Power & data quality: garbage in, garbage out

The principle

Every metric Grepiac computes, CP, W′, IF, Load, zones, Running Effectiveness, CdA, is only as good as the data going in. Garbage in, garbage out. A clean, well-calibrated file produces analysis you can coach off; a noisy or mis-scaled one produces confident-looking nonsense.

Implausibly high bike power

If an athlete's bike power looks far too high (or their modeled CP/W′ come out unrealistic), suspect the input, not the athlete:

• Single-sided power meter doubling incorrectly, or a left/right scaling error.
• Bad / missing calibration: no zero-offset, drift of tens of watts.
• Wheel-on (or "dumb") trainer reporting estimated power from a speed curve, often optimistic and tyre-pressure dependent.
• A trainer's power treated as outdoor, or two power sources fighting.

The fix is at the source: calibrate before each ride, use a consistent power source, and prefer a properly-zeroed dual-sided or hub/trainer power meter for the numbers that feed thresholds.

Run power

Garmin/optical run power reads ~30% high vs Stryd and shouldn't be mixed with it, see Run power: Stryd vs Garmin.

What to do when a file is off

• Trim dead/stationary time so it doesn't dilute averages (Trimming).
• Set a distance override for a mis-measured open-water swim.
• If thresholds were modeled from bad data, set measured CP/W′ so the match-burn and zone analysis are trustworthy.

Good inputs lead to trustworthy trends. Treat a single suspicious session as a data question first, a fitness question second.