None

18-07-22 | sg | lmr_depth^ | diff |
LLR: -0.12 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | Use depth - depth/8 for reduction calculation. |

18-07-22 | sg | tweak_reductions | diff |
LLR: -0.67 (-2.94,2.94) [0.00,4.00] |
sprt @ 10+0.1 th 1 | Tweak reduction formula: 0.88*d+0.12. |

18-07-22 | sg | lmr_depth | diff |
LLR: -1.44 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | Use depth + depth/8 for reduction calculation. |

18-07-22 | sg | lmr_depth^ | diff |
LLR: -1.54 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | Use depth + depth/12 for reduction calculation. |

18-07-22 | sg | lmr_depth | diff |
LLR: -2.26 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | Use depth - depth/12 for reduction calculation. |

18-07-22 | sg | lmr_nm^ | diff |
LLR: -2.96 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | No history reduction after null move if stat score is negative. |

18-07-22 | sg | lmr_depth | diff |
LLR: -2.96 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | Use depth+1 for reduction calculation. |

18-07-22 | sg | lmr_depth^ | diff |
LLR: -2.95 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | Use depth-1 for reduction calculation. |

18-07-22 | sg | lmr_depth^^ | diff |
LLR: -2.96 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | Use newDepth+1 for reduction calculation. |

18-07-22 | sg | lmr_depth^ | diff |
LLR: -2.95 (-2.94,2.94) [0.00,4.00] |
sprt @ 10+0.1 th 1 | newDepth >= 3. |

18-07-22 | sg | lmr_depth | diff |
LLR: -2.96 (-2.94,2.94) [0.00,4.00] |
sprt @ 10+0.1 th 1 | Use newDepth for reduction calculation. |

18-07-22 | sg | lmr_rewrite | diff |
LLR: 2.95 (-2.94,2.94) [-3.00,1.00] |
sprt @ 10+0.1 th 1 | Check a non-functional LMR rewrite for non-regression (if this passes i open a PR). This patch unifies the check that reductions doesn't becomes negative. In current master for captures a special handling was used. This was a source of errors if you add there new rules and doesn't correctly handle the check. This rewrite allowed also to simplify the capture branch. |

18-07-22 | sg | lmr_depth | diff |
LLR: -2.95 (-2.94,2.94) [0.00,4.00] |
sprt @ 10+0.1 th 1 | Use newDepth in LMR depth condition (including non-functional LMR rewrite). |

18-07-22 | sg | lmr_nm^ | diff |
LLR: -2.97 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | Less reductions for quiet moves after null move. |

18-07-22 | sg | lmr_nm | diff |
LLR: -2.95 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | Less reductions for non-quiet moves after null move. |

18-07-22 | sg | lmr_nm | diff |
LLR: -2.96 (-2.94,2.94) [0.00,5.00] |
sprt @ 60+0.6 th 1 | LTC: Less reductions after null move. |

18-07-22 | sg | prune_nm | diff |
LLR: -2.96 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | Less move pruning after null move. weight=8. |

18-07-22 | sg | lmr_nm^^ | diff |
LLR: -2.96 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | No history reduction after null move. |

18-07-22 | sg | prune_nm^ | diff |
LLR: -2.96 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | Less move pruning after null move. |

18-07-22 | sg | lmr_nm^^ | diff |
LLR: -2.95 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | No tt capture reduction after null move. |

18-07-22 | sg | lmr_nm | diff |
LLR: -2.95 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | No LMR after null move. |

18-07-22 | sg | lmr_nm | diff |
LLR: 2.96 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | Less reductions after null move. |

18-07-22 | sg | lmr_nm^ | diff |
LLR: -2.95 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | No quiet stat comparison reduction after null move. |

18-07-22 | sg | lmr_nm | diff |
LLR: -2.97 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | No reduction at cut node after null move. |

18-07-22 | sg | prune_nm | diff |
LLR: -2.96 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | No move pruning after null move. |

18-07-21 | sg | playoutAll | diff |
LLR: -2.96 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | Call playout for two first moves. |

18-07-21 | sg | playoutAll | diff |
LLR: -2.96 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | My try on Stefano's playout idea. Based on the stable version.Call playout for each root move |

18-07-21 | sg | lmr3 | diff |
LLR: -2.96 (-2.94,2.94) [0.00,4.00] |
sprt @ 60+0.6 th 1 | Speculative LTC (low throughput) to check scaling from this >50K yellow (framework nearly empty). Use quiet stat score comparison rules for all moves. |

18-07-21 | sg | lmr2 | diff |
LLR: -2.95 (-2.94,2.94) [0.00,5.00] |
sprt @ 60+0.6 th 1 | Speculative LTC (low throughput) to check scaling of this >60K yellow test (framework nearly empty). Use escape rule for all moves. |

18-07-21 | sg | lmr3 | diff |
LLR: -2.96 (-2.94,2.94) [0.00,4.00] |
sprt @ 10+0.1 th 1 | Use quiet stat score comparison rules for all moves. |

18-07-21 | sg | lmr2 | diff |
LLR: -2.96 (-2.94,2.94) [0.00,4.00] |
sprt @ 10+0.1 th 1 | Use escape rule for all moves (skip cut node condition). |

18-07-21 | sg | lmr3^ | diff |
LLR: -2.95 (-2.94,2.94) [0.00,4.00] |
sprt @ 10+0.1 th 1 | Use quiet stat score rule for all moves. |

18-07-21 | sg | lmr3 | diff |
LLR: -2.95 (-2.94,2.94) [0.00,4.00] |
sprt @ 10+0.1 th 1 | Use capture stat score for all moves. |

18-07-21 | sg | lmr2 | diff |
LLR: -2.95 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | Use escape rule for all moves. |

18-07-21 | sg | lmr2^ | diff |
LLR: -2.95 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | Use cut node rule for all moves. |

18-07-21 | sg | lmr2 | diff |
LLR: -2.94 (-2.94,2.94) [0.00,4.00] |
sprt @ 10+0.1 th 1 | Use cut node and escape rule for all moves. |

18-07-21 | sg | lmr | diff |
LLR: -2.28 (-2.94,2.94) [0.00,4.00] |
sprt @ 10+0.1 th 1 | Use move count rule for all moves. |

18-07-21 | sg | lmr2^ | diff |
LLR: -2.19 (-2.94,2.94) [0.00,4.00] |
sprt @ 60+0.6 th 1 | Speculative LTC of this this struggling >130K test to check scaling (low throughput). Use move count rule for all moves. |

18-07-21 | sg | lmr | diff |
LLR: -0.29 (-2.94,2.94) [0.00,4.00] |
sprt @ 10+0.1 th 1 | Use tt capture rule for all moves. |

18-07-21 | sg | lmr2 | diff |
LLR: -2.18 (-2.94,2.94) [0.00,4.00] |
sprt @ 60+0.6 th 1 | Speculative LTC of this this struggling >130K test to check scaling (low throughput). Use tt capture rule for all moves. |

18-07-21 | sg | lmr2 | diff |
LLR: -2.95 (-2.94,2.94) [0.00,4.00] |
sprt @ 10+0.1 th 1 | Using move count as well as tt capture rule for all moves struggles with >100K games. Now try to move them together outside of the non-capture branch. |

18-07-21 | sg | lmr_checks | diff |
LLR: -2.97 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | Less reductions only for non-captures. |

18-07-21 | sg | lmr^ | diff |
LLR: -2.95 (-2.94,2.94) [0.00,4.00] |
sprt @ 10+0.1 th 1 | Use pv exact rule for all moves. |

18-07-21 | sg | lmr_checks^ | diff |
LLR: -2.97 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | Less reductions only for captures. |

18-07-20 | sg | stat_bonus_root_depth_n | diff |
LLR: -2.95 (-2.94,2.94) [0.00,5.00] |
sprt @ 60+0.6 th 1 | LTC: Extend further to depth 19. |

18-07-20 | sg | tweak_stat_bonus | diff |
LLR: -2.96 (-2.94,2.94) [0.00,5.00] |
sprt @ 60+0.6 th 1 | Speculative LTC because ym stat_bonus_root_depth patch struggles. STC was yellow but the change effects only depth 18 so try LTC to trigger more often this bonus (low throughput). Try now extend to depth 18 with special value from current formula. |

18-07-20 | sg | stat_bonus_root_depth_n | diff |
LLR: -2.96 (-2.94,2.94) [0.00,5.00] |
sprt @ 60+0.6 th 1 | LTC: Sorry, I'am an idiot and used the wrong depth bounds. retest the fixed version. At STC i doesn't expect a gain but hope it is at least not a regression. Try best formula but use special value for depth 18 |

18-07-20 | sg | stat_bonus_root_depth_n | diff |
LLR: 2.96 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | Extend further to depth 19. |

18-07-20 | sg | tweak_stat_bonus | diff |
LLR: -2.95 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | Try now extend to depth 18 with soecial value from current formula. |

18-07-20 | sg | tweak_stat_bonus2 | diff |
LLR: -2.95 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | Extend furhter to depth 19. |

18-07-20 | sg | stat_bonus_cubic | diff |
LLR: -2.96 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | Extend my best cubic formula to depth 18 by using special value. |

18-07-20 | sg | stat_bonus_root_depth_n | diff |
LLR: 2.95 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | Sorry, I'am an idiot and used the wrong depth bounds. retest the fixed version. At STC i doesn't expect a gain but hope it is at least not a regression. Try best formula but use special value for depth 18 |

18-07-20 | sg | stat_bonus_root_depth_n | diff |
LLR: -1.90 (-2.94,2.94) [0.00,5.00] |
sprt @ 60+0.6 th 1 | Speculative LTC because this version is only different for depth 18 which should kick in more at LTC and it failed yellow at STC (low throughput).try best formula but use special value for depth 18. |

18-07-20 | sg | stat_bonus_root_depth_n | diff |
LLR: -2.96 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | try best formula but use special value for depth 18. |

18-07-20 | sg | stat_bonus_root_depth_n | diff |
LLR: -2.96 (-2.94,2.94) [0.00,5.00] |
sprt @ 10+0.1 th 1 | Now use no cap but lower factor. |