, fixed-fractional and Kelly-style bankroll rules.

120/min

webhook limit

Each custom strategy accepts bounded signals using Kalshi tickers and idempotent client IDs.

Product configuration and worker limits verified in the DepthFeed Paper Trading implementation on July 26, 2026. Eligibility depends on currently available Kalshi markets and plan access.

Kalshi paper trading at a glance

Fills: Displayed live book — buys at ask, sells at bid
Settlement: Automatic at resolution ($0 / $1)
Signal webhook: POST /paper/hook/{token} · 120/min
Strategies per plan: 1 free · 5 Quant · 15 Research · 40 Desk
Starting balance: $100–$1,000,000 virtual (default $10,000)
Mark cadence: ~90 seconds
Lab assets: BTC · ETH · SOL · XRP · DOGE · BNB · HYPE when eligible
Lab windows: 15m · 1h · 24h
Entry rules: Time range · level cross · dip recovery
Sizing: Fixed dollars · fixed fraction · Kelly-style
Risk controls: Take profit · stop loss · max open · trailing drawdown
Research record: Equity curve · positions · trade ledger · CSV export

How Kalshi paper trading works

Choose the environment that matches the question

Kalshi's official demo and DepthFeed Paper Trading answer different questions. Use the official demo to test demo credentials, authentication, request formatting and exchange API behavior with mock funds. Kalshi explicitly notes that demo prices and market behavior may not reflect real markets. Use DepthFeed when the research question is whether a fixed strategy behaves acceptably when its signals are priced from displayed production-market books and retained as one forward record. DepthFeed does not submit an order to either Kalshi environment.

Define the strategy before collecting results

Write down the asset, market window, Yes or No side, entry rule, size, exits and failure threshold before starting. For example: buy Yes on eligible Kalshi BTC 15-minute markets when the displayed price recovers from below 0.35 to 0.50, risk no more than 5% of current equity per entry, and halt at a 20% trailing drawdown. This converts a trading idea into a falsifiable test and prevents the rule from drifting after every win or loss.

Run a server-side lab rule without Kalshi credentials

Lab mode monitors eligible Kalshi crypto markets and evaluates three entry families on the paper worker cadence: time-before-close inside a price range, a cross above or below a configured level, or recovery after a defined dip. Choose Yes or No, the stake or bankroll-sizing method, optional take-profit and stop-loss levels, maximum open positions and a drawdown circuit breaker. Because it is a DepthFeed simulation, no Kalshi account, demo key or production credential is required.

Connect custom Kalshi logic by webhook

Webhook mode accepts custom signals from a model, service or trading bot. Send the action, venue, Kalshi ticker, dollar size and an optional client ID to the strategy's secret endpoint. The token is shown once and can be rotated; the client ID makes a retry idempotent. A caller cannot provide the fill price. DepthFeed reads the displayed Kalshi book and records the virtual action, preventing the signal generator from grading itself with an unrealistically favorable price.

Model the Yes/No book consistently

Kalshi's order book is expressed through Yes and No prices that complement each other. DepthFeed normalizes the displayed quote needed for the selected side, uses the ask for a virtual buy and the bid for a virtual sell or close, and records the market ticker with the trade. This captures the observed spread at decision time. It does not assume that the midpoint was executable, but it also does not walk multiple levels or consume real contracts.

Carry the same assumptions out of the backtest

A Backtest Lab rule can prefill a new paper strategy with the same venue, asset, market window, side, entry condition, size and risk controls. Review the configuration before creating it and preserve the original version. The historical replay and the forward test should differ because the samples differ, not because the rule was quietly optimized after seeing paper outcomes. When a field changes, start a separately named version and compare the two records.

Measure bankroll behavior and failure modes

Fixed-dollar sizing tests a constant notional but can continue risking the same amount as equity falls. Fixed-fractional sizing adjusts exposure with the current bankroll. Kelly-style sizing estimates a fraction from a stated edge, probability or empirical hit rate and caps it at a chosen maximum. Take-profit, stop-loss, concurrency limits and the trailing drawdown circuit breaker expose whether apparent edge survives concentration and losing streaks. A strategy that trips its circuit breaker is flattened and halted for review.

Review complete resolved samples, not dashboard snapshots

The strategy view combines virtual cash, equity, realized and unrealized P&L, positions, marks and the trade ledger. Separate open exposure from realized outcomes and inspect whether profit comes from one contract, one market window or one short regime. The newest 50,000 trades are retained for up to 180 days, while CSV export provides a durable record. Save the test configuration and date range with the export so the result can be independently reproduced.

Use a staged live-readiness decision

A successful paper test is evidence about signal behavior under the simulator's assumptions; it is not evidence that a production Kalshi order will receive the same fill. Before considering live execution, compare intended size with full displayed depth, include applicable fees and slippage, test authentication and order lifecycle behavior in Kalshi's official demo, and define a much smaller live canary. DepthFeed keeps those stages separate and never converts the paper ledger into a real Kalshi position.

Worked example: test a Kalshi recovery rule

This example separates API testing from strategy validation. Its settings illustrate a reproducible process and are not a profitability claim or trading recommendation.

1State the market hypothesis
Suppose the hypothesis is that a sharp decline in the Yes price on a Kalshi BTC 15-minute market sometimes overreacts, and a recovery to 0.50 carries useful continuation information. Define the dip threshold, recovery threshold, eligible series, maximum time remaining and what would invalidate the idea. Choose the minimum number of independent resolved markets before reviewing results so an early streak cannot end the experiment.
2Use Kalshi demo for the exchange integration question
Separately confirm that the intended live system can authenticate, discover markets, submit and cancel orders, process order updates and handle errors in Kalshi's official demo environment. That validates the integration surface. Do not use demo P&L as proof of the strategy because Kalshi states that demo prices and behavior may not reflect real markets.
3Create the DepthFeed forward test
Select Kalshi, BTC, the 15-minute window, Yes and the dip-recovery rule. Preserve the thresholds from the backtest, choose a virtual starting balance, cap simultaneous positions and set the trailing drawdown circuit breaker. If the logic is more complex than the lab rule, create a webhook strategy and send a valid Kalshi ticker with each signal.
4Inspect how each virtual fill was graded
When the recovery condition occurs, verify the recorded ticker, timestamp, selected side, price, shares and notional. A virtual buy is priced from the displayed ask rather than an optimistic midpoint; a close uses the bid. If a signal is retried, use the same client ID so the endpoint records it once. Investigate stale or missing quotes instead of substituting an estimated price.
5Compare the backtest with unseen outcomes
After enough markets settle, export the paper trades and compare entry distribution, expectancy, drawdown, losing streaks and concentration with the original historical test. Segment by price band and time remaining. If the forward result is materially worse, test whether spread, regime change or selection differences explain it before changing the rule.
6Require both strategy and operations gates
A live canary should require two independent passes: the DepthFeed paper record meets the precommitted performance and drawdown criteria, and the Kalshi demo integration handles authentication, lifecycle events and errors reliably. Then apply conservative live fees and size-dependent slippage, confirm account and jurisdiction requirements, and start far below the size used in the virtual account.

Where Kalshi paper trading differs from live execution

Displayed production-market quotes improve relevance, but the result remains a model. Treat each limitation as a required adjustment or operational test before live use.

Top-of-book pricing does not guarantee fillable size

A displayed ask or bid can be smaller than the intended live order. The paper engine does not reserve that liquidity or walk the ladder. A production order may fill partially, cross several prices or remain unfilled. Review full Yes/No depth at the intended notional for any size-sensitive strategy.

Maker queue and cancellation behavior are not simulated

Resting live orders compete for queue position, and displayed contracts can disappear before an order arrives. The paper record models an action from the quote observed by the simulator; it does not estimate maker fill probability, cancellation races or the exact sequencing of exchange events.

The approximately 90-second cadence limits the questions it can answer

Lab evaluation and position marks are deliberately slower than an execution system. They are useful for strategy-level forward testing but cannot validate sub-second signals, reaction latency or the exact path within each interval. Use event-level historical replay and dedicated instrumentation for those questions.

Production fees and order lifecycle remain external

Paper P&L does not include every cost or failure mode of an authenticated Kalshi order. A live estimate should account for applicable fees, spread, depth slippage, rejected or delayed requests, partial fills, maintenance windows and reconciliation between local state and exchange state.

Overlapping markets reduce effective sample size

Several positions tied to the same BTC move are correlated even when their tickers differ. Count independent market exposures and regimes, not merely rows in the trade ledger. A large trade count can still represent a small amount of distinct evidence.

DepthFeed is independent from Kalshi

DepthFeed does not operate Kalshi's demo, issue Kalshi credentials, place Kalshi orders or guarantee live results. The simulator is a research tool, not financial advice; venue rules, account eligibility and the decision to risk money remain the user's responsibility.

Start a paper strategy with no real funds

Explorer includes one concurrent paper strategy at no cost. Create a rule in the dashboard or connect your own bot by webhook; no card or trading-wallet connection is required.