Beni vs Kalshi
Beni thinks Kalshi's crypto numbers look cooked. Kalshi employee IcoBeast says he's mixing up two products. The replies got heated, so we went through the posts and filings to see what holds up.
Catch me up ↓About 90 seconds. You don't need to know how trading works.
Added Sep 21 The $1 parlay argument ↗01 / Missed the thread?
Start with these five cards. If something sounds off, the receipts are further down.
Kalshi lets people trade on real-world outcomes, such as whether Bitcoin will be above a certain price at a set time. It also has crypto perpetual futures, usually called perps. Those let traders take positions on prices with leverage, meaning their exposure can be bigger than the cash they put in. [25]
On September 19, Kalshi employee IcoBeast posts a chart: Kalshi at 96.7%, Polymarket at 3.3%. Beni says the numbers include wash trading, where trades look like real activity without a genuine change in market risk. IcoBeast asks why someone would bother when trading fees cost money. [1][2][3]
Beni points to fee rebates and trading volume far bigger than the value of positions still open. He also questions the small positions on the public leaderboard and the incentives for market makers, firms that offer to buy and sell. Later, he adds tables where the same ETH trade sizes appear over and over. [5][22]
IcoBeast's answer is that the chart measures prediction contracts, but Beni's rebates and turnover numbers are about perps. He also backs the way volume is counted and points out that the incentive programs are public. [16]
The incentives are there in the filings, and Beni's screenshots show the numbers he's questioning. We still don't know who controlled those trades or whether they were fake. His promised follow-up is still pending. [22]
A quick detour before the numbers
Volume counts trading over a period. Open interest counts positions still open. The same amount of exposure can change hands over and over, so those two numbers can be very different.
[26][25]Imagine different traders passing these positions around. Move the slider and watch the volume climb.
There's nothing fake about that example. Wash trading means making activity look real without genuine market risk. To know what happened at Kalshi, we'd need to look at the trades themselves. [27]
02 / Who's involved?
Beni is accusing Kalshi of faking volume. He writes the thread and the follow-up analyses; his bio describes him as a former quant and startup cofounder.
IcoBeast is the main person arguing back. He says he works at Kalshi and built its crypto-markets program, and his profile says his views are his own. [4]
Kalshi is the exchange whose numbers started all this. Its filings and help pages let us check parts of the argument. IcoBeast's personal replies don't amount to an official statement from the company.
Artemis is the data provider named on the chart that started the argument. The screenshot compares Kalshi with Polymarket, not with every crypto exchange. [1][29]
Jump Trading shows up because of a reported deal to supply liquidity. That still doesn't tell us whether Jump was on either side of the trades Beni is questioning. [9]
@j0hnwang gets tagged when Beni promises a follow-up on 15-minute and hourly contracts. The announcement is in the timeline. We couldn't find Wang's specific remarks in the sources we checked. [18]
03 / How the replies went
The posts, in order. Pick a day if you're catching up on just that bit. All times are UTC.
We've checked the posts. Whether their accusations hold up is a separate question.
IcoBeast posts the 96.7% Kalshi versus 3.3% Polymarket chart. The tooltip is dated September 18. This is the post Beni replies to.
Read the original ↗Beni says three-quarters of the volume is wash traded. IcoBeast asks why anyone would pay trading fees to fake it. When Beni questions his Kalshi promotion, IcoBeast points out that he works there. [3][4]
Read the original ↗Beni lays out his case: the rebate program, ETH volume compared with open interest, the leaderboard, Jump, and the way contracts are counted. He puts the source links in the final post.
Read the original ↗Beni mocks IcoBeast for not answering his points. IcoBeast's detailed reply arrives later that day, after this post.
Read the original ↗IcoBeast says Beni is mixing up prediction markets and perps. He pushes back on the description of self-clearing members, explains that the incentive programs are public, and defends how volume is reported.
Read the original ↗Beni now says 99% of perp volume is fake, although that percentage hasn't been verified. He promises a thread on 15-minute and hourly prediction contracts and tags @j0hnwang. [18]
Read the original ↗Beni says he's received nonpublic information and needs a lawyer before he can share it. He pushes the thread back by 24 to 48 hours. We haven't seen the private information he's referring to.
Read the original ↗Beni says Bloomberg and other outlets have contacted him. Even if that's right, it doesn't mean they've published an investigation confirming his accusations.
Read the original ↗Beni asks Coffeezilla to help go through his data. There's an invitation at this point, but no evidence that Coffeezilla has started investigating.
Read the original ↗@retardmode claims $136M actually traded against $1.91B of reported volume. A chart accompanies the post, but no linked dataset lets us reproduce those daily totals.
Read the original ↗Car quotes the same figures, adds the 7.1% framing, and projects them across a month. The counting question is worth unpacking; repeating the totals doesn't independently confirm them.
Read Car's post ↗Follow the $1 example ↓Beni posts an image of TestFlight beside an app labelled Kalshi Demo. It's a suggestive screenshot, but it doesn't show that he's accessed internal systems or private data.
Read the original ↗Beni posts tables of repeated ETH trade sizes and says he's backed up his data. He also passes on an anonymous tip about market-maker volume deals, calling it a rumor himself.
Read the original ↗Daniel Sapkota says a market on Mamdani becoming the 2028 nominee had $1 transactions every two seconds, adding up to about $2M between August 1 and September 19. Beni shares the post. We haven't reproduced Daniel's calculation, so this is a separate allegation to check. [36]
Read the original ↗Beni says the conduct could amount to fraud or manipulation if Kalshi took part, directed it, or knowingly incentivized it. That is his conditional legal argument; the post doesn't report a legal finding.
Read the original ↗Beni posts a screenshot that appears to show Coffeezilla following him back and sending a brief greeting. It looks like contact, but there's no announced investigation or endorsement in it.
Read the original ↗Beni joins in with a joke about other exchanges. It was his latest visible post when we checked at 04:20 UTC, and it doesn't add evidence about the trades.
Read the original ↗IcoBeast had already pushed back on anonymous fake-volume accusations on September 15. That was before his direct exchange with Beni. We can't tell from the post whether he had Beni in mind.
[33]04 / Bring up the receipts
Pick the part you're curious about. We've put Beni's argument next to the source and our read of it.
Beni says the fee objection misses a rebate: eligible makers receive 0.3 basis points while takers pay 0.3. Add those together and, he argues, the two sides could trade with zero combined exchange fees. [6]
Kalshi's September 2 filing with the CFTC sets out those rates for eligible self-clearing members trading crypto perps. They're in the document's final terms. The CFTC is the US derivatives regulator, but this is Kalshi's filing, not a regulatory verdict on the trading. And 'self-clearing member' is a membership category, not another name for a market maker Kalshi handpicked. [23]
This example uses $1 million of eligible notional, meaning the exposure the trades represent. 0.3 basis points = 0.003%. We're adding the two sides' exchange fees. That doesn't tell us whether the same person controlled both sides, or whether there were other trading costs.
Kalshi has to announce when the September program starts. The earliest allowed start is September 16 at 17:00 ET. It runs no later than December 31 unless changed or ended. Those are the published terms; they don't tell us what any particular trader received.
Beni also links a separate August 4 volume incentive filing. It allows rewards on selected markets, including futures, with eligibility and reward terms set for each market. Several groups are excluded, including members with market-maker agreements. This filing alone doesn't tell us whether rewards were running on a particular crypto prediction market. [24]
The older filing's date is unclear in the tweet. The document itself says August 4, 2026.
Before this dispute, CFTC staff warned that some volume rewards can encourage wash trading. They also said well-designed incentives can improve liquidity. Their August 12 advisory called for clear terms and monitoring aimed at these risks. It didn't accuse Kalshi or assess Beni's evidence. Read the advisory ↗
Beni's ETH-perp screenshot shows both figures, and the division checks out. But it compares a whole day's trading with the positions still open at one moment. [7]
That doesn't mean each position traded 174 times, or that traders typically held positions for eight minutes. Positions can open, close, and change throughout the day. [26]
Beni points to the leaderboard: how does that much volume fit with a top displayed position of just $17,598? His screenshot also says "Join leaderboard". We don't know how complete the list is, who opted in, or whether it includes institutions. We can't treat it as a list of every position on the exchange. [8]
In Part 1.5, Beni says repeated $5,500 ETH trades make up roughly half the activity. These are the figures in his attached tables. [22]
The tables put different contract counts under the $5,500 label as ETH's price changes. They also show a high 'loop%', but the post doesn't include the full calculation or the trade file behind it.
The second image includes BTC too. Its most common trade sizes are different: the repeated $5,500 claim is about ETH.
Kalshi counts prediction-market volume in contracts traded. Beni objects to putting a dollar sign on that count; IcoBeast says Polymarket uses the same notional convention. Parlays bundle several picks into one contract. Their tiny prices can make the gap between payment and payout look huge. [25][10][16]
Buy 1,000 YES contracts at 0.1¢ each. You pay $1. Each can pay up to $1, so the maximum payout is $1,000. That's also $1,000 of payout notional. Kalshi's rules allow sub-cent combo prices. [43][46]
The $999 assumes a newly opened, fully collateralized pair without offsets. Existing positions can change hands, and eligible hedges can reduce collateral requirements. Reported volume doesn't mean that much fresh cash arrived with every trade. [44][45]
On September 20, @retardmode posts a $136M versus $1.91B comparison. Car quotes it 81 minutes later and calls the smaller figure just 7.1% of reported parlay volume. Neither post links the data behind those daily totals. [41][42]
Claimed; not independently reproduced. The ratio is 7.12%. We haven't confirmed what the smaller figure counts, including which side's payments it uses.
Car's roughly $4B versus $57B monthly comparison is the daily pair multiplied by 30. It's a projection, not a measured month. The attached chart also puts 48% of contract volume in parlays with 11 or more legs. That isn't 48% of tickets, and the image gives no date range or source. [41][42]
Gambling Insider reported $17.83B of August combo notional versus $1.09B using one side's price. The Event Horizon, citing TickerTracker, put combos at 61% of volume over September 12 and 13, with 47% of combo volume in combinations with 11 or more legs. These are different periods. They give context, but don't verify the viral daily totals. [47][48]
Imagine buying 100 YES contracts. Change the price and you change what you pay, but you've still traded 100 contracts. They can settle for at most $100 in total; at the starting price of 30 cents, you pay $30.
Bloomberg reported that Jump would receive equity for supplying liquidity to Kalshi and Polymarket. [28]
Beni argues that this relationship gives a liquidity provider a reason to want strong growth numbers. That suggests a possible incentive, but it doesn't connect Jump to a suspicious trade. [9]
In Part 1.5, Beni says an unnamed source told him about volume deals with market makers. He doesn't share a contract or name the source in that post. He calls it a rumor. [22]
05 / IcoBeast replies
After the early insult, IcoBeast posted a more detailed response on September 20. It's worth reading alongside the accusations.
Read his full response ↗He says the original chart covers prediction markets, while the rebate and turnover arguments are about perps.
He says crypto prediction markets don't receive rebates. The separate filings don't, on their own, tell us which incentives were running in any particular market.
He says exchanges commonly offer incentives. He also defends fair access to self-clearing membership and the use of contract counts to report notional volume.
06 / Your take
You've seen the argument. Which take holds up?
We checked the public posts and confirmed that the incentive programs exist. The figures and tables discussed here are visible in Beni's screenshots. Kalshi's glossary explains the different products and what it means by volume.
We don't know who controlled the trades, whether risk really changed hands, which incentives were paid, or whether anyone coordinated artificial activity.
For the trading words
A contract you can buy or sell on the answer to a question. In a binary market, a winning contract generally pays $1 at settlement and a losing one pays $0.
Short for perpetual futures. It's a derivative that lets you take a position on a price without a fixed expiry date. It's a different product from a yes/no prediction contract.
How much trading happens over a period. Check the units: a figure might count contracts, cash paid, or the exposure the trades represent (notional).
The positions still open at a particular moment. It doesn't count all the trading that's happened along the way.
A maker puts an order on the market for someone else to trade against. A taker trades against that order. A rebate returns or credits some trading fees.
Activity that looks like real trading but doesn't genuinely change risk or market position. Repeated trade sizes alone don't prove it.
These explanations draw on the original documentation linked here. [25][26][27]
07 / Check for yourself
All 48 of them: the posts, the filings, the definitions and the news coverage. The original ten-post thread is in here too.
19 Sep · 17:09 UTC · Artemis screenshot: two-platform crypto-volume comparison.
19 Sep · 19:37 UTC
19 Sep · 20:18 UTC
19 Sep · 20:40 UTC
19 Sep · 23:50 to 23:51 UTC
19 Sep · 23:50 to 23:51 UTC
19 Sep · 23:50 to 23:51 UTC
19 Sep · 23:50 to 23:51 UTC
19 Sep · 23:50 to 23:51 UTC
19 Sep · 23:50 to 23:51 UTC
19 Sep · 23:50 to 23:51 UTC
19 Sep · 23:50 to 23:51 UTC
19 Sep · 23:50 to 23:51 UTC
19 Sep · 23:50 to 23:51 UTC
20 Sep · 04:04 UTC
20 Sep · 15:31 UTC · His own account, rather than a statement from Kalshi.
20 Sep · 15:35 UTC
20 Sep · 16:19 UTC
20 Sep · 18:57 UTC
20 Sep · 19:32 UTC · Beni says Bloomberg contacted him. That doesn't mean Bloomberg confirmed his claims.
20 Sep · 23:00 UTC · Image shows TestFlight and Kalshi Demo icons.
21 Sep · 01:18 UTC · Beni's analysis and screenshots. We haven't reproduced the results from raw data.
2 Sep 2026 · Appendix B, PDF pp. 7 to 9: scope, fees, exclusions, and monitoring.
4 Aug 2026 · A separate program. Kalshi has to announce which markets qualify and what the rewards are.
Updated 4 Sep 2026
Accessed 21 Sep 2026
Accessed 21 Sep 2026
9 Feb 2026 · Background on the deal. It doesn't verify Beni's accusations.
Accessed 21 Sep 2026 · We haven't independently recalculated the chart's figures.
21 Sep · 02:35 UTC · Covers Beni's allegation without independently auditing the trades.
21 Sep 2026 · Republishes Odaily's report. It isn't a separate finding.
21 Sep 2026
15 Sep · 18:14 UTC · We don't know whether he was referring to Beni.
20 Sep · 19:47 UTC
21 Sep · 01:55 UTC · Describes repeated $1 activity. We haven't reproduced the calculation.
21 Sep · 02:01 UTC
21 Sep · 02:54 UTC · His argument about what could be illegal. No court or regulator's finding is reported here.
21 Sep · 03:34 UTC · Appears to show contact. We haven't found an announced investigation.
21 Sep · 03:46 UTC · Beni responds to someone else's joke. It adds no evidence about the trades.
12 Aug 2026 · Advice published before this dispute. It doesn't assess Kalshi.
20 Sep · 20:00 UTC · Daily totals and chart provenance not independently reproduced.
20 Sep · 21:21 UTC · Quotes source 41; the monthly numbers are a projection.
Product documentation · Up to $1 per contract; a combo has its own order book.
Product explanation · The two sides fund the payout in a newly opened pair.
Product documentation · Eligible hedged positions can reduce collateral requirements.
Technical documentation · Valid prices depend on the market.
Updated 9 Sep · $17.83B notional versus $1.09B using one side’s price. Different period from the viral posts.
15 Sep · TickerTracker-based analysis of the 12 and 13 September weekend.
Research cutoff: 21 September 2026, 04:20 UTC. The main chronology was checked through that time. On September 21 we added a focused review of the parlay posts, earlier reporting and contract rules. That addition isn't a fresh check of every developing thread.
We pieced together the original ten-post thread and the replies leading up to it, then read the later posts and IcoBeast's detailed response. We inspected the attached images and checked the cited regulatory filings and product definitions. The links go to the original X posts. We retrieved post text and timestamps through a public mirror, and used Thread Reader and live X pages to find posts and cross-check them.
The labels tell you what we checked. When a label says we've checked a post or document, it doesn't mean we've verified every claim in it. We checked the numbers visible in screenshots, but didn't reproduce them from exchange trade records. We didn't have private account data, the alleged nonpublic material, or access to the anonymous source.
Odaily, AiCoin, and outlets republishing their summaries show how the allegation spread. They don't count as separate confirmations of wash trading. When we mention journalist contact, legal advice, or data backups, we're reporting what Beni said.
In the material we reviewed, we didn't find a formal Kalshi statement resolving this episode, a relevant regulatory finding, or independently audited proof of wash trading. That tells you what we found; it doesn't prove that no such material exists. We also couldn't verify that the promised full follow-up on short-duration contracts had been published by the cutoff.
This site is independent of Beni, Kalshi, Polymarket, Artemis, and Jump. We label our own analysis. The screenshots are the work of their original publishers.