About

Why this exists

I'm Roger. I've run all my life, I coach myself, and I train on hilly ground in the Cotswolds, where most pace tools stop making sense. I wanted to know what pace each session should be, and whether I'd hit it. I've never been satisfied with the answers on offer.

None of the mainstream tools take pacing seriously. Strava gives you heart rate zones but no guidance on how to set them. Garmin does the opposite: it sets your thresholds automatically, and they move without explanation, never say how sure they are and often just look wrong. At the other end there's hardware. I've tried a Tymewear strap, and it helped a little, but for me it wasn't worth the money. You can test your own blood lactate, but it's fiddly and easy to get wrong. Or you can book a lab test, which for me means a 50-mile drive and £300.

What Nerd Athletic is going to be

Most running apps are built around your activities. They log what you did, add it up and show it back to you. Nerd Athletic is built around you. It treats the runs you already do, from any watch that exports a FIT file, as evidence about your physiology. From them it estimates:

  • your critical speed (CS)
  • your first lactate threshold (LT1)
  • your durability: how much you fade over long efforts

Under the hood are statistical models built on current exercise physiology research.

Your training paces come from that model. Sessions are set as a percentage over or under CS, LT1 or race pace, not as zones. Every number comes with a range and an explanation of where it came from, and you can leave out any run you don't trust. When there isn't enough recent evidence, it says so rather than guessing. The model should get sharper the longer you use it.

I've been testing an early version on my own training, and the results are promising. While I build the rest, I'll write about what I'm learning here and release free tools to help runners like me get more out of their training.

Building in public

I'm building this in the open, and I'd like you to come along. I'll write up what I learn and what works. I'll also write up what doesn't, including the times the model turns out to be wrong.

First comes the physiology, starting with a series on critical speed. It covers what CS is, how to calculate it, what happens in the body above and below it, how to train with it and where the model breaks down. Then I'll write about how I'm building the model of the runner:

  • how messy watch data becomes usable evidence
  • how to allow for hills
  • how to put an honest range on an estimate

I'm testing it on eight years of my own training data first, then on a few friends. Then I'll publish the results, including how well its predictions match real races and training sessions.

What you'll find here

  • Physiology: energy systems, CS, LT1 and what the research actually says, with plenty of paper reviews. I'm not a doctor or a sports scientist, but I think there's value in a technical person working through this carefully and showing the working.
  • Modelling, and a bit of code: FIT files, the maths behind the tools, and the frustrations and breakthroughs along the way.
  • Training: turning the physiology into sessions. That means sessions set from CS and LT1, what 80/20 training really means, and getting past the headlines to the evidence.

Who it's for

It's for self-coached runners who want to get faster and are happy to dig into the detail. They race, run most days, pack their running kit for holidays and sync their watch the moment they stop. If you're newer to running and can see yourself getting into it, you're welcome too. I'll try to keep things accessible, even when they get a bit full-on.

Why write it up

Writing is the best way I know to learn. It also forces me to filter out bad information, because I owe readers more than rubbish. A lot of running content online is low-effort, even from people with impressive credentials. There's room for someone to take these topics slowly and be honest about what we know and what we don't.

On AI

I've written software for over 25 years, and AI tools like Claude are part of how I build. I brief them the way I'd brief a junior engineer and review everything they produce. I stay responsible for the design and the maths. I write the first draft of every post by hand, and AI sometimes helps with phrasing. I've also always loved an em dash, so if you spot one, it isn't proof that a machine wrote the piece.

About me

My background spans computer science, chartered accountancy, auditing banks, startups and a decade in blockchain and crypto, much of it leading product and engineering teams. Nerd Athletic is my antidote to all that: something tangible that I love, with a simpler goal. Run faster.

Get in touch

If any of this interests you, email me at rojeee [at] gmail [dot] com. Ask a question, or tell me what I've got wrong. I read everything you send.

Cheers,
Roger

Subscribe

Get the next post by email.

New posts when I publish, and first access to the app when it launches, with a founding price for the first 100 subscribers.

Subscribe on Substack

Opens Substack. Unsubscribe any time.