I’ve built a few pacing models, and they’re great for spotting where you’re under‑ or over‑running, but the data has to match my body, not just the pack. A model that tells me to drop 30 seconds on mile three could kill me if I’m already low on glycogen or my joints are tired. So I treat the numbers as a baseline, not a law. If the model says I should stay 5 seconds ahead, I’ll try it, but if my heart rate spikes or my cadence drops, I’ll adjust on the fly. Predictive pacing is a tool, not a cheat code.