Niubang copy speed and stability: how it is measured and the real figures (updated weekly)
How long NB Copy Engine takes from receiving a master's new trade to the copy account's broker confirming the fill, as p50 and p95, with sample sizes, fills and rejections for the last 7 and 30 days. Figures come from the copy engine's execution records, are updated every Monday and contain no customer identifiers.
Niubang editorial · Updated
How it is measured
For every copied new trade, NB Copy Engine writes an execution record: when it received the master's new-trade push, when it sent the order to the copy account's broker, when the broker confirmed, and whether the order was filled or rejected.
Copy latency = the engine receives the master's new-trade push → the copy account's broker confirms the fill. It does not include the time before the master's broker pushes the trade to the engine (that depends on the master's broker and the engine cannot measure it), so measured from the master's click the real gap is somewhat longer.
Order round trip = the engine sends the order to the copy account's broker → the broker confirms. It mainly reflects the speed of the copy account's broker server.
p50 is the median: half of the orders are faster. p95 means 95% of orders are faster. Latency covers filled new trades only; closing trades are not included.
The sample is every copied new trade on the site, including Niubang's own internal accounts. Only totals are shown here, with no account, customer, master or strategy identifiers.
Latest figures
The figures below are recalculated automatically every Monday. On days when masters were stopped or followers paused there are very few samples; the sample size is shown as is, and small samples swing a lot, so read them together with the 30-day column.
Updated 2026-11-11 (recalculated every Monday, UTC dates)
| Last 7 days | Last 30 days | |
|---|---|---|
| Period | 2026-11-04 – 2026-11-11 | 2026-10-12 – 2026-11-11 |
| Copy orders sent | 71 | 28,971 |
| Filled (fill rate) | 41 (57.7%) | 26,012 (89.8%) |
| Rejected for account reasons | 29 | 2,856 |
| Rejected by broker servers | 0 | 81 |
| Network or program errors | 1 | 22 |
| Copy latency samples | 34 | 22,484 |
| Copy latency p50 | 940 ms | 494 ms |
| Copy latency p95 | 1,645 ms | 1,547 ms |
| Order round trip p50 | 775 ms | 320 ms |
| Order round trip p95 | 1,541 ms | 1,062 ms |
| Brokers / strategies involved | 4 / 3 | 13 / 14 |
Reading stability
Fill rate = fills ÷ copy orders sent. Rejections fall into two groups: account reasons (insufficient margin, position limit reached; the customer needs to act, for example by depositing or lowering the multiplier) and broker-server reasons (rate limits, service unavailable, outright rejection; if these stay high, consider another server). Network or program errors are counted separately.
Copy speed and fill prices are also affected by price movement, spreads and the broker's execution model; see “Cross-broker copy trading: symbol suffixes, lot sizes and latency”.
Copy trading involves risk, and a strategy's past performance does not guarantee future returns.