Prediction Market Paper Trading Guide

Nearly every stock broker hands you a paper trading account on day one. Prediction markets don't. Kalshi runs a demo environment most traders have never heard of, Polymarket has no official simulator at all, and the third-party tools that fill the gap are thin. That's a problem, because prediction market paper trading matters more than paper trading stocks ever did: in these markets, fees and slippage decide whether an edge is real, and the only cheap way to find out is to simulate honestly.

This guide covers what paper trading means in prediction markets, how Kalshi's demo environment works, what Polymarket traders use instead of a demo account, where simulated fills quietly lie to you, what a serious paper setup must track, and how Claw Arbs ships a built-in paper trading mode that rehearses arbitrage and value betting end to end with zero risk.

What Is Prediction Market Paper Trading?

Paper trading means running a strategy with simulated money while recording every entry, fee, and settlement as if it were real. You watch real markets, log the fills you would have gotten, and measure whether the strategy makes money before a single dollar is exposed.

In prediction markets it comes in two flavors, and the difference matters:

Sandbox environments. A separate copy of the exchange with a fake balance. Kalshi's demo environment works this way. It's ideal for testing API integration and order mechanics, but the prices you trade against are not production order flow.
Shadow mode against live prices. The strategy runs against real production quotes, but fills are simulated and logged instead of sent to the exchange. This is what a spreadsheet does crudely and what a built-in paper mode does properly. It answers the question that matters: would this strategy have made money on the real market today?

Both are useful. A sandbox tells you your code works. Shadow mode tells you your edge works. Most traders need the second answer far more than the first.

Why Paper Trading Matters More Here Than in Stocks

In equities, paper trading mostly tests psychology and mechanics. Commissions are mostly zero, spreads on liquid names are a penny wide, and a simulated fill on a large-cap stock is close enough to a real one. The open question is whether your thesis is right.

Prediction markets flip that. The open question is whether your edge survives the costs, because the costs are large relative to the edges people actually trade:

Kalshi fees follow a tiered schedule that peaks near 50 cents, exactly where competitive sports contracts trade. At the default tier, a contract near the middle of the range costs roughly 1.7 cents per contract in fees.
Polymarket charges no trading fee, but it runs on-chain (USDC on Polygon) and every transaction costs gas, which stings on small positions.
Spreads run 1 to 3 cents on plenty of books. If your gross edge is 3 cents, the spread alone can be your entire profit.

Run the numbers on a typical cross-venue arb. Kalshi offers Yes on a game at 47 cents while the opposite side costs the equivalent of 50 cents on Polymarket. Total outlay: 97 cents for a position that pays $1 at resolution whichever way the game goes. That's about a 3% gross edge. Kalshi's fee takes roughly 1.7 cents of it, gas takes a sliver more, and the 3% gross edge is really about 1.3% net. Still worth trading, but the costs consumed more than half the edge before slippage even entered the picture.

Paper trading prediction markets isn't testing whether your probabilities are right. It's testing whether your edge survives fees, spreads, and slippage. Most edges don't.

That's why demo trading prediction markets with honest fee accounting is not optional. A strategy that looks profitable on raw prices can be a reliable loser after costs, and paper trading is the cheapest way to find out.

Kalshi Paper Trading: The Demo Environment

Kalshi is the US-regulated (CFTC) prediction market that prices contracts in cents, and it offers something most venues don't: an official demo environment. The Kalshi demo account lives at demo.kalshi.co, separate from production, with its own API keys and a simulated balance to trade with.

That makes Kalshi paper trading the natural starting point if you're building your own bot. You can test authentication, order placement, order types, cancellations, and error handling without a cent at risk. If your code has a bug that submits ten orders instead of one, the demo environment is where you want to meet it.

Two caveats. First, the demo is a separate environment, so treat it as an integration testbed, not proof that your edge is real. The order flow you trade against there is not production liquidity, and profits in a sandbox don't imply profits on the live exchange. Second, the details change over time, so check Kalshi's current developer docs rather than relying on any third-party writeup, including this one.

If you're evaluating existing software instead of building your own, our Kalshi trading bot guide covers the API, order types, and what separates a weekend script from a production system.

Polymarket Paper Trading: No Official Simulator

Polymarket has no demo account, and the architecture explains why. It's a crypto prediction market where trades settle in USDC on Polygon, and there's no sandbox chain mirroring production order flow. You can't paper trade on the venue itself, so traders improvise:

Third-party simulators. A handful of Polymarket simulator tools exist that let you open hypothetical positions. They're thin: basic position tracking against last-traded prices, usually no orderbook depth, no gas modeling, and no settlement logic. Fine for testing a directional hunch, useless for testing an arbitrage or value betting edge.
Manual tracking. Record the timestamp, the best ask, and the size available at that price, then compute what the fill would have cost including gas. A spreadsheet works, and it's honest as long as you log depth, not just price.

Manual tracking has a failure mode worth naming: you write down the price you saw, not the price you would have gotten. If the best ask showed 55 cents with $40 of depth behind it and your strategy wanted a $400 position, your real fill was several cents worse than your notebook says. Paper records that ignore depth quietly overstate every result.

The Limits of Paper Trading: Simulated Fills Are Optimistic

Every paper trading setup, including good ones, shares one bias: simulated fills are optimistic. Three things no simulator reproduces:

Queue position. On a real orderbook, other orders sit ahead of yours at the same price. Paper mode assumes you were first in line.
Partial fills. The simulator fills your whole order at the quoted price. Live, you might get 30 of the 100 contracts you wanted, with the rest filling worse or not at all.
Adverse selection. The fills you get most easily are the ones a sharper trader was happy to give you. When the price is about to move against you, your order fills instantly; when it's about to move your way, it doesn't. Paper mode fills both cases equally.

Here's what that does to real numbers. Take the arb from earlier: paper mode records a fill at 47 cents, a 3-cent gross edge, and about 1.3 cents net after the roughly 1.7-cent Kalshi fee. Now run the same trade live. By the time your order arrives, the 47-cent ask is gone and you fill at 48 cents. The gross edge drops to 2 cents, the fee barely moves, and the net edge lands around 0.3 cents. One cent of slippage erased about three-quarters of the profit.

So apply the honest rule: a paper edge shrinks live. If a strategy barely clears breakeven on paper, it loses money live. You want paper results with enough margin that realistic slippage still leaves a profit, which is why serious systems let you charge yourself a slippage penalty inside the simulation.

What a Good Paper Trading Setup Must Track

Spreadsheet or software, a paper trading setup only means something if it records the right fields:

Entry price and timestamp. The exact quote at the moment of the signal, not the price a minute later when you got around to logging it.
Available depth. The size at your price level. A simulated fill is only realistic up to the quantity that was actually on the book.
Fees at your tier. Kalshi's fee depends on the contract price and your volume tier, and it peaks near 50 cents. Model the fee on every fill, not as a month-end adjustment.
A slippage assumption. Decide up front how many cents of slippage you charge yourself per fill and apply it consistently. Zero is not a defensible number.
Settlement. Track every position to resolution. A paper trade isn't complete until the market resolves and the P&L includes the fee on winnings.

You can model net edge after fees for any contract price with our free prediction market calculator.

Kalshi Demo vs Polymarket Simulators vs Claw Arbs Paper Mode

Here's how the three realistic options for demo trading prediction markets compare:

Aspect	Kalshi Demo	Polymarket Options	Claw Arbs Paper Mode
What it is	Official sandbox, separate from production	Third-party simulators or a spreadsheet	Simulated execution built into the bot
Balance	Simulated, with its own API keys	Hypothetical, tracked by hand	Simulated
Prices	Demo environment, not production order flow	Real quotes only if you record them yourself	Real live prices via WebSocket feeds (~25ms)
Fees and settlement	Check Kalshi's current docs	Only if you do the math yourself	Fees modeled on every fill, auto-settles on resolution
Arbitrage rehearsal	Kalshi only	Polymarket only	12 venues, cross-venue arbs end to end
Cost	Free	Free	Free during the current testing period

The pattern is clear: the Kalshi demo is an integration sandbox, the Polymarket options are bookkeeping, and a built-in paper mode is the only option that rehearses a complete strategy against live production prices.

How Claw Arbs Paper Trading Mode Works

Claw Arbs is arbitrage and value betting software that both detects and executes. Its paper trading mode isn't a bolted-on toy; it's the same pipeline with the final step swapped from real orders to simulated ones.

When paper mode is on:

The full detection pipeline runs against live prices. WebSocket feeds on Kalshi and Polymarket update in about 25ms, and an 11-book sharp odds aggregator supplies the reference lines. Nothing about detection is simplified.
Every simulated fill is recorded with fees modeled. The P&L you review is net of Kalshi's tiered fees and gas assumptions, not a gross number that flatters the strategy.
Positions settle automatically when markets resolve. The step most spreadsheets skip happens on its own, so your paper record is made of complete, closed trades.
The safety systems stay on. Kill switch, exposure caps, naked-leg tracking, and cooldowns all apply in paper mode. You rehearse the real system, not a simplified copy of it.

Execution mode is a switch, not a rebuild: run paper only, live only, both at once, or detection only. Running both at once is underrated. Keep live stakes small while paper mode runs the full strategy size, then compare the two P&L streams to see exactly what slippage and missed fills cost you in the real world.

Claw Arbs ships with 12 venues pre-installed: Kalshi, Polymarket, Polymarket US, Predict.fun, and SX Bet on the prediction market side, plus Sportmarket, Stake.com, BC.Game, Cloudbet, Bet105, Pinnacle (PS3838), and BetInAsia. So the strategies worth rehearsing, like Kalshi vs Polymarket arbitrage and prediction market value betting, run end to end on paper: detection, simulated execution, fee accounting, and settlement.

Over 1,000 users run it, and it's free during the current testing period, so paper mode costs exactly what a simulator should: nothing. The free arbitrage betting bot guide covers what's included.

Graduating From Paper to Live

Paper trading is a phase, not a destination. The sensible path from simulated money to real money looks like this:

Set a gate before you start. For example: two weeks of paper trading with positive net edge after modeled fees and a 1-cent slippage penalty. If paper can't clear the bar, live won't.
Go live at small stakes. Size positions so a bad week is tuition, not damage. The goal of the first live stretch is data, not profit.
Compare live fills to paper fills on the same signals. This is the single most useful diagnostic in trading automation. If paper says 47 cents and live averages 48, you've measured your true slippage and can set thresholds to match.
Watch slippage per venue. It isn't uniform. If one venue consistently fills worse than the simulation, raise the minimum edge you demand there rather than abandoning the strategy.
Scale only when live tracks paper. If live net edge lands within about half a point of paper over a meaningful sample, size up gradually. If the gap is wider, fix thresholds and sizing first.

Speed decides how much slippage you eat between detection and fill. Closing that gap is the whole point of automated arbitrage execution, and it's why Claw Arbs goes from detection to order in about 20ms.

The Bottom Line

Prediction market paper trading is the cheapest audit your strategy will ever get. A Kalshi demo account proves your integration works. A Polymarket simulator or a spreadsheet can log a hunch. But the only test that predicts live performance runs your full strategy against live prices with honest fees, honest depth, and a slippage assumption you'd defend out loud.

That's what Claw Arbs paper mode delivers: the complete detection and execution pipeline, real prices, simulated money, automatic settlement, and the same safety systems you'll trade with live. Rehearse until the numbers hold up. Then flip the switch.

FAQ

Does Kalshi have a paper trading or demo account?

Yes. Kalshi runs a demo environment at demo.kalshi.co that is separate from production. It uses its own API keys and a simulated balance, so you can test order logic and bot plumbing without risking real money. Details change over time, so check Kalshi's current developer docs before building against it.

Is there a Polymarket simulator or demo account?

No official one. Polymarket trading happens on-chain with USDC on Polygon, and there is no sandbox that mirrors it. A few third-party simulators exist but they are thin. Most traders either track hypothetical fills manually or use software with a built-in paper trading mode, like Claw Arbs.

How accurate is prediction market paper trading?

Paper trading is optimistic by design. Simulated fills assume you get the quoted price with no queue position, no partial fills, and no adverse selection. Expect your live edge to come in below your paper edge, especially on thin markets. Treat paper results as an upper bound, not a forecast.

Does Claw Arbs include a paper trading mode?

Yes. Claw Arbs ships with a built-in paper trading mode that runs the full detection pipeline against live prices with simulated money. It records every simulated fill with fees modeled, settles positions automatically when markets resolve, and can run alongside or instead of live execution.

Prediction Market Paper Trading: Test Strategies Before Risking Real Money