Kalshi review: a 0.07 taker fee, referral credits that expire in 7 days, and no deposit of ours behind any of it

We have not funded a Kalshi account, so nothing on this page tells you how fast Kalshi pays out or how it answers a complaint. What we can check is the arithmetic. The taker fee on Kalshi's published schedule, version 2026-08-04.a, is roundup(0.07 × C × P × (1 − P)), which makes 100 contracts at $0.50 cost $1.75 in fees and the same 100 contracts at $0.05 cost $0.34, or 6.8% of the money at risk. Our own paper log stands at 10 trades, 6 closed, net −$28.30 on $500 deployed as of 4 August 2026.

The referral terms, read at source on help.kalshi.com on 5 August 2026, pay non-withdrawable trading credits that expire 7 days after issuance. Everything below is sorted into what we verified and what we could not.

This review has three sources and none of them is a funded account

That is the whole evidence base. Anything about withdrawal speed, hold triggers, app stability under load or support quality is absent from this page because we have not bought the right to say it.

The taker fee is 7 cents on the dollar of a coin flip, and it gets worse as the contract gets cheaper

The formula on the schedule is:

fee = roundup(0.07 × C × P × (1 − P))

where C is the number of contracts and P is the price in dollars. Round-up is to the next whole cent.

One caveat on sourcing before the numbers, because this site's whole argument is that you should be able to check us. Kalshi publishes that schedule as a rate-limited PDF rather than as crawlable HTML, and no stable URL for it is recorded in our source file, so the version stamp 2026-08-04.a is the only handle we can give you. It is the single claim on this page we cannot hand you a link to.

Work it at three prices, 100 contracts each time:

The second and third rows are the point. In dollars the fee is identical, $0.34, because P × (1 − P) is symmetric around $0.50. But one position ties up $5.00 and the other ties up $95.00, so the same fee is 6.80% of one and 0.36% of the other, a factor of 19.

Divide the formula through by the stake and the reason falls out. Stake is C × P, so:

fee ÷ stake = 0.07 × C × P × (1 − P) ÷ (C × P) = 0.07 × (1 − P)

The contract count cancels. The price does not. Fee as a share of what you put up is 0.07 × (1 − P), which rises the cheaper the contract gets. The longshot corner of the board is the most expensive place on the exchange to take liquidity, which is the opposite of how it looks when you are staring at a 4-cent quote.

Rounding makes small longshot orders worse again

Round-up to the next cent is trivial on a large order and brutal on a small one. One contract at $0.03 owes 0.07 × 1 × 0.03 × 0.97 = $0.0020, which rounds up to $0.01. You paid 3 cents for the contract and 1 cent to the exchange: 33% of stake. One contract at $0.50 owes $0.0175, rounds to $0.02, and pays 4.0% instead of the unrounded 3.5%.

So the fee schedule punishes cheap contracts twice, once through 0.07 × (1 − P) and again through the rounding, and the two stack hardest exactly where a new trader goes first.

Kalshi's own documentation is not much help here. Their HTML fee page is a stub that punts to that rate-limited PDF, so there is no fee table, no per-price grid and no worked example anywhere on the indexable web. On kalshi fees (Volume 800, Traffic Potential 300, KD 9, all Ahrefs), our platform audit records a 404 page holding position 9. Those three figures cannot be re-pulled today: our Ahrefs API units are exhausted until 13 August 2026, so read them as of the audit rather than as of now. A fee calculator is on our build list at /kalshi/fees/ for that reason.

Break-even is 0.07 × (1 − P), so any single edge threshold is wrong somewhere

Because break-even as a fraction of stake is 0.07 × (1 − P), the edge you need before a Kalshi taker order is worth doing depends entirely on where you enter:

This is one leg, entry only. It is also the reason a flat "1% edge" rule is not a rule at all: 1% clears the bar at $0.90 and misses it by more than 5 points at $0.10, on the same number.

We have this wrong on our own site as you read it. Our scanner and methodology pages still publish a flat 1.0% minimum edge threshold, and under the formula above a flat threshold cannot be right at more than one price. The fee-adjusted figures meant to replace it are 3.63% for a Kalshi leg, 2.56% for the Polymarket 0.05 categories and 2.04% for the 0.04 ones, and those are not right either: each is a single number where the honest answer is a curve. Both get replaced with a per-price calculation before the scanner publishes a public row. Our own correction log is at /methodology/.

Adding a Kalshi leg turned our paper book from +$4.73 to −$3.48

Five positions in the paper log were opened on 28 July 2026 with $500 deployed, priced on Polymarket at the verified per-category rates, and then re-costed a second time as a two-leg cross-venue trade with a Kalshi leg added at 0.07.

Worth saying plainly what these are, because a two-leg cross-venue position looks like arbitrage and is not. Our scanner finds momentum, not arbitrage. Intra-market arbitrage is dead on modern Polymarket: Gamma prices sum to exactly $1.00, so there is no gap to take. The second leg here is a cost test, not a free lunch, and the point of the table is what that cost does.

The Kalshi leg cost $8.21 across five positions and flipped a book that was $4.73 up at entry to $3.48 down at entry, before any market moved. Five observations is not a rate and we are not publishing it as one. It is five worked examples, and you can check every line against the formula above.

The bottom row is the clearest case. At an entry of $0.2965, a Kalshi taker leg needs 0.07 × (1 − 0.2965) = 4.92% of stake in edge just to pay for itself. That trade was opened on a claimed 1.20%. It was not a bad outcome, it was negative arithmetic at the moment of entry.

Full log at /trades/.

The referral pays credits that expire in 7 days and are not cash

Read direct from the Referral Program FAQ on help.kalshi.com on 5 August 2026:

• What the referrer gets is not cash. The FAQ's own wording: "Referral credits aren't cash, they're designed for trading on the platform." Only profits you subsequently make trading with them become withdrawable.
• They expire 7 days from issuance unless otherwise specified, or they are forfeited.
• The code window is 72 hours from account creation and before the first deposit, whichever comes first. Deposit before entering a code and the code is dead.
• US only. International users are not eligible.
• A volume gate exists. The referred person has to meet the trading requirement shown in their own Rewards section. Account-specific, visible in-app only.
• A lifetime cap exists. The amount is not published.
• No payout amount and no percentage are published anywhere. The FAQ says reward amounts vary by programme and can change between referrals.

Set that against the other venue for scale, not as a recommendation. Polymarket's referral documentation publishes 10% of net taker fees on direct referrals and 5% indirect, behind a $10,000 lifetime volume gate, ending at 30 days or Platinum tier. Taker fees are the entire base there, because Polymarket never charges makers anything: that is zero, not a discount, so a referred trader who only posts resting orders generates nothing at all. Numbers you can argue with, in either direction. Kalshi publishes none.

Three of those terms, the 7-day expiry, the 72-hour pre-deposit window and the non-withdrawable credit mechanic, are the ones most likely to cost somebody the bonus they signed up for, and they are the three least discussed on the pages that rank for Kalshi promo terms.

There is no Kalshi publisher programme to join, which is worth knowing when you read Kalshi reviews

We tried to read kalshi.com/affiliates four times across two research passes. Every attempt returned HTTP 429. No content from that URL has ever been obtained by this project, so we cannot tell you what it says.

What is documented elsewhere is that Kalshi runs named partnerships with CNN, CNBC (multi-year, exclusive), FOX (News, Business, Weather and One), Madison Square Garden, the Chicago Blackhawks and Men in Blazers. Negotiated brand and data deals. No application form, no rate card, no self-serve tier that we have found. Our source file records that partner list without naming the document it came from, and Kalshi publishes no partner page we can link, so treat the list as reported rather than verified at source.

Two things follow. First, a Google AI Overview we read on 5 August 2026 described Kalshi as running custom promotional partnerships with select media publishers, which reads like something you could apply to and actually describes CNN, CNBC and FOX. Second, our competitor audit found five domains running byte-identical code with per-host-branded Kalshi referral codes. That is not five publishers who signed up. It is consistent with one operator negotiating once and deploying across five media properties, because there was nothing to sign up for.

If you are reading a Kalshi review with a code in it, that is the structure you are reading inside.

How this page is funded

PredictionEdge currently takes no referral revenue, and whether it ever will is undecided. There is no Kalshi code on this page. In this specific case the arithmetic makes it academic anyway: Kalshi's programme is an ordinary user refer-a-friend scheme that pays a US trader in expiring credits, and we are neither a US user nor a real-money trader.

What we could not verify

This is the honest half of the review.

• Whether Kalshi charges makers at all. Kalshi's newsroom says resting orders are fee-exempt. Kalshi's help centre says maker fees are charged. Our source file records the contradiction without a URL for either document, so we are naming it rather than citing it, and every fee figure on this page should be read as a taker figure only. This is a Kalshi question, not a general one: on Polymarket the answer is settled, makers are charged zero.
• Whether anything is charged at settlement. Our reading of version 2026-08-04.a records the taker formula. It does not tell us what happens when a contract settles rather than being sold.
• Special-event fee tiers. Kalshi acknowledges these exist for elections, awards ceremonies and large sporting championships. The numbers are not published anywhere public.
• The sub-penny rounding and whole-cent rebate accumulator on docs.kalshi.com. Documented, but written for engineers rather than traders. We have not worked out what it does to a small trader's bill, and it may soften the rounding effect described above.
• The referral payout amount, the rev-share percentage, the size of the volume gate, and the lifetime cap. All in-app only. The FAQ page also does not state which states are excluded.
• Everything about money movement. Withdrawal timing, what triggers a KYC hold, and support response times are the dominant complaints in this category and we cannot speak to any of them without a funded account.
• Kalshi's regulatory position, federal or by state. Every legal claim on this site links its originating document, we do not yet hold those documents, and the state pages are not written. So this review makes no claim about whether Kalshi is legal where you are.
• Two reputation figures in our research file. One records 174 of 214 complaints going unanswered on a BBB profile (81%), another a Trustpilot rating of 1.9 out of 5. Our file does not name which platform's profile either came from, so we are not attaching them to Kalshi until we re-read the profiles ourselves.
• A quotation attributed to a Massachusetts Attorney General complaint, concerning Kalshi writing its own contract rules and determining settlement without an independent intermediary. It sits in our research file with no case caption and no docket number, we have not read the filing, and so we are not repeating it as a finding.
• Our own trade log does not fully reconcile. The corrected record says 10 trades, 6 closed, net −$28.30 on $500 deployed as of 4 August 2026. A separate note describes 5 open positions against the same $500. The two have not been reconciled. The figure we stand behind is the first one, and this sentence exists so you can discount the rest accordingly.

Questions people actually type

How much does Kalshi charge per trade? The taker fee is roundup(0.07 × C × P × (1 − P)), with C contracts at price P. A 100-contract order at $0.50 pays $1.75 on a $50.00 position, or 3.50% of stake. Whether makers pay anything is contradicted between Kalshi's newsroom and its help centre, so we cannot answer for resting orders.

Why do cheap contracts cost more in fees? Because fee divided by stake reduces to 0.07 × (1 − P), and that rises as the price falls. At $0.95 the fee is 0.35% of your money. At $0.05 it is 6.65%. Rounding up to the next cent makes small longshot orders worse still: one contract at $0.03 pays $0.01, which is 33% of stake.

Is the Kalshi referral bonus real money? No. Kalshi's Referral Program FAQ states referral credits are not cash and only the profits you make trading with them become withdrawable. They expire 7 days from issuance, the programme is US-only, and a lifetime cap exists whose amount Kalshi does not publish.

I already deposited. Can I still enter a referral code? No. The code has to go in within 72 hours of creating the account and before the first deposit, whichever comes first. Both conditions apply, so a deposit made on day one closes the window even though the 72 hours have not run out.

Have you traded on Kalshi with real money? No, and that is the first sentence of this review for a reason. Our record is a paper log: 10 trades, 6 closed, net −$28.30 on $500 deployed as of 4 August 2026. It is published in full at /trades/, losses included, which is the only reason it is worth anything.

Does adding a Kalshi leg to a cross-venue trade pay? In the only five positions we have costed, no. One leg netted +$4.73 at entry; adding a Kalshi leg at 0.07 to each netted −$3.48, a swing of $8.21 on $500 deployed. That is five observations, not a rate, and we publish the arithmetic per trade rather than a percentage.

What is the minimum edge worth trading on Kalshi? It depends entirely on entry price, which is why a flat threshold misleads. Covering one taker leg needs 0.35% at $0.95, 3.50% at $0.50 and 6.65% at $0.05. A trade we opened at $0.2965 on a claimed 1.20% edge needed 4.92% to cover a Kalshi leg and could never have won.

---

Next, if you are still deciding: is Kalshi legit and is Kalshi gambling at the same stage, and Kalshi vs Polymarket once you want the mechanical comparison rather than the bonus comparison.