Technical reference for the math and data pipeline underneath RideTool. You don't need any of this to use the product, but it's here if you want to understand exactly how the numbers are calculated.
The rest of the docs speak rider: fitness, freshness, effort. This section drops the translation and shows the machinery those words come from. If words like "exponentially weighted average" make you close the tab, close the tab. Your emails and your Me Card will keep working exactly the same.
The formulas page covers the training-load model RideTool is built on: how each ride gets an effort score, how those scores roll up into the long-term fitness curve and the short-term fatigue curve, and how the gap between them becomes the freshness read in your emails. The sync page covers the plumbing: what happens when the same ride arrives from Strava and your Garmin, which copy wins and why, and what triggers a recompute of your fitness history.
The technically curious. Riders who saw "fitness is on the rise" in a Friday email and wanted to know what, precisely, went up. Engineers who won't trust a number until they've seen the formula. Anyone who's used other training software and wants to check that RideTool's math matches what they already know.
I published this because I wouldn't trust a black box with my own training data, and I don't expect you to either. The math is standard, published sports science, the same Banister and Coggan models the big platforms use. RideTool's job is the plain-English layer on top, and this section is where you can check my work underneath it.