Q2 2026 CEX Trading Data Review: Who's Slacking Off? Who's Inflating Their OI?

Author: danny

In these past few months of the bear market, has Binance been surpassed by OKX? After the implementation of MiCA, has it affected Bybit and Bitget? In the face of regulation, have Hyperliquid's trading volume and TVL been impacted?

About three months ago, we conducted a data analysis of trading volume (Vol/PoR) and open interest (OI/PoR) for several centralized exchanges (CEX) using on-chain data: first for trading volume, and then for open interest. The conclusions back then were somewhat abstract; interested readers can check it out themselves.

Q1 2026 Data: "Leverage Hijacking Plan: When We Examine Exchanges with OI / PoR, Who's Naked?"; "Who's Swimming Naked? An In-Depth Investigation into the Authenticity of Trading Volume and Market Share of Crypto Exchanges."

Some might say that a single snapshot doesn't tell much, or that once incentive activities stop and fees return to normal, the data will naturally revert. What’s really going on? Tracking it once will reveal the truth.

This article mainly discusses three things:

• Re-running Vol/PoR with Q2 data;
• Re-running OI/PoR;
• Comparing Q1 and Q2 to see whose ratios are converging and whose are deteriorating.

1. The Same Ruler, But It Has Improved Against the Trend

The formulas remain unchanged:

• Vol/Reserve = Total trading volume over 30 days ÷ 30 ÷ Core reserve assets (BTC + ETH + USDT + USDC)

• OI/PoR = Total open interest ÷ Core reserve assets

The benchmark is still Hyperliquid, the trading venue with the highest cost of fabrication. This logic has been proven in the previous two articles, so I won’t repeat it here.

However, the performance of Hyperliquid's data this quarter can be described as: growing against the trend.

Hyperliquid's TVL rose from $4.88 billion to $6.11 billion (+25%), and the 30-day contract trading volume increased from $206.8 billion to $237.7 billion (+15%). Funds are flowing in faster than trading volume is increasing, so the benchmark ratio dropped from 1.44x to 1.30x. With a 15% tolerance range, the threshold for Q2 is 1.49x (Q1 was 1.66x).

Some might say that the threshold has dropped, so were those exchanges with abnormal data wronged? Therefore, all determinations in this article have undergone sensitivity testing: using both the new Q2 threshold and the old Q1 threshold of 1.66x.

2. Looking at the Denominator First: Reserves Across the Industry Are Shrinking

The denominator of the ratio is reserves. All data comes from the PoR pages of various exchanges (snapshot from June 2026), converted at BTC $63,001 / ETH $1,769: three observations:

First, reserves are generally shrinking. Binance -17%, OKX -21%, Bybit -29%. There are two forces at play here: one is the price of cryptocurrencies—BTC has dropped to the $63k range, so the asset denominated in PoR naturally shrinks in USD; the second is the actual outflow of funds. The direction is basically consistent: the entire industry's liquidity pool is receding.

Second, Bitget's excess reserves are retreating. In Q1, we mentioned that Bitget held 237% in BTC excess reserves, "We have money, we are not running away." In Q2, this number dropped to 156%. A significant portion of the platform's own BTC has been withdrawn from the reserve pool. This is a minor issue—after all, it’s their own money—but it directly impacts our ratios, which will be discussed later.

Third, MEXC's excess reserves remain unchanged. The BTC reserve ratio is 269%, almost unchanged from Q1's 270%. But note a detail: MEXC users have only deposited 4,699 BTC. An exchange claiming a monthly trading volume of nearly $400 billion has less than 5,000 BTC deposited by all users combined. However, some say that MEXC excels in altcoins, which is also a reasonable explanation.

Data period: June 6, 2026 - July 6, 2026 (30 days) | Source: PoR pages of various exchanges, CoinGlass, CoinGecko, DefiLlama. Note: This article only samples these eight exchanges and does not represent the entire market share; actual conclusions will certainly have discrepancies!

3. Vol/PoR: Rechecking Trading Volume Dimensions

The 30-day trading volume is calculated using CoinGecko's daily series adjusted for the daily BTC price (source change explanation can be found in limitations), with a benchmark of 1.30x and a threshold of 1.49x:

Binance remains the entity that needs no explanation. 0.59x, with every $1 in reserves corresponding to $0.59 in average daily trading volume, and a contract dimension of 0.51x, which is less than half of the benchmark. The ratio is higher than Q1's 0.44x—but this is due to the denominator shrinking by 17%, not an explosion in the numerator. The spot ratio of 0.078x is almost identical to Q1's 0.081x: there still lies a massive amount of silent funds on the platform, with what should be deposited being deposited, and what should be settled being settled. This is the form that a mature exchange should take.

OKX is clean. The contract ratio is 0.93x, and the total ratio is 1.01x—each $1 in reserves corresponds to nearly $1 in average daily trading volume. Moreover, its data shows a 0% deviation in cross-validation, the most solid among the eight. Reserves shrank by 21%, and trading volume also decreased, with both the numerator and denominator receding together, stabilizing the ratio.

HTX is unexpectedly healthy. 1.07x, below the threshold. The only note is that its self-reported trading volume is about 50% higher in cross-validation—but a downward adjustment would only make it healthier, so the determination is unaffected. Its real issue lies not in trading volume, but in OI, which will be discussed in the next section.

Bybit is on the line. 1.42x, just a step away from the threshold. Its reserves shrank by 29% this quarter, the most among the eight—while the numerator (trading volume) showed no abnormal movement, it is the denominator collapsing that has pushed the ratio up. This exchange is now the best observation point for "denominator risk": the trading volume dimension is temporarily innocent, but if reserves continue to flow out without a corresponding decrease in trading volume, it will find itself on the other side next quarter.

KuCoin, has it landed this season? This is the most intriguing set of numbers in the entire article: KuCoin's monthly spot trading volume dropped from $65 billion in Q1 to $32.4 billion, halving; the spot ratio fell from 0.98x to 0.536x, and the contract ratio dropped from 1.27x to 0.98x—now below the on-chain benchmark. The total ratio of 1.52x is only 0.03 above the threshold, and suspicious trading volume has basically dropped to zero. In Q1, we attributed its abnormal spot trading to trading incentive activities. Now that the activities have tapered off, the data has responded accordingly—KuCoin has confirmed our judgment from three months ago with its actions.

Bitget has slid from health into the gray area—but this one must be discussed separately. The contract ratio rose from 1.33x in Q1 to 1.68x, and the total ratio is 1.90x, which looks like deterioration. But don’t rush to convict: its BTC excess reserves dropped from 237% to 156%, with a significant portion of the denominator being withdrawn by the platform itself. Simply put, the deterioration of the ratio can occur in two ways—either the numerator rises, or the denominator is pulled down—Bitget has at least half of its decline coming from the second route. This is not called volume manipulation; this is called withdrawing self-owned funds. Cross-validation shows its trading volume may be about 30% overstated, and if recalculated after downward adjustment, the contract ratio would fall back to around 1.2x, within the average range.

Gate. Q1: Total ratio 2.03x; Q2: Total ratio skyrocketed to 3.25x, with the contract ratio rising from 1.75x to 2.77x. What’s worse: reserves shrank by 19%, but trading volume hardly decreased.

MEXC. In Q1, we attributed its 2.95x contract ratio to trading incentives and zero-fee activities; Q2 data shows: contract ratio 3.73x, close to three times the on-chain benchmark. Interestingly, its spot ratio did indeed drop from 0.81x to 0.447x—the incentives on the spot side seem to have truly been retracted, but all the firepower has shifted to contracts. Therefore, the conclusion can only be revised: this is not an activity; this is a business model. An exchange where users have only deposited 4,699 BTC has generated a monthly contract trading volume of $354.4 billion—behind every user BTC stands $75 million in monthly contract flow.

4. OI/PoR: Rechecking Stock Dimensions

Trading volume is flow, while OI is stock—every open contract requires real money locked as margin. CoinGlass's Q2 quarterly report has not yet been released (the quarter has just ended), so this section uses a real-time snapshot from July 4, 2026, and there will certainly be differences from the Q1 quarterly average.

Benchmark: Hyperliquid OI/TVL = 1.16x ($7.09 billion / $6.11 billion).

MEXC's OI has already caught up with Bybit. $9.69 billion vs $9.64 billion—an exchange with $3.2 billion in reserves and another with $9.9 billion in reserves have almost the same amount of open contracts. OI/PoR 3.06x, the highest in the entire field. In Q1, we said, "You can manipulate volume without spending money, but you cannot create OI out of thin air"—now either MEXC's users are all fully leveraged, or...

Gate. 2.33x, compared to Q1's 2.25x not only has it not converged, but it has also risen. $9.07 billion in OI is pressing on $3.9 billion in core reserves—Q1 we warned that "under extreme market conditions, Gate's liquidation system may face pressure," and this statement remains unchanged, with a stronger tone.

KuCoin and HTX share a common anomaly: their position-to-transaction ratios are absurdly high. KuCoin stands at 3.74, while HTX is at 4.44—this means that the open contracts on these platforms would take three to four days of total transaction volume to turn over completely. In contrast, Binance has a ratio of 0.81, and Hyperliquid is at 2.35 (due to incentives for on-chain orders). A significantly higher position than transaction volume typically indicates two things: zombie positions or artificially inflated open interest (OI) figures. However, KuCoin's OI remains stubbornly at 2.76x—transaction volume is easy to collect, but existing positions are hard to hide.

V. Quarter-on-Quarter Comparison

Let's look at the data from two quarters side by side.

Suspicious transaction volume quarter-on-quarter (exceeding threshold):

Some might argue that the suspicious total increased from $295 billion to $519 billion, suggesting that the lowered threshold could have created a statistical illusion. This is not the case. As mentioned earlier, even using Q1's old threshold of 1.66x, MEXC and Gate's suspicious volumes remain at $240 billion and $180 billion, respectively. The real driving factor is that in a bear market, the denominator (reserves) has decreased, while the numerator (reported transaction volume) has not only remained stable but has subtly increased. Funds are retreating, but reports refuse to reflect this, revealing the discrepancy.

Market share comparison over three periods:

The real share of the top three has risen from 75.6% to 79.3%. Bear markets have always favored concentration—when the tide goes out, funds instinctively flow to the pools least likely to encounter trouble.

VI. RWA Interlude: Is the Heat in US Stocks Reflecting in Data?

This quarter, tokenized US stocks and RWA trading have surged in crypto exchanges. MEXC's homepage carousel features not new memes, but stocks like SKHYNIX, KIOXIA, and SAMSUNG—those chip stocks that have skyrocketed during the storage supercycle; Bybit has launched a TradFi section, showcasing gold, crude oil, and the Nasdaq on the main screen; and Gate has "Stocks" sitting alongside Polymarket in its navigation bar. Retail investors are trading perpetual stocks on crypto exchanges, and all this volume counts as "derivatives trading volume."

This raises a reasonable question: Some might say that your denominator only counts BTC+ETH+USDT+USDC, while their trading volume now mixes in US stocks and gold; could the high ratio be unfairly influenced by RWA? Or could MEXC and Gate's "deterioration" simply be a result of benefiting from the tokenization of US stocks?

We pulled all contract data from both exchanges and separated stocks, indices, precious metals, and crude oil from crypto assets:

Three observations:

First, the story is true, but the scale is not entirely convincing. MEXC is indeed serious about this line—200 RWA contracts, with a full range of storage stocks (Micron at $23 million, SanDisk at $17 million, SK Hynix at $11 million, ADATA, Kioxia) included, and even a SpaceX Pre-IPO contract. However, all of this only accounts for 4.6% of the total derivatives volume. To take the most extreme sensitivity: if we were to write off MEXC's RWA trading volume, the contract ratio would drop from 3.73x to 3.56x—still close to three times the on-chain benchmark.

Second, the true king of RWA trading is gold, not US stocks. In MEXC's RWA volume, gold (XAU+XAUT) accounts for over 40%, with the SPX500 index second. The so-called "tokenized US stock trading boom" currently resembles more of a "precious metals safe-haven trading boom."

Third, while the denominator should be noted, it is still quite small today. Perpetual stocks use USDT as margin, so their reserve consumption is similar to ordinary contracts, making the formula fair for them. The real gap in metrics lies in the spot market: tokenized stocks held by users (like xStocks) are not included in the core reserves of BTC+ETH+USDT+USDC. For example, in the case of Gate, its total reserves amount to $8.18 billion, with core reserves at $3.88 billion; the difference contains GT, altcoins, and tokenized stocks. Even if we expand Gate's denominator to total reserves—the most lenient calculation—the total ratio still stands at 1.55x.

Conclusion: Three Numbers

3.73x ------ MEXC's Q2 contract trading volume/reserve ratio, compared to 2.95x in Q1. Spot incentives have indeed retreated, with all firepower shifting to contracts—this is not just activity; it's a business model.

3.25x ------ Gate's Q2 total ratio, compared to 2.03x in Q1. Reserves shrank by 19%, while reported transaction volume remained unchanged, and OI is still more than double the reserves.

79.3% ------ The adjusted combined real market share of the top three (Binance + OKX + Bybit), compared to 75.6% in Q1. In a quarter where the tide recedes, funds are voting with their feet, accelerating oligopolization. Exchanges still trying to catch up should consider one thing: the way to catch up is to deepen the pool, not just to lengthen the transaction volume.

Appendix: Methodology and Limitations

Data source switch. Q1 spot data used Newhedge, while derivatives used CoinGlass snapshot estimates; Q2 standardized to CoinGecko's 30-day daily transaction series (converted to USD based on the daily BTC price). Daily series are cleaner than "single-day snapshot × 30," but absolute values for quarter-on-quarter comparisons should be discounted. All derivatives data have undergone normalization cross-validation with CoinGlass: six exchanges showed deviations within 12%, with Bitget (+28%) and HTX (+53%) being higher; the text has been processed for sensitivity and the determination strength adjusted accordingly.

OI is a real-time snapshot. CoinGlass's Q2 quarterly report has not been released; this article uses the OI snapshot from 2026-07-04, which is not a quarterly average and differs from Q1 metrics.

PoR conversion prices are standardized. Snapshot dates range from June 1 to June 30; this article uniformly converts based on the latest prices (BTC $63,001 / ETH $1,769), and price fluctuations between snapshot days may introduce errors.

Disclosure of metric differences. HTX combines stablecoins into the "USDs" metric; Gate's USDC is hidden on the second page of the asset list ($117 million, already counted).

Benchmark drift. Hyperliquid's benchmark dropped from 1.44x to 1.30x; all outlier determinations have undergone sensitivity testing using Q1's old threshold. The 30-day benchmark data is based on July 4, differing by two days from the transaction volume window, which is a conservative treatment (if synchronized, the benchmark would be lower and the determination stricter).

RWA/TradFi classification is based on symbol rules. Stocks, indices, precious metals, and crude oil contracts are separated from CoinGecko's contract data based on symbol patterns (STOCK suffix, XAU/SPX/NAS, etc.) for the single-day snapshot on 2026-07-04, and the proportion may fluctuate with market conditions.

Reserve data comes from self-disclosure. If reserves are overstated, the ratio will be lower, and suspicious transaction volume will be underestimated—estimates are conservative.