Author: danny Source: X,@agintender

The trading volume of crypto exchanges may be the biggest lie in the industry. This article uses on-chain data to infer the real monthly trading volumes of the eight major CEXs, and the conclusion reached is counter-intuitive. Could Binance, OKX, and Bitget’s Market Shares Be Undervalued?

The cost of brushing is almost zero - no handling fees for market makers, no on-chain settlement required for API matching, and VIP rebates can even allow you to "earn more the more you trade". The result is that if you open any exchange ranking website, half of the numbers you see may be ghost transactions.

But decentralized exchanges on the chain are different. Each transaction requires real margin, real gas fees, and the entire process is transparent. This gives us a yardstick.

We collected official reserve data and 30-day trading volume data from eight major CEXs including Binance, OKX, Bybit, Bitget, Gate.io, MEXC, KuCoin, HTX and Hyperliquid. We used Hyperliquid’s on-chain data as a “benchmark” to infer how much of each exchange’s monthly trading volume is true.

The conclusion is actually not unexpected: of the $361 billion in trading volume reported by an exchange in the past 30 days, it is estimated that more than half is water; Binance’s real market share is not 39%, but may be close to 46%; the entire industry may have more than $300 billion in monthly trading volume that is fictitious.

30-day average Vol/Reserve ratio ranking

Data period: February-March 2026 (30 days) | Source: Official Proof of Reserves, DefiLlama, CoinGlass, Newhedge of each exchange

Note: This article only uses a sample of these eight exchanges for research. It does not represent the entire market share. The actual conclusion will definitely have errors!

1. Benchmark: Why is Hyperliquid the ruler?

To judge whether the trading volume of CEX is real, you first need a "benchmark". We choose Hyperliquid.

Hyperliquid is currently the largest on-chain perpetual contract DEX, with $4.88 billion in TVL, almost all of which is USDC. The cost of counterfeiting its transaction volume is high: each transaction requires locking USDC margin, which has real capital costs; OI is also at the forefront; Gas and transaction fees are inevitable; the chain is completely transparent, and every transaction can be audited.

After all, they are all exchanges, operating in the same field, with similar trading targets. It stands to reason that user habits and fund usage efficiency will be similar. As a "contract trading-based" platform, Hyperliquid's capital utilization rate should be higher than that of CEX, so Hyperliquid is used as the benchmark.

Our core methodology is to calculate the "average daily trading volume/reserve ratio" for each exchange:

For specific theory, see: https://x.com/agintender/status/1992906704925884898?s=20

Formula = 30-day total trading volume ÷ 30 ÷ core reserve assets (BTC+ETH+USDT+USDC)

Hyperliquid’s total contract trading volume in the past 30 days was $206.8 billion (defiLlama API precise data), with an average daily volume of $6.89 billion, corresponding to $4.88 billion TVL, and an average daily ratio of 1.44x.

This 1.44x is the ceiling of "honest trading". Considering that CEX has more diverse fund uses than pure contract DEX (including spot, financial management, lending, etc.), we give CEX a 15% tolerance range and relax the threshold to 1.66x.

2. Data basis: Each CEX reserve proves calibration

Before inferring the trading volume, the reserve data must be correct - it is the denominator of the formula.

The following analysis all uses official PoR data, and the reserve rate is as follows:

Note: Bitget (BTC 237%), MEXC (BTC 270%) and Gate.io (BTC 147%) all hold BTC far exceeding user deposits. This is because the platform’s own funds make the denominator larger, so our analysis is actually more tolerant to them.

3. Panorama of 30-day trading volume

Data source

• Monthly spot trading volume: Newhedge (March 2026 data)

• Monthly trading volume of derivatives: average of multiple snapshots of CoinGlass × 30 (marked as estimate)

• Hyperliquid: DefiLlama API 30 days data

4. Breakdown of contract trading volume

Contracts are the absolute main force in CEX trading volume, and their authenticity directly determines the credibility of the total volume.

Contract vs spot two-dimensional anomaly detection (bubble chart)

Binance’s contract trading volume is most likely real. 0.36x is well below the on-chain benchmark - every $1 of reserves corresponds to only $0.36 of average daily contract volume. There is a large amount of "silent funds" on the platform, which is a typical manifestation of a healthy ecology.

Bitget and KuCoin are basically normal in the contract dimension. Both have lower contract ratios (1.33x and 1.27x) than the Hyperliquid benchmark.

Gate.io is in a gray area. The contract ratio of 1.75x is 22% above the benchmark. Taking into account the 15% tolerance interval (threshold 1.66x), Gate.io just exceeded the line. However, Gate.io holds 147% of BTC excess reserves, and proprietary trading may partly explain the high contract volume.

MEXC has the strongest signal. The contract ratio of 2.95x is 2.1x higher than the Hyperliquid benchmark. (Later investigation found that it should be caused by MEXC’s trading incentive activities)

5. Independent signals of spot trading volume

30-day spot data (from Newhedge) provides a verification dimension independent of derivatives:

MEXC and KuCoin have much higher spot daily ratios (0.81x and 0.98x) than other exchanges. Gate.io (0.278x) is within reasonable range.

Compare Binance (0.081x) and OKX (0.059x) - most of the funds are in a "precipitated" state, and users save them for long-term holding or financial management. This is exactly what a large, mature exchange should look like.

6. Reverse prediction of real trading volume

For exchanges whose daily average total ratio exceeds the tolerance threshold (1.66x), we will treat the excess as "suspicious trading volume":

Reserve rate heat meter

MEXC’s monthly suspicious transaction volume is approximately $201 billion. This means that MEXC has an average of approximately $6.7 billion in suspicious transactions every day. (Later investigation found that this should be due to MEXC’s trading incentive activities)

KuCoin’s moisture is concentrated in the spot. The contract dimension is normal, but the monthly spot volume of $65 billion is significantly higher than the $2.2 billion reserve.

Gate.io is suspected of having mild moisture. The contract ratio of 1.75x is just above the threshold, and the total ratio of 2.03x is in the "zone of concern". The estimated monthly trading volume of approximately $54 billion (approximately 18%) may be on the high side. However, Gate.io’s BTC excess reserves (147%) and overall reserve rate of 122% show healthy fundamentals – the suspicion of moisture is more likely to come from market maker incentives rather than systemic fraud.

The trading volumes of Binance, OKX, Bybit, HTX, and Bitget are basically credible. All five have contract ratios below the Hyperliquid benchmark.

7. The truth about CEX market share

Among the eight largest CEXs, Binance’s monthly market share is 39.1% ($1,459b / $3,725b). In recent years, there have been many voices saying that Binance is "losing market share" - exchanges such as MEXC and Gate.io are rapidly encroaching.

Note: This percentage is not the percentage of the entire market, but the percentage in the report of this article. This article only samples these eight exchanges, and the actual conclusion will definitely have errors!

But if the suspicious transaction volume is excluded and recalculated:

Market Share - Reported vs Adjusted

Binance’s real monthly market share is nearly 46%. It’s not that Binance is losing share, but that some competitors are using false trading volumes to create the illusion of “catching up.”

The top three companies together account for 76% of the real market. Binance (45.7%) + OKX (17.3%) + Bybit (12.6%) = 75.6%. Crypto trading is far more concentrated than reported data suggests—the head effect is “diluted” by false trading volume.

MEXC's true share may be only 5%. The reported 9.7% was down by nearly half. Gate.io’s adjustment was mild (7.9% → 7.5%), and its performance is relatively healthy among mid-sized exchanges.

8. Industry Impact

Real trading volume vs suspicious trading volume (monthly)

MEXC ~$201 billion + Gate.io ~$54 billion + KuCoin ~$40 billion = $295 billion/month suspicious transaction volume. If hundreds of small and medium-sized exchanges are included, the industry's overall monthly false trading volume may be on the order of $350-450 billion.

For ordinary users: Don’t be misled by “transaction volume ranking”. An exchange with a monthly trading volume of $361 billion may have a real trading volume of only $160 billion. When selecting trades, look at the vol/reserve ratio, not the absolute volume.

For project parties: When selecting a currency listing location, look at the actual order book depth rather than the reported transaction volume. Probably more than half of the monthly reported numbers are non-existent transactions.

For regulators: Hyperliquid benchmarking provides a reproducible, quantitative testing tool. With the accumulation of exchange data on the chain, 30-day and 90-day analysis will make fraud more and more difficult to hide.

9. Methodology and limitations

Hyperliquid benchmarks may be on the high side. All user funds of the pure contract platform are used for transactions, and the capital efficiency is naturally higher than that of CEX. 1.44x may be a high benchmark for CEX, meaning our estimate may be conservative.

Excess reserves blur the boundaries. The self-owned funds of Bitget (BTC 237%), MEXC (BTC 270%) and Gate.io (BTC 147%) push up the ratio but do not equal wash volume.

Derivatives 30d are estimates. CEX's 30-day derivatives trading volume is based on CoinGlass' multiple snapshots and is not precise historical data. Spot 30d from Newhedge Precise monthly data.

Reserve data comes from self-disclosure. If reserves are overstated, our ratios will be low and suspicious transaction volumes will be understated.

Conclusion: four numbers

0.44x – Binance’s 30-day average daily volume/reserves ratio. Almost a "clean slate". When the industry is filled with false prosperity, Binance is the one who can stay in the first place without brushing the volume.

56% - The possible proportion of moisture in MEXC's reported monthly trading volume. Monthly trading volume of $361 billion to reserves of $3.2 billion, a daily average ratio of 3.76x is 2.6 times the benchmark. Approximately $201 billion in monthly trading volume may be fictitious.

46% – Binance’s adjusted real monthly market share. On the surface, it’s only 39%, but after stripping away the water from its competitors, Binance’s true dominance is closer to half. Some of the exchanges that claim to be chasing Binance are just chasing their own shadows.

76% – The adjusted combined real monthly market share of the top three companies. This market is more oligarchic than retail investors think - and fake trading volume has been masking this fact.

Postscript

We initially used DefiLlama’s on-chain tracking data, but after cross-validating the official Proof of Reserves of each exchange, we found large deviations - DefiLlama generally underestimated reserves, with deviations ranging from 5% to 240%.