The Gate We Trusted Because We Built It
PlayerGem's availability gate passed candidates it should have stopped — and the flaw was invisible precisely because...
Read itLatest Articles
The Gate We Added After the Fact and Called Foresight
PlayerGem's account of a gate invented after a miss, retrofitted into the process, and mistaken for prior judgment un...
Read itThe Gate We Rebuilt After a Retired Player
PlayerGem's availability gate passed a player who had retired mid-season — a miss that exposed how roster status and ...
Read itThe Gate We Passed Twice Before We Got It
PlayerGem's availability gate failed twice in the same way before the shop understood what it was actually measuring ...
Read itThe Gate We Closed Right for the Wrong Reason
PlayerGem closed a gate correctly — the player was genuinely unavailable — but the reasoning behind the decision was ...
Read itThe Gate We Closed Too Late
PlayerGem's account of an availability gate that ran eight days past its failure point, the ratings it contaminated, ...
Read itThe Shrinkage That Was Hiding a Real Signal
PlayerGem's shrinkage routine pulled a genuine outlier back toward the base rate for months before we noticed the mec...
Read itThe Corpus Built on the Wrong Stat Version
PlayerGem rebuilt a basketball corpus around a refined shot-creation metric and spent months grading ratings against ...
Read itWhen the Corpus Was Right and the Gates Were Not
PlayerGem's gates cleared candidates the corpus had already flagged as thin — a specific week where the two layers of...
Read itChasing a Pattern That Was Two Data Points
PlayerGem's account of rating a basketball player on a two-game pattern, why the corpus gates failed to catch it, and...
Read itWhat a Good Month Nearly Cost Us
PlayerGem's Misses desk on how a strong run of accurate ratings quietly eroded the shop's calibration discipline — an...
Read itThe Rating We Should Never Have Published
PlayerGem reconstructs the single rating that cleared every internal gate, earned our highest confidence label, and w...
Read itThe Bug That Looked Like a Finding
For six weeks we believed we had discovered something real about mid-season performance. We had discovered an off-by-one error in a date comparison.
Read articleThe Week Everything Was Confidently Wrong
Our worst week was not a week of bad luck. Every rating failed for the same reason, which means it was a single mistake made many times.
Read articleWe Graded Ourselves Generously for Months
Nobody cheated. The grading logic contained several small, individually defensible choices that all happened to point the same direction.
Read articleAll The Articles
- The Gate We Trusted Because We Built It
- The Gate We Added After the Fact and Called Foresight
- The Gate We Rebuilt After a Retired Player
- The Gate We Passed Twice Before We Got It
- The Gate We Closed Right for the Wrong Reason
- The Gate We Closed Too Late
- The Shrinkage That Was Hiding a Real Signal
- The Corpus Built on the Wrong Stat Version
- When the Corpus Was Right and the Gates Were Not
- Chasing a Pattern That Was Two Data Points
- What a Good Month Nearly Cost Us
- The Rating We Should Never Have Published
- The Bug That Looked Like a Finding
- The Week Everything Was Confidently Wrong
- We Graded Ourselves Generously for Months