# Performance methodology

> Copilink publishes transparent, product-specific internal processing benchmarks for Standalone and NinjaTrader 8, with the environment, sample, metric boundaries, and results documented separately.

Last verified: 2026-09-14
Canonical page: https://copilink.com/research/methodology
Product: both
Markdown of: https://copilink.com/research/methodology


## Why this methodology is useful

- Product-specific results for Copilink Standalone and Copilink for NinjaTrader 8
- Clearly defined measurement points, sample sizes, environments, and dates
- Copilink processing reported separately from broker and network time

## Copilink Standalone

0.48 ms median from the leader's fill reaching Copilink to the first follower order being sent, and 1.76 ms median until every follower's order is sent, measured on more than 1,000 copied fills on users' own computers. Broker, network and exchange time are not included.

Timing starts when the leader's fill reaches Copilink and ends when the follower order is sent to the broker, measured on more than 1,000 copied fills on users' own computers, on Tradovate and Rithmic, Windows and macOS.

## Copilink for NinjaTrader 8

~1.6 ms average internal processing latency inside NT8 across about 70 live and simulated trades with five concurrent followers on an AMD Ryzen 7 2700X and 16GB DDR4.

Measured from leader `OnOrderUpdate` / `OnExecutionUpdate` to submitting corresponding orders to five follower accounts inside NT8.

## Measurement scope

The published figures measure Copilink's internal processing. Broker acceptance, network transit, exchange matching, and market fills are separate execution stages, which keeps the benchmark specific and reproducible.

## Sources

- https://copilink.com/research/methodology
