ChainCatcher has published an analysis by Zhou on where arbitrage may exist in on-chain stock perpetuals, using SK Hynix as the clearest recent example and asking whether the same playbook can be applied to CXMT.
The piece starts with a trading recap that has been circulating on X. In that post, trader GodpanSen said he began trading cross-market spreads tied to SK Hynix in June and made more than $600,000 in a little over a month. ChainCatcher says the exact positions and profit figures cannot be verified from public data, but adds that the market structure and the mechanics described in the post do not show obvious flaws at a high level.
At the same time, CXMT is set to list on Shanghai’s STAR Market next Monday, and its on-chain pre-IPO perpetual has already started trading. The article says the deal is one of the biggest IPO names in the market recently after the SK Hynix ADR theme, and investors are now debating whether the SK Hynix approach can be reused for CXMT.
What the SK Hynix trader was really monetizing
According to the report, SK Hynix effectively had at least five price tracks running at the same time: the Korean common stock, the Nasdaq ADR, a Hong Kong-listed leveraged ETF, stock perpetuals on centralized exchanges, and a HIP-3 contract on Hyperliquid. Those instruments reference the same company, but they do not share the same settlement currency, trading hours or pricing mechanics. The trader’s edge, in the article’s telling, came from those mismatches.
The first setup appeared after Binance listed the contract in June. Before that, there had only been one perpetual curve on Hyperliquid. Once Binance joined, there were finally two prices that could be compared directly.
GodpanSen said that on one weekend he found the Binance SK Hynix contract trading $30 above Hyperliquid. He attributed that gap to differences in funding rules across venues and reasoned that if the funding paid before closing the trade was less than $30 per share, the spread was profitable. He opened a 1,000-share position. When the cash market reopened on Monday and the spread narrowed, he said the trade produced a net gain of $15,000 after funding costs.
The second and larger strategy involved buying Korean spot shares and shorting a rich perpetual. The article notes that many crypto traders cannot easily access Korean equities, so positioning on the derivatives side tends to skew long. At extreme points, the perp premium was said to have exceeded $40 per share. GodpanSen used an Interactive Brokers account to buy the Korean stock while shorting the inflated perp, neutralizing directional stock exposure and waiting for the basis to compress. He also collected funding paid by longs to shorts. That leg, according to the post cited by ChainCatcher, generated more than $120,000.
The third method focused on differences in how exchanges charge funding. GodpanSen said prices were often ordered with Binance above OKX and OKX above Hyperliquid. After backtesting, he concluded that, excluding sharp pre-market and after-hours rallies, one week of funding on OKX was nearly 1 percentage point richer than on Binance. He therefore shifted positions to OKX when prices on the two venues were close, saying he earned $170,000 in funding there. He estimated the same position would have brought in only about $100,000 on Binance.
He attributed that gap to OKX’s index design, claiming it did not build its own pricing algorithm and instead weighted Hyperliquid and Binance. ChainCatcher cautions that OKX’s official documentation does not describe the index that way. The documentation only states that the index uses multiple price sources and that some components are smoothed with an EMA.
Some trades went beyond pure basis capture
The article says the fourth trade was effectively a bet that exchange rules would change. During a stretch when SK Hynix stock fell sharply, the Binance contract traded more than $40 above the underlying stock. In the recap, the trader argued that under a normal formula the 8-hour funding rate should have been above 1%, but Binance capped a single funding period at 0.5%, while Hyperliquid settled every hour. That opened a roughly $30 spread between the two venues.
GodpanSen said he believed the situation would not last. He shorted Binance and went long Hyperliquid with a position of nearly $10 million and an average spread of about $25 per share. He said Binance then changed funding to every four hours that same afternoon, the spread narrowed, and he closed in two batches for more than $150,000 in profit.
ChainCatcher says that account does not line up neatly with Binance’s public notices. Binance said in a June 1 announcement that the three perpetuals launched with funding-rate limits of plus or minus 2%. The exchange then narrowed the band to plus or minus 0.50% at 00:15 on July 15 and changed settlement frequency from every eight hours to every four hours at the same time. In other words, the trader’s description of the rate already being locked at 0.5% before the rule change does not match the published notice. The article says what he likely encountered was an effective cap in practice rather than the formal announced limit.
The fifth trade involved a Hong Kong leveraged ETF. GodpanSen said that on a Friday when the Korean market was closed but Hong Kong was open, a 2x long SK Hynix ETF listed in Hong Kong fell more than 20 points in a single day, implying a drop of more than 10% in the stock, while crypto perps tied to SK Hynix were down only 5% over the same period. He bought the discounted ETF and shorted the SK Hynix perpetual on Binance, expecting convergence once Korea reopened on Monday.
He calculated the hedge ratio as 100 ETF shares to 1.19 shares of stock, but in practice used a 1-to-1 hedge and intentionally left about 20% of the position as naked long exposure. He said he bought 480,000 ETF shares and shorted 5,000 shares via the contract. On Monday, the stock fell only 5% and then moved higher, allowing him to unwind the trade for more than $200,000.
The recap also included an FX loss. To simplify the purchase of Korean spot shares through Interactive Brokers, he borrowed KRW. The crypto contracts, however, referenced the U.S. dollar value of the stock. That left him with hidden exposure to the KRW/USD exchange rate. Because the trade was opened near a recent low in KRW and the won later appreciated sharply over the following month, he said the FX leg alone lost $60,000.
Why the setup is harder than it looks
ChainCatcher groups the strategy into three broad categories.
- The first is moving spread exposure between two perpetuals. If the same underlying trades at different levels on different exchanges because of different funding rules, a trader can try to capture that gap as long as cumulative funding paid while holding the position stays below the initial basis. In that case, both legs are derivatives, and what is being hedged is time and venue dislocation more than the asset itself.
- The second is buying spot and shorting the perp to earn funding while staying roughly market-neutral. The article stresses that market-neutral does not mean risk-neutral. Funding rates float. They may stay positive when retail positioning is crowded on the long side, but if short demand becomes dominant and funding turns negative, the short side stops collecting and starts paying.
- The third is exploiting mismatches during closed-market windows. SK Hynix trades across Korean cash equities, Hong Kong leveraged ETFs, a Nasdaq ADR and 24/7 crypto contracts. Those products do not open and close together. For the trade to work, prices have to converge rather than keep moving in the same direction. Leveraged ETFs also suffer from daily rebalance drag, so tracking error grows with holding time. The article says this makes them more suitable for narrow windows such as a single weekend.
The report argues that while the mechanism can look elegant on paper, the operational threshold is high. A trader needs functioning access to Korean equities, FX rails, broker capacity and multiple exchange margin accounts. On top of that, execution depends on close attention to platform-specific details such as funding intervals, funding caps, index construction and the moments when those differences are likely to matter most.
ChainCatcher also notes that the last two trades in the SK Hynix example were not pure spread trades in the strict sense. One effectively relied on a view about whether Binance would alter its funding framework. The other depended on whether the discount in the Hong Kong ETF would normalize on Monday. Both introduced judgment about exchange behavior and short-term price direction.
GodpanSen said only 20% to 30% of capital could be used for arbitrage. The article adds that the amount of capital required to support this style of trading is itself a filter. Some users said that even if they copied the framework, fully capturing the headline profits would be difficult. The piece says that once rule changes and directional judgment are added, the risk profile becomes far less uniform and much more game-like.
Why CXMT is not a straightforward replay
The article then shifts to CXMT, which has become the next obvious focal point for traders searching for similar opportunities.
Hyperliquid has already listed a CXMT-USDC perpetual. As of publication, the contract was trading at $6.3896, equivalent to about RMB 43.26. That was roughly 25% below its peak, yet still about five times the IPO price. Based on post-offering total shares outstanding of 66.881 billion, the implied market capitalization was about RMB 2.94 trillion.
That premium is an order of magnitude above what the article says was seen in the SK Hynix case, but it does not automatically create cleaner arbitrage.
With SK Hynix, there were Korean shares, a Nasdaq ADR, a Hong Kong leveraged ETF and multiple exchange-traded perps. Each of GodpanSen’s methods required at least two price curves. A perp-versus-perp spread needs two derivative markets. Funding capture needs spot and perp. Trading closure mismatches needs an ETF and a perp.
Before listing, CXMT has only one visible curve on Hyperliquid. The article says several on-chain positions have already been built on the short side, but there is no spot leg available to hedge against. Those positions are effectively betting that the price will compress lower after listing, not arbitraging between two quoted markets.
After listing, access to the underlying shares may remain a major constraint. According to the article, the STAR Market’s RMB 500,000 asset threshold, together with QFII quota restrictions, will keep most overseas investors out of the cash equity. ChainCatcher says that barrier is functionally similar to the problem crypto traders faced with Korean spot shares in the SK Hynix setup.
The bigger structural difference lies in price-limit rules. Under Shanghai Stock Exchange rules cited in the article, newly listed STAR Market stocks face no price limits for the first five trading days. From the sixth trading day, limits of plus or minus 20% apply. If the cash stock gets locked at a limit-up or limit-down level, the convergence mechanism can be interrupted outright.
There is also the currency layer. In a long-spot, short-perp structure around CXMT, traders are dealing with a chain from RMB to USD to USDC. The article notes that RMB is not freely convertible, onshore and offshore exchange rates can diverge, and QFII inflows and outflows come with separate quota and remittance restrictions.
As for spread trading between exchanges, ChainCatcher says that part could in theory be copied if more venues list CXMT stock perpetuals after the IPO and a second or third curve appears. The same is true for trading-hour mismatches.
Still, the report says the playbook is now public and widely discussed. That means the information edge is thinner, and any spread that does emerge may be compressed faster.
More fragmented markets create more gaps, and more hidden risk
In the final section, ChainCatcher argues that market fragmentation can make arbitrage look more common while making actual execution harder. Once the same underlying is split across cash equities, depositary receipts, leveraged ETFs and several perpetual venues, differences in index methodology, settlement cycles, funding caps and trading hours naturally create gaps.
But spotting a spread and monetizing it are very different things. Between those two steps sit account access, cross-market rails, margin allocation and risk control. If any of those pieces are missing, the strategy breaks down.
The article says these trades may appear to earn highly certain money, but much of the risk sits outside price. In the early stage of a new market, arbitrage often means betting on the rule set itself. Gaps may come from rough market design rather than from clear pricing mistakes. ChainCatcher places recent discussions around Polymarket arbitrage in the same bucket.
It also says such opportunities have a visible half-life. As mechanisms are refined, liquidity deepens and more participants enter, the available spread tends to shrink. For professional traders, rule differences can create arbitrage. For retail traders, the same differences can become a source of risk.
The report closes by citing remarks made yesterday by New Huo Technology economist Fu Peng. He said capital-market pricing reflects expectations and can move well ahead of current fundamentals. Tight capacity and strong orders in the present do not by themselves imply that stock prices will keep rising, because stocks trade the future.
Fu added that many young traders in Korea can post strong profits one day and large losses the next. In his view, the problem is not corporate operations or supply-demand conditions in the industrial chain, but excessive leverage built up inside the market. ChainCatcher says that warning applies to arbitrage traders as well.