Articles
Power pacing, CdA, courses and engine validation. Written for people who take their numbers to the start line.
2026-10-20What is a good CdA? The real spread across 8 triathletes and 71 ride filesOn a Tri/TT bike, the median from training ride files is 0.234, and race files come out a little lower, at 0.224. This article uses de-identified statistics from 8 athletes and 71 files on AeroPacer: where your number falls, why road bike riders have to start from a lookup table, and how many minutes a 0.01 drop in CdA is worth over a 70.3 and a full distance (226).
2026-09-21Wind tunnel test vs race power file: one rider's CdA measured two ways, 0.8% apartIn January 2026 Pengcheng Li had his race equipment measured in the HKUST wind tunnel at CdA 0.2123 m². Six months later AeroPacer solved 0.214 m² from his IRONMAN Hamburg power file, a difference of 0.8%. This article lays out both measurements in full, including why the two were never supposed to be identical and what could still make the agreement luck.
2026-09-05Which of the four plans? Start by counting this year's racesBuilding a plan is free. A Race Pass buys one race's weather, output files and your own CdA. CdA Pro turns your CdA into long-term data. The Coach plan lets one person run a whole squad. This article explains how the four plans differ, why the lines are drawn where they are, and which one is for you.
2026-09-03What is CdA? Why your drag area decides most of your bike legCdA, also called drag area, is your drag coefficient multiplied by your frontal area. It measures how much air you have to push aside as you ride. At 37 km/h on a flat road, 84% of your power goes into pushing air, and CdA is the input most likely to throw off a finish-time prediction. This article covers how it is calculated, what 0.01 is worth in minutes, and how to get your own number from a power file.
2026-09-03Which IRONMAN courses are fastest? How the course speed score worksStart at 100 and take off four deductions: Terrain, Technical, Heat and Wind. Each has a clear formula and a cap, and the weather comes from actual records for the same time of year over the last five years. This article lays out the formula, then runs it on seven courses.
2026-09-01How accurately does AeroPacer calculate your bike split? Replaying 64 real power filesI fed the actual power and that day's weather from 64 real power files, ridden by nine people, back through the engine. The median bike split error is 2.1%, and 86% land within ±5%. This tests the physics model, not a pre-race prediction. The method, the sample, the charts and the full list of races are all laid out here.