What unusually small, repetitive executions can show
The original September analysis examined public MEXC executions to look for unusual size distributions and repeated quote amounts. Those patterns are useful screening signals. Public execution data alone, however, cannot establish common account ownership, self-trading, or the intent behind a trade. Calling a pattern proven wash trading requires evidence that this dataset does not contain.
In this guide
- What unusually small, repetitive executions can show
- The original research window and its provenance
- What the published size comparison reported
- Repeated amounts are a screening signal, not proof
- Why a small maximum needs a coverage check
- Price movement does not settle the question
- A stronger investigation checklist
- How to use the downloadable pair tables
- Questions readers often ask
This revised guide preserves the original dated research record, explains its limits, and shows how to investigate the pattern without confusing an unusual distribution with an identified offender.
The original research window and its provenance
The original publication described 8–15 September 2026 and reported 55,737,832 executions across 1,188 MEXC pairs. The figures below are the original published snapshot, not fresh October measurements. The publication did not retain a complete reproducibility manifest identifying every time-boundary and inclusion rule.
During the 11 October review, a bounded live-database check did not reproduce that historical week. The MEXC check timed out under its five-second statement limit; a separate earlier retention check placed the oldest observed live MEXC trade in October. The downloadable pair tables were regenerated on 6 October. Consequently, those current downloads are not evidence that the September totals have been independently rerun.
This distinction matters: an archived published result can be discussed as such, but a refreshed dataset must not be presented as the original source behind it. Future snapshots need a capture time, exact interval boundaries, coverage definition, and immutable version before they can support a repeatable comparison.
What the published size comparison reported
| Pair | Published executions | Published observation |
|---|---|---|
| BTCUSDT | 1,173,870 | Mean size about 2,924 USDT; median about 173 USDT |
| ETHUSDT | 1,060,762 | Mean size about 2,562 USDT; median about 375 USDT |
| PLBUSDT | 803,702 | 99.2% of executions below 5 USDT; largest about 297 USDT |
| CTPUSDT | 656,462 | 99.2% below 5 USDT; 99th percentile 3.87 USDT; largest about 87 USDT |
These reported distributions are markedly different. A median much lower than the mean suggests a long upper tail. A very high share below 5 USDT with a small observed maximum suggests many small fills within the captured sample. Neither result identifies how many accounts or people participated.
Do not use BTC and ETH as universal definitions of legitimate trading. Pair liquidity, price, lot size, minimum notional, market-maker behaviour, fees, promotions, and participant mix can differ. A useful comparison controls those differences rather than assuming every healthy market must look like BTC.
Repeated amounts are a screening signal, not proof
The original CTP table reported 33,022 executions rounded to 1.20 USDT and 26,962 rounded to 1.10 USDT. Repetition at this frequency is worth examining. Before attributing it to manufactured turnover, check how the amounts were rounded and what the exchange’s order-size constraints were.
Two distinct high-precision amounts can become the same two-decimal value. Minimum order values, lot increments, order splitting, and standard algorithmic sizing can also concentrate fills. Histograms need their bin definition: a repeated cent value is not the same evidence as an identical full-precision amount repeated with identical timing.
Public executions also represent fills rather than necessarily complete orders. An algorithm can split a parent order into many pieces, or one order can match several counterparties. Counting fills as distinct investors makes the interpretation stronger than the evidence permits.
Why a small maximum needs a coverage check
A low observed maximum is only the largest execution in the captured dataset. It is not proof that the market never executed a larger trade. Verify the complete interval, feed interruptions, filters, duplicate handling, and exchange message semantics before comparing maxima.
An apparent hard ceiling may reflect an order rule, an execution strategy, a filter, or incomplete capture. A repeatable ceiling over several complete intervals would be more informative than one screenshot. It still would not identify who was on both sides.
Price movement does not settle the question
The original CTP interval reported prices from 0.000708 to 0.000823 USDT, a low-to-high increase of approximately 16.24%. A market can move materially while most captured fills are small. Thin depth, quote changes, external price discovery, and fragmented execution can all contribute.
This combination does not by itself prove that volume was invented. Conversely, substantial reported turnover does not guarantee that an order of comparable size could execute near the last price. Historical turnover and current order-book depth answer different questions.
A stronger investigation checklist
- Define the cohort. Record exact venue, pair, quote currency, UTC start, exclusive end, and whether timestamps refer to exchange execution or receipt.
- Audit coverage. Document missing intervals, filters, and duplicate treatment before calculating distributions.
- Keep precision visible. Compare full-precision and rounded amounts; record histogram bins and market lot/notional constraints.
- Compare like with like. Examine another complete interval for the same pair and, where asset identity is verified, another venue.
- Separate fills from actors. Execution counts and size categories cannot identify account ownership or establish self-matching.
- State the conclusion at the evidence level. “Unusually concentrated small executions” is supportable from a distribution. “Identified wash trader” needs additional ownership and conduct evidence.
The MEXC public-trade specification is the reference for the feed’s fields and trade-side values. It does not provide account identities in the public execution stream.
How to use the downloadable pair tables
Use the refreshed tables as a separate exploratory snapshot. Inspect their generation date and available columns, and do not merge them into the September narrative without checking the interval. A pair ranking can help prioritise manual inspection; it cannot turn a heuristic into an accusation.
For a cross-venue example, read the MEXC and Gate comparison. For the arithmetic behind buy/sell summaries, read the execution-volume guide.
Questions readers often ask
Does 99.2% of fills below 5 USDT prove wash trading?
No. It establishes an execution-size concentration in the stated dataset. Ownership, self-matching, and intent require additional evidence.
Does a repeated 1.20-USDT amount prove one bot produced the trades?
No. Check rounding, constraints, and timing. Several strategies can produce similar sizes, and the public feed does not identify a bot operator.
Is high turnover the same as deep liquidity?
No. Current spread and available depth must be examined separately. Repeated execution can generate substantial turnover without equally substantial resting liquidity.
Are the current downloadable CSVs the September archive?
No. The files were regenerated on 6 October. They should be treated as a different snapshot, not a preserved source for the original week.
What changed from the original interpretation?
Categorical claims that these tests rule out normal activity were removed. The observed patterns remain useful, but the article now separates those observations from conclusions that public fills cannot prove.
