Compare execution behaviour before comparing venues
The original MEXC–Gate study compared a week of public executions across the two venues. It found large differences in execution sizes, activity, and aggressive-side proportions. Those differences can help you ask better questions about a pair’s trading conditions. They do not establish that one venue is universally better, that two ticker symbols necessarily represent the same asset, or that a historical price difference was executable arbitrage.
In this guide
- Compare execution behaviour before comparing venues
- Scope and historical-data limits
- Example: the reported CTP execution-size contrast
- Whole-venue totals are affected by market coverage
- Count imbalance is not volume imbalance
- Weekly VWAP gaps are not simultaneous prices
- Large volume ratios need precision and identity checks
- A useful seven-metric comparison
- Questions readers often ask
Scope and historical-data limits
The original publication described 8–15 September 2026, with 55,579,298 MEXC executions, 17,180,706 Gate executions, and 270 shared pairs. The sum is 72,760,004 executions. Rounded to one decimal million, that is 72.8 million, not the original article’s 72.9 million.
These are the original published historical totals, not freshly re-created October observations. A companion MEXC article reported 55,737,832 executions and a different CTP count for what it described as the same week. Without a retained cohort manifest, the difference cannot be confidently attributed to a particular filter or boundary. The two results should not be silently treated as one identical dataset.
On 11 October, bounded live checks did not reproduce the September cohort. Gate’s retained CTP sample began in October and contained zero executions in the September interval checked. The MEXC check encountered a statement timeout. Both downloadable research packs were regenerated on 6 October, so the current files are separate snapshots rather than immutable September evidence.
The tables below preserve the original published comparisons with that limitation. They support a discussion of methodology, not a claim that the old raw dataset has now been independently validated.
Example: the reported CTP execution-size contrast
| Metric | MEXC published snapshot | Gate published snapshot |
|---|---|---|
| Execution count | 657,199 | 31,622 |
| Share below 5 USDT | 99.2% | 2.5% |
| Median execution value | 1.05 USDT | 10.30 USDT |
| Largest execution | About 87 USDT | About 1,799 USDT |
This is a large distributional contrast in the published sample. It does not prove that MEXC executions were fake or that Gate participants were distinct human investors. Minimum values, lot increments, incentives, order splitting, and coverage must be checked. An execution count is not an account count.
A cross-venue comparison is useful only after verifying that both markets refer to the same underlying asset and comparable instrument. The same ticker can be reused. Contract, network, asset description, and deposit/withdrawal mapping are stronger identity evidence than a matching symbol.
Whole-venue totals are affected by market coverage
The original report included 1,188 MEXC pairs and 359 Gate pairs. Summing each venue’s full captured universe compares different market sets. A higher total count may reflect more pairs rather than more executions in the same assets.
The shared-pair cohort is a better starting point for a paired comparison, but it still needs matched time boundaries and complete coverage on both feeds. Uniform symbol coverage does not establish equal uptime, identical filtering, or identical duplicate handling.
Keep these questions separate:
- What happened across everything captured from each venue?
- What happened on the same verified assets during the same interval?
- What could a specific order actually execute against now?
The first two are historical dataset questions. The third requires current quotes, available size, and costs.
Count imbalance is not volume imbalance
The original BTX example reported buy-labelled execution shares of 50.2% on MEXC and 17.3% on Gate. The difference is 32.9 percentage points. It describes a difference in classified execution counts if that is the table’s denominator.
It does not establish that 50.2% of quote turnover was buying, that holders on one venue accumulated while holders on the other sold, or that the same traders behaved differently. Many small buy fills can coexist with fewer larger sells.
Check the side definition in the MEXC trade documentation and Gate public trade specification. Similar-looking field names need their own documented meaning before they are compared. A roughly balanced BTC summary does not by itself validate every pair’s side mapping.
Weekly VWAP gaps are not simultaneous prices
A venue’s weekly VWAP weights execution prices by that venue’s traded quantity. If activity concentrates at different times, two weekly VWAPs can differ even when contemporaneous quotes are close. A weekly difference is therefore not an immediately executable spread.
The original AIN examples did not support the accompanying claim that daily prices differed by roughly 0.1% throughout the week. Using MEXC as the denominator, 0.1026 versus 0.1059 is approximately 3.22%, while 0.1412 versus 0.1429 is approximately 1.20%. State the formula and direction; percentage-point language is inappropriate for these raw price levels.
To investigate a potential executable difference, compare timestamp-aligned bids and asks for the same asset and usable size. Then account for fees, transfer conditions, balances, and settlement timing. The live spread tool provides separate current-market context; it does not retrospectively make a weekly average tradable.
Large volume ratios need precision and identity checks
A venue can report much more turnover for a shared pair. Record the exact underlying values before presenting a precise ratio. Dividing rounded display values may not reproduce a ratio calculated from unrounded data, so distinguish exact ratios from approximate examples.
For a pair with a very small denominator, the multiple can become huge without implying deep liquidity on the numerator venue. Examine absolute turnover and current depth as well as the ratio. One exceptional pair should not become a general claim about all markets on either exchange.
The original WXT comparison showed widely separated prices, but stable prices alone do not establish whether two markets represent the same token. Identity requires independent asset metadata. If identity remains unresolved, exclude the pair from actionable cross-venue spread conclusions.
A useful seven-metric comparison
For an appropriately matched cohort, record execution count, total quote turnover, median execution value, a high percentile, small-execution share, aggressive-buy share with its denominator, and VWAP. That is seven metrics, correcting the original section’s count of six.
Add the interval, timezone, timestamp basis, pair-identity check, missing-data notes, and precision used. Preserve the source snapshot rather than overwriting it with the next generated CSV. The resulting table remains an evidence summary, not a verdict about account ownership or misconduct.
Questions readers often ask
Does the same ticker prove the assets match?
No. Verify the asset, contract and network where applicable, and the exchange’s market description.
Can a weekly VWAP gap be traded immediately?
No. It is a historical weighted-average comparison. Current executable bid/ask prices and size are separate inputs.
Does more small trading prove manipulation?
No. It warrants investigation of constraints, precision, coverage, and behaviour. Public executions do not reveal common account ownership.
Why do the two September articles have different MEXC counts?
The retained published record does not establish the reason. Do not silently choose one explanation or claim a rerun resolved it. Exact cohort metadata is required.
What do the downloadable tables represent now?
They are refreshed research tables with their own generation date. They are not a preserved copy of the September raw execution cohort.
