Calibration

Calibration is how Worth Radar keeps itself honest. It compares what the app predicted (profit, resale value, effort) against what actually happened on the flips you've completed — and surfaces any systematic bias.

What the report shows

A read-only predicted-vs-actual breakdown — overall, by item category, and by recommendation — so you can spot patterns like "resale value is consistently over-predicted on electronics." It's built from the outcomes you record in the Flip Tracker.

It feeds back into scoring

Beyond reporting, calibration gently nudges future predictions toward observed reality with a small, bounded per-category correction. Two safeguards keep this from misbehaving:

• It's dormant until there's enough history — with only a handful of sold outcomes it changes nothing, so early estimates aren't jerked around by a tiny sample.
• The correction is bounded — it can only nudge within a limited range, and it fails neutral (no change) on any error, so calibration can never corrupt or break scoring.

What you need to do

Just record your real outcomes as you complete flips. The more actual results accumulate, the sharper — and more locally accurate — your scores become over time.