---
title: "Discussion on \"Measuring Latency in .NET Without Fooling Yourself: Stopwatch, Percentiles and Coordinated Omission\" | Hashnode"
description: "Discussion on \"Measuring Latency in .NET Without Fooling Yourself: Stopwatch, Percentiles and Coordinated Omission\". Why the mean hides your worst latency, and how a test loop hides a one-second freeze"
url: "https://hashnode.com/posts/measuring-latency-in-net-without-fooling-yourself-stopwatch-percentiles-and-coordinated-omission/6abe6f61ad1bec9939b37493"
---

[Feed](https://hashnode.com/feed) /

Discussion

# [Measuring Latency in .NET Without Fooling Yourself: Stopwatch, Percentiles and Coordinated Omission](https://fix-api.hashnode.dev/measuring-latency-in-net-without-fooling-yourself-stopwatch-percentiles-and-coordinated-omission?utm_source=hashnode&utm_medium=feed)

Why the mean hides your worst latency, and how a test loop hides a one-second freeze

Oct 1 · 11 min read · fix-api.hashnode.dev  
Read the article

-   [#csharhp](https://hashnode.com/tag/csharhp)
-   [#dotnet](https://hashnode.com/tag/dotnet)
-   [#fintech](https://hashnode.com/tag/fintech)
-   [#trading](https://hashnode.com/tag/trading)
-   [#tutorial](https://hashnode.com/tag/tutorial)

## Written by

[Sergiy Lutsak](https://hashnode.com/@serglutsak)

@serglutsak · Dnepr

  
[View profile](https://hashnode.com/@serglutsak)

HFT systems developer since 2000. Founder of HFT Software (HFT Forex Copier, HFT Arbitrage Platform). Writing about FIX protocol, latency and low-level C#.

## Responses 2

Markdown supported · ⌘↵ to post

[Arnold Holm](https://hashnode.com/@stratcorealpha)

[Oct 8](https://hashnode.com/posts/measuring-latency-in-net-without-fooling-yourself-stopwatch-percentiles-and-coordinated-omission/6abe6f61ad1bec9939b37493/comment/6ac79db9ee187fa2254072ac)

The fixed-rate test should also keep timed-out and rejected orders in the report. A histogram of completed fills alone can look faster when the slowest requests disappear. Record each intended request and its final status, then show the timeout/rejection counts beside the latency percentiles. That makes the tail visible even when no fill arrives.

[indiainfranotes](https://hashnode.com/@indiainfranotes)

[Oct 1](https://hashnode.com/posts/measuring-latency-in-net-without-fooling-yourself-stopwatch-percentiles-and-coordinated-omission/6abe6f61ad1bec9939b37493/comment/6abe72ebd2ffa21befa28bf6)

Settlement green closes the demo. Audit still wants a tip that survives the vendor UI flip.

Curious which India DPI stack keeps that tip queryable without a second shadow ledger.

marker1001e
