Stackruns

Coherent 0.2.10: 3670 ± 52 over 7 anchors

Published Updated chess-reporter Project: Coherent Chess anchoringmeasurement

Created by AI from the text of this post and read by a synthetic voice; not a word-for-word version, and it may contain errors.

Project measurement, not in the register: an anchoring against foreign engines with a listed rating. The figures come from the data delivery (entry M-NNNN One Elo measurement: a build anchored against foreign engines with a listed rating, on a measuring machine of its own. Numbered in the order delivered; the register run it measured is named with it. The series continues the entries E-0001 to E-0012 and starts at M-0013.); the build measured is the one from run T-0174 of 9 October 2026. The reservations below are those delivered by the day of publication; the full and current account is on the playing-strength page.

Coherent 0.2.10 played 3,584 games against 7 foreign engines with a published rating, akimbo 1.0.0, Black Marlin 9.0, PlentyChess 0.3.0, Minke 6.0.0, Stormphrax 8.0.0, Cinder 0.6.1 and Reckless 0.9.0, 512 against each at 120+1. The result is 3670 ± 52 (anchored to CCRL Computer Chess Rating Lists, a public ranking of chess engines from games under standard conditions. blitz ratings), ± 7 of it statistical. From 14 October 2026 that is the figure at the head of this site; it replaces the anchoring of 0.2.6 at 3620 ± 52.

The runs

Coherent 0.2.10: the anchors and the published figure — Project measurement, not in the register Project measurement, not in the register. Seven horizontal intervals, one per opponent, on a scale of anchored playing strength: akimbo 1.0.0 at 3656 with a half-width of 17, Black Marlin 9.0 at 3657 with a half-width of 17, PlentyChess 0.3.0 at 3646 with a half-width of 18, Minke 6.0.0 at 3750 with a half-width of 20, Stormphrax 8.0.0 at 3659 with a half-width of 17, Cinder 0.6.1 at 3675 with a half-width of 18, Reckless 0.9.0 at 3651 with a half-width of 17. Below them the published figure for 0.2.10, 3670 with a band of 52, shaded behind every row. The plus-minus values are half-widths of nominal 95 % intervals; the published band also carries the transfer term. 3600 3620 3640 3660 3680 3700 3720 3740 3760 3780 akimbo 1.0.0 3656 ± 17 Black Marlin 9.0 3657 ± 17 PlentyChess 0.3.0 3646 ± 18 Minke 6.0.0 3750 ± 20 Stormphrax 8.0.0 3659 ± 17 Cinder 0.6.1 3675 ± 18 Reckless 0.9.0 3651 ± 17 0.2.10 published 3670 ± 52 playing strength anchored to CCRL blitz ratings
Each row is one opponent: the value anchored through it, with its 95 % half-width. The bottom row is the published figure for 0.2.10, 3670 ± 52; the shaded strip carries its band, transfer term included, across all rows. Project measurement, not in the register.
OpponentListedGamesScore rateDifferenceFrom it
akimbo 1.0.03567 ± 1051262.50 %+88.74 ± 14.043656 ± 17
Black Marlin 9.03590 ± 1051259.47 %+66.63 ± 13.533657 ± 17
PlentyChess 0.3.03592 ± 1251257.71 %+54.04 ± 13.253646 ± 18
Minke 6.0.03594 ± 1351271.00 %+155.51 ± 15.363750 ± 20
Stormphrax 8.0.03746 ± 1251237.79 %−86.57 ± 12.003659 ± 17
Cinder 0.6.13763 ± 1351237.60 %−88.02 ± 12.683675 ± 18
Reckless 0.9.03766 ± 1151234.08 %−114.59 ± 13.013651 ± 17

All half-widths at the 95 % level, as the delivery states them; the column From it is the anchor plus the measured difference. The listed ratings are from the CCRL Blitz Rating List as of 2026-09-21, 2026-09-26.

How the figure is formed

Recalculated from the anchors: the 7 anchored values, averaged with weights inverse to their variance, give 3667.4 with a statistical half-width of ± 6.7; the delivery rounds it to ± 7. With the measuring side's transfer term the band is ± 51.4; the delivery names 3670 ± 52, and this site publishes the delivery's figure. The anchored values contradict one another (chi-square 78.55 at 6 degrees of freedom, bound 12.59; half-widths read as 95 % half-widths).

Against 0.2.6 7 anchors were played in both campaigns, akimbo 1.0.0, Black Marlin 9.0, PlentyChess 0.3.0, Minke 6.0.0, Stormphrax 8.0.0, Cinder 0.6.1 and Reckless 0.9.0. Per anchor the difference of the two measured differences is +27.90 ± 20.81, +34.11 ± 19.35, +42.45 ± 20.01, +117.53 ± 20.48, +50.62 ± 19.81, +32.39 ± 19.96 and +52.36 ± 21.29; combined, +51 ± 8. The per-anchor values contradict one another (chi-square 52.24 at 6 degrees of freedom, bound 12.59). That is the measured change: the opponent's listed rating and the transfer term cancel in it. The 50 points between the two headline figures are not a measured gain; they carry both bands in full.

Coherent 0.2.10 against 0.2.6: the change per shared anchor — Project measurement, not in the register Project measurement, not in the register. Seven horizontal intervals, one per anchor played in both campaigns, on a scale of Elo difference against 0.2.6: akimbo 1.0.0 at +28 with a half-width of 21, Black Marlin 9.0 at +34 with a half-width of 19, PlentyChess 0.3.0 at +42 with a half-width of 20, Minke 6.0.0 at +118 with a half-width of 20, Stormphrax 8.0.0 at +51 with a half-width of 20, Cinder 0.6.1 at +32 with a half-width of 20, Reckless 0.9.0 at +52 with a half-width of 21. Below them the combined value, +51 ± 8. A dashed vertical line marks zero, no change. 0 +20 +40 +60 +80 +100 +120 +140 akimbo 1.0.0 +28 ± 21 Black Marlin 9.0 +34 ± 19 PlentyChess 0.3.0 +42 ± 20 Minke 6.0.0 +118 ± 20 Stormphrax 8.0.0 +51 ± 20 Cinder 0.6.1 +32 ± 20 Reckless 0.9.0 +52 ± 21 combined +51 ± 8 difference of the two measured differences against 0.2.6, in Elo
Per anchor played in both campaigns, the difference of the two measured differences against 0.2.6; the bottom row combines them to +51 ± 8. The dashed line is zero: no change. The opponent's listed rating and the transfer term cancel in each row. Project measurement, not in the register.

What belongs to this figure

  1. Measured outside the formal registration procedure on the arena measurement station under the export rule EV-1 specified in advance. The coherent account, under which agent sessions also run, holds unrestricted administrator permissions without a password requirement there (clarification of permissions pending); with administrator permissions, running blocks were readable and the register could be altered, so this entry does not prove that no one could have modified the result. Every result line of this measurement path is recorded twice, independently, with its checksum: export rule, register header and pin time were recorded within 60 s of the pin reading, prior to the first counted block; counted are only blocks that started after the pin reading and ended before the final reading.
  2. The calibration of the total band is not confirmed from the complete raw data.
  3. The evidence for the scale showing that this book maps one-to-one onto the list at 2+1 comes from a round robin with three openings per pairing across ten colour-swapped rounds. The grouping is not accounted for in the calculation provided (intraclass correlation approximately 0.30, factor 1.93), so the evidence is approximately half as strong as reported. This is a reservation regarding the model assumption, not regarding the error calculation: Coherent's own anchoring runs had a distinct opening per colour pair (256 of 256 per anchor), and the statistical half-width of 6.7 remains unaffected by this.
  4. Anchor reservation: 1 of 7 anchors (akimbo 1.0.0) lies more than 100 Elo Strength difference to the opponent, estimated from the games, with a 95-percent half-width where the oracle reported one. Never an absolute rating. below the version measured.
  5. Anchor reservation: host offset not measured. Played on the measuring station arena, one game per physical core of an AMD Ryzen 9 9950X (with AVX-512); the nominal time 120+1 is effectively a longer time control there than on the virtualised hosts of the earlier entries (most recently 64 vCPU, KVM). The difference from the earlier entry is not a measured change.
  6. Anchor reservation: Minke 6.0.0 lost 208 of 512 games on time forfeit (124) or termination (84), all on the side of Minke; Coherent had no time forfeit. The export rule includes these games as established beforehand. Without Minke, the version measured would diagnostically stand roughly 10 to 13 Elo lower (mean of the remaining six anchored values 3657), within the band. Added on 14.10.2026 at the request of the website.
  7. Statistical half-width plus transfer uncertainty: a given band with open calibration, without confirmed 95-percent coverage. A round robin among the five anchors showed that an individual published rating deviates under foreign conditions. From that, ± 51 remains (1.96 times 36.9 divided by root 2, deliberately the value for two anchors); a shared error does not shrink with the number of anchors (here seven). Together with the ± 6.7 from games and list errors, that yields ± 52.

Without a provenance record in the delivery: akimbo 1.0.0, Black Marlin 9.0, PlentyChess 0.3.0, Minke 6.0.0, Stormphrax 8.0.0, Cinder 0.6.1 and Reckless 0.9.0.

Conditions

120+1, book 8moves_v3 with 34,700 positions, each colour pair with an opening of its own, 64 MB hash, one thread, 15 concurrent games, no adjudication, a fixed number of games without SPRT Sequential probability ratio test. Games are played until the accumulated evidence reaches one of two bounds; then the test stops.; 124 losses on time. The measurement is dated 14 October 2026, 06:26 UTC.

Generated from the data delivery. The recalculation is the site's; every other figure is the measuring side's.