tag:famoe.ly,2026:run-feedfamoe.ly pipeline runsWhat the moe. prediction pipeline did, one entry per run that had something to report.2026-07-20T17:54:57Zfamoe.ly pipelinetag:famoe.ly,2026-07-20:run-1754573 nights scored, called 18 of 29 expected2026-07-20T17:54:57Z2026-07-20T17:54:57Z<p>Run at 17:54 UTC (0m00s). All sources answered. 381 shows in the archive.</p><table><tr><th align="left" valign="top">New data</th><td>override +1 &#183; setlistfm +1 &#183; no new songs in vocabulary</td></tr><tr><th align="left" valign="top">Unrecognized names</th><td>On the Run</td></tr><tr><th align="left" valign="top">Payout</th><td>07/16: hit 5 of 12 played &#183; expected 9.8 &#183; recall@30 0.417<br>calibration 0.51</td></tr><tr><th align="left" valign="top">Payout</th><td>07/17: hit 8 of 15 played &#183; expected 9.9 &#183; recall@30 0.533<br>calibration 0.81</td></tr><tr><th align="left" valign="top">Payout</th><td>07/18: hit 5 of 8 played &#183; expected 9.7 &#183; recall@30 0.625<br>calibration 0.52</td></tr><tr><th align="left" valign="top">Calibration streak</th><td>over-promised 10 of the last 10 scored nights</td></tr><tr><th align="left" valign="top">Model</th><td>out-of-sample 0.270 &#177;0.021 (-0.001 , inside the error bar)<br>vs naive frequency 0.157</td></tr><tr><th align="left" valign="top">Hyperparameters</th><td>halflife 200 / blend 0.3 / gap 0.35 / era 0.5 (frozen)</td></tr><tr><th align="left" valign="top">Board top-10</th><td>in: Not Coming Down, Head, George, 32 Things, Puebla<br>out: Yellow Tigers, In Stride, Rebubula, Ups and Downs, Lazarus<br><small>full top-30 churn on the <a href="https://famoe.ly/scorecard/">Scorecard</a></small></td></tr></table>tag:famoe.ly,2026-07-19:run-201407Feed starting up, 379 shows in the archive2026-07-19T20:14:07Z2026-07-19T20:14:07Z<p>Run at 20:14 UTC (0m00s). All sources answered. 379 shows in the archive.</p><table><tr><th align="left" valign="top">Most recent scored night</th><td>07/12: hit 5 of 8 played &#183; expected 8.1 &#183; recall@30 0.625<br>calibration 0.62</td></tr><tr><th align="left" valign="top">Calibration streak</th><td>over-promised 10 of the last 10 scored nights</td></tr><tr><th align="left" valign="top">Model</th><td>out-of-sample 0.271 &#177;0.020<br>vs naive frequency 0.147</td></tr><tr><th align="left" valign="top">Hyperparameters</th><td>halflife 200 / blend 0.3 / gap 0.35 / era 0.5 (frozen)</td></tr><tr><th align="left" valign="top">Board top-10</th><td>unchanged</td></tr></table>