How to Make Money on Kalshi: Probability, Costs, and Risk

To make money on Kalshi over a meaningful sample, you must buy outcomes for less than their eventual value or sell them for more, after accounting for fees, the bid-ask spread, slippage, taxes, and mistakes in your forecast. That sounds simple; consistently doing it is not. Event outcomes are uncertain, other traders react to much of the same information, and even a sound forecast can lose on any single contract.

Think in probabilities, not certainties. If a YES contract can settle at $1 and the best executable ask is $0.43, a 43% forecast is not an edge. Your estimate must be above the all-in break-even probability, and the difference must be large enough to survive reasonable model error. The displayed last price is not necessarily the price available for the number of contracts you want.

Many event contracts present two sides, YES and NO, tied to an objective question. The market's rules define the measurement source, threshold, observation window, and settlement process. Read those rules before interpreting the title. A contract that sounds like it asks whether an event “happens Friday” may actually use a named publication, a specific timezone, and a fixed release window. Being right about the headline but wrong about the contract definition still loses.

For a simplified $1-settlement binary bought and held to resolution, buying YES at price q produces a maximum gross gain of $1 − q if YES resolves and a maximum gross loss of q if NO resolves. Multiply both by the number of contracts, then subtract applicable costs. A NO position has the same logic using its own executable price. Positions closed before resolution instead realize the difference between entry and exit fills, less costs; the final event outcome no longer determines that closed position's result.

Kalshi's official explainer describes prediction-market pricing, while its market-rules guidance explains why contract-specific terms matter. The CFTC's prediction-market education page also emphasizes rules, costs, financial risk, and using only risk capital.

Start with the current order book. The ask is the lowest displayed price at which someone is offering to sell; the bid is the highest displayed price at which someone is offering to buy. Their difference is the spread. A marketable order usually crosses that spread. A larger order may consume several price levels, so its volume-weighted average fill can be worse than the top quote. That difference is slippage.

Then inspect the current venue fee schedule. Kalshi publishes a fee explainer, but this guide intentionally does not quote a rate: fee treatment can depend on the product, order, price, or current schedule and can change. Calculate fees from the official information shown when you trade. Also consider exit costs if your plan depends on closing before settlement.

A quick break-even formula for a simplified YES contract held to a $1 or $0 resolution is: purchase price per contract plus per-contract costs. For example, if 100 contracts cost $42 and you model $1.20 of total costs, the all-in outlay is $43.20. Dividing by the $100 possible settlement value gives a 43.2% break-even probability. If your estimate is 44%, the apparent edge is only 0.8 percentage points—probably too fragile to survive ordinary estimation error.

Better-defined research

Some traders specialize in a narrow domain and build a repeatable forecast from primary data, base rates, and a documented model. Specialization can make it easier to notice when a market overweights a vivid headline or ignores how the official measurement is constructed. The test is not whether the narrative sounds persuasive; it is whether forecasts made the same way are calibrated over time.

Faster interpretation, not prohibited information

Scheduled releases can move prices quickly. A prepared trader may know in advance which table, timestamp, or calculation matters and can interpret public information efficiently. That is different from trading on material nonpublic information or violating venue rules. Build your process around lawful public sources and review the applicable rules.

Relative-value consistency

Related contracts sometimes imply probabilities that cannot all be reasonable together. If a set of mutually exclusive ranges has executable prices whose all-in total is far from 100%, or if a broad event looks cheaper than a narrower event contained within it, investigate. Do not assume the difference is free arbitrage: rule wording, settlement sources, timing, fees, thin depth, and incomplete fills can explain it.

Patient execution

A forecast can be right and the trade still be poor because the entry price was too high. A limit order sets a price boundary and may capture a better entry, but it may never fill. It can also fill precisely when informed traders are moving the other way—adverse selection. The official limit-order guide explains the order mechanics; it does not turn passive orders into guaranteed bargains.

These examples show why win rate alone is inadequate. A trader can win often but lose money if the wins are small and the losses are large. Another can be right less than half the time and still have positive expectancy if the prices and payoff sizes compensate. Track forecast, fill, size, costs, and outcome together.

1. Translate the rule into a test. Write the exact condition that makes YES resolve, the source that decides it, the relevant time window, and any edge cases. If you cannot explain the test in one paragraph, do not trade it yet.
2. Set a prior. Start with a relevant historical base rate before reading the freshest headline. Record why the comparison set is relevant and how much uncertainty it has.
3. Update with primary evidence. List evidence that raises the probability and evidence that lowers it. Avoid counting three articles based on the same original source as three independent signals.
4. State a range. Record a central probability and a credible low/high range. Compare the conservative end—not only the center—with the all-in break-even price.
5. Inspect executable depth. Use bids, asks, and size at each level. Kalshi's order-book guide and its developer documentation are useful references for how displayed book data is represented.
6. Precommit size and exit rules. Define maximum loss, thesis-invalidating evidence, time-based review points, and whether the default plan is to hold or exit. Do this before the price moves.
7. Journal the decision. Save the rule version, timestamped quotes, forecast, evidence, intended order, actual fills, and later outcome. Review the process even when a bad decision happens to win.

For a deeper implementation of forecasting and journaling, use the Kalshi trading-strategies framework. It is intentionally separate from this profitability explainer: the framework focuses on generating and testing decisions, while this guide focuses on the economics that any strategy must clear.

Set a risk budget in dollars before converting it to contracts. For a simple YES purchase held to resolution, the purchase outlay plus costs is a practical starting estimate of maximum loss. If your risk budget is $25 and the ask is $0.62, 40 contracts cost $24.80 before fees. That leaves almost no cost buffer, so the permitted size must be lower. Do not round up simply because the platform accepts another contract.

Also cap exposure by theme. Five contracts tied to the same election, inflation release, storm, or court decision are not five independent ideas. If the common assumption fails, all can lose together. Calculate a scenario loss across correlated positions and compare it with your total risk limit.

• Use money you can afford to lose; never use rent, emergency savings, or borrowed funds.
• Reduce size when rule interpretation, model inputs, or executable liquidity is uncertain.
• Do not increase a position merely to recover a prior loss.
• Keep enough uncommitted capacity that one surprise does not force unrelated exits.
• Judge sizing by the planned worst case, not by the most likely outcome.

The CFTC's event-contract education recommends understanding costs and risks, reviewing contract-specific rules, monitoring open positions, and trading only with risk capital. Refract Funding's own risk disclosure applies to its simulated program and should be read before using it.

• Confusing confidence with probability. “I think YES” is not a numerical estimate and cannot be compared with a price.
• Ignoring the contract rules. A correct real-world prediction may not match the named source, threshold, or observation period.
• Chasing a move. New information can improve a forecast while the new price worsens the trade. Recalculate instead of assuming momentum confirms the thesis.
• Using non-executable prices. Midpoints and last trades can make historical results look cleaner than the fills the book offered.
• Overtrading a small edge. Costs and forecast noise can consume a narrow advantage, especially when repeatedly crossing the spread.
• Oversizing correlated ideas. Different tickers can still depend on one underlying event or source.
• Outcome bias. A lucky win does not validate a bad process, and a foreseeable loss does not automatically invalidate a well-priced decision.
• Moving the thesis after entry. If every adverse fact gets explained away, the original exit rule was not real.

The cure is not more activity. It is a smaller number of auditable decisions, a large enough sample, and honest review. If you have not recorded your forecast before seeing the outcome, you cannot reliably measure calibration afterward.

A useful practice period is not a contest to maximize pretend profit. Pick one market family, write forecasts before entry, include modeled costs, use available depth rather than midpoint fills, and cap risk. Review at least 30 resolved forecasts before drawing conclusions, and more when the strategy trades infrequently or outcomes are highly correlated.

Refract Funding provides a simulation that references published market data. Its trades, fills, balances, positions, Evaluations, and funded-stage accounts are simulated. Review the program guide, terms, and risk disclosure rather than treating an article or hypothetical result as a promise. Passing an Evaluation does not guarantee funded access or a payout; program rules, eligibility, review, regional availability, and approval apply.