Regression Tests

All of the information below has been generated from the results of tests performed on the Fishtest framework. All data and graphs can be downloaded here. Please get involved if you have questions or suggestions!

Current Testing Criteria

• 1 Thread 60 seconds + 0.6 seconds for 60,000 games (21/11/2019 - current)
• 8 Threads 30 seconds + 0.3 seconds for 40,000 games (13/12/2018 - current)
• 8moves_v3.pgn opening book (09/11/2013 - current)

Past Testing Criteria

• 1 Thread 60 seconds + 0.6 seconds for 40,000 games (02/01/2016 - 21/11/2019)
• 1 Thread 60 seconds + 0.5 seconds for 40,000 games (13/10/2013 - 02/01/2016)
• 1 Thread 60 seconds + 0.5 seconds for 20,000 games (04/03/2013 - 13/10/2013)
• 8moves_GM.pgn opening book (10/04/13 - 01/11/13)
• varied.bin opening book (04/03/13 - 07/04/13)

---

Current Development

Date	Version	1 Thread	8 Threads
02/07/21	Stockfish 14
26/07/21	master vs Stockfish 14 Update default net to nn-26abeed38351.nnue [differences] 17	Elo: 2.74 ±0.8 nElo: 9.85 ±2.8   PairsRatio: 1.24 [raw statistics]
15/08/21	master vs Stockfish 14 New NNUE architecture and net [differences] 26	Elo: 9.31 ±0.9 nElo: 29.61 ±2.7   PairsRatio: 1.75 [raw statistics]	Elo: 6.08 ±0.9 nElo: 22.66 ±3.4   PairsRatio: 1.64 [raw statistics]
31/08/21	master vs Stockfish 14 Update default net to nn-735bba95dec0.nnue [differences] 39	Elo: 15.04 ±0.9 nElo: 47.86 ±2.7   PairsRatio: 2.54 [raw statistics]
15/09/21	master vs Stockfish 14 Update default net to nn-13406b1dcbe0.nnue [differences] 46	Elo: 16.64 ±0.9 nElo: 49.53 ±2.7   PairsRatio: 2.45 [raw statistics]	Elo: 12.39 ±0.9 nElo: 45.59 ±3.2   PairsRatio: 2.80 [raw statistics]
06/10/21	master vs Stockfish 14 Capping stat bonus at 2000 [differences] 57	Elo: 18.68 ±0.9 nElo: 57.70 ±2.7   PairsRatio: 3.05 [raw statistics]	Elo: 11.74 ±0.9 nElo: 43.60 ±3.2   PairsRatio: 2.70 [raw statistics]
18/10/21	master vs Stockfish 14 Simplify probCutCount away [differences] 67	Elo: 17.69 ±0.9 nElo: 54.78 ±2.7   PairsRatio: 2.84 [raw statistics]
23/10/21	master vs Stockfish 14 Adjust ButterflyHistory decay parameter [differences] 78	Elo: 17.87 ±0.9 nElo: 57.35 ±2.6   PairsRatio: 3.15 [raw statistics]	Elo: 10.27 ±0.9 nElo: 40.55 ±3.2   PairsRatio: 2.67 [raw statistics]
28/10/21	Stockfish 14.1 Official Release [differences] 80
05/11/21	master vs Stockfish 14 Tweak initial aspiration window [differences] 89	Elo: 18.71 ±0.9 nElo: 60.14 ±2.6   PairsRatio: 3.39 [raw statistics]
23/11/21	master vs Stockfish 14 Less futility pruning [differences] 98	Elo: 19.35 ±0.9 nElo: 60.39 ±2.6   PairsRatio: 3.27 [raw statistics]	Elo: 13.28 ±0.9 nElo: 49.26 ±3.2   PairsRatio: 3.13 [raw statistics]
28/11/21	master vs Stockfish 14 Refine futility pruning for parent nodes [differences] 103	Elo: 24.44 ±0.9 nElo: 76.95 ±2.5   PairsRatio: 4.86 [raw statistics]
07/12/21	master vs Stockfish 14 Update default net to nn-63376713ba63.nnue [differences] 118	Elo: 26.99 ±0.9 nElo: 83.13 ±2.5   PairsRatio: 5.47 [raw statistics]	Elo: 17.80 ±0.9 nElo: 64.76 ±3.1   PairsRatio: 4.76 [raw statistics]
14/12/21	master vs Stockfish 14 Remove NNUE scaling term [differences] 126	Elo: 26.45 ±0.9 nElo: 82.28 ±2.5   PairsRatio: 5.53 [raw statistics]
22/12/21	master vs Stockfish 14 Update default net to nn-ac07bd334b62.nnue [differences] 139	Elo: 28.65 ±0.9 nElo: 85.61 ±2.5   PairsRatio: 5.52 [raw statistics]	Elo: 17.66 ±0.9 nElo: 64.60 ±3.0   PairsRatio: 4.79 [raw statistics]
10/01/22	master vs Stockfish 14 Adjust pruning constants [differences] 148	Elo: 30.51 ±1.0 nElo: 89.03 ±2.5   PairsRatio: 5.67 [raw statistics]
29/01/22	master vs Stockfish 14 Do stats updates after LMR for captures [differences] 159	Elo: 32.07 ±1.0 nElo: 93.05 ±2.5   PairsRatio: 6.18 [raw statistics]	Elo: 22.31 ±1.0 nElo: 76.78 ±3.0   PairsRatio: 6.02 [raw statistics]
10/02/22	master vs Stockfish 14 Update architecture to "SFNNv4" [differences] 166	Elo: 34.88 ±1.0 nElo: 100.85 ±2.5   PairsRatio: 7.44 [raw statistics]	Elo: 24.37 ±1.0 nElo: 83.29 ±2.9   PairsRatio: 7.51 [raw statistics]
17/02/22	master vs Stockfish 14 Tune search at very long time control [differences] 168	Elo: 33.07 ±1.0 nElo: 95.49 ±2.5   PairsRatio: 6.47 [raw statistics]	Elo: 23.83 ±1.0 nElo: 82.31 ±2.9   PairsRatio: 7.46 [raw statistics]
19/03/22	master vs Stockfish 14 Remove ttPv tree shrinking [differences] 180	Elo: 35.01 ±1.0 nElo: 102.62 ±2.5   PairsRatio: 8.16 [raw statistics]
17/04/22	master vs Stockfish 14 Decrease LMR at PV nodes with low depth [differences] 189	Elo: 36.69 ±1.0 nElo: 107.69 ±2.4   PairsRatio: 9.49 [raw statistics]	Elo: 26.09 ±1.0 nElo: 89.84 ±2.8   PairsRatio: 9.59 [raw statistics]
18/04/22	Stockfish 15 Official Release [differences] 190

---

Fishtest Progression

---

Version Progression

1 Thread	8 Threads

---

30 Day Average

1 Thread	8 Threads

---

Draw Ratio vs Elo

1 Thread	8 Threads

---

Fishtest History

Stockfish 3 Development (01/03/13 - 30/04/13)

Date	Version	1 Thread
22/09/12	Stockfish 2.3.1
01/03/13	Fishtest Introduced
04/03/13	master vs Stockfish 2.3.1 Increase see prune depth [differences] 226	Elo: 15.00 ±2.8 nElo: 26.02 ±4.8 [raw statistics]
11/03/13	master vs Stockfish 2.3.1 Be more aggressive on trying to finish iterations [differences] 227	Elo: 15.49 ±2.8 nElo: 26.38 ±4.8 [raw statistics]
11/03/13	master vs Stockfish 2.3.1 Check for easy move just once [differences] 228	Elo: 13.42 ±2.8 nElo: 23.19 ±4.8 [raw statistics]
16/03/13	master vs Stockfish 2.3.1 Further increase SEE prune depth [differences] 232	Elo: 17.77 ±2.8 nElo: 30.53 ±4.8 [raw statistics]
24/03/13	master vs Stockfish 2.3.1 Update bestValue when futility pruning (2) [differences] 237	Elo: 16.71 ±2.8 nElo: 28.92 ±4.8 [raw statistics]
30/03/13	master vs Stockfish 2.3.1 Set IID half way between d/2 and d-4 [differences] 241	Elo: 18.76 ±2.8 nElo: 32.46 ±4.8 [raw statistics]
03/04/13	master vs Stockfish 2.3.1 Double Impact of Gain tables [differences] 242	Elo: 15.44 ±2.8 nElo: 26.21 ±4.8 [raw statistics]
06/04/13	master vs Stockfish 2.3.1 Increase null verification threshold to 12 plies [differences] 246	Elo: 17.11 ±2.8 nElo: 29.62 ±4.9 [raw statistics]
07/04/13	master vs Stockfish 2.3.1 Rescale UCI parameters to 100 [differences] 249	Elo: 19.02 ±2.8 nElo: 32.29 ±4.8 [raw statistics]
10/04/13	master vs Stockfish 2.3.1 De-templetize Position::is_draw() [differences] 257	Elo: 24.13 ±2.8 nElo: 41.10 ±4.8 [raw statistics]
19/04/13	master vs Stockfish 2.3.1 Skip a couple of popcount in previous patch [differences] 262	Elo: 28.27 ±2.9 nElo: 47.69 ±4.9 [raw statistics]
26/04/13	master vs Stockfish 2.3.1 Fix a crash introduced few days ago [differences] 270	Elo: 33.67 ±2.9 nElo: 56.61 ±4.9 [raw statistics]
28/04/13	master vs Stockfish 2.3.1 Temporary revert "Expose EvalInfo struct to search" [differences] 273	Elo: 30.86 ±2.9 nElo: 51.82 ±4.9 [raw statistics]
30/04/13	Stockfish 3 Official Release [differences] 275

Stockfish 4 Development (30/04/13 - 20/08/13)

Date	Version	1 Thread
30/04/13	Stockfish 3
16/05/13	master vs Stockfish 3 Use two counter moves instead of one [differences] 28	Elo: 22.11 ±3.0 nElo: 35.78 ±4.8 [raw statistics]
23/05/13	master vs Stockfish 3 Bunch of 3 small patches [differences] 33	Elo: 26.70 ±2.1 nElo: 42.53 ±3.4 [raw statistics]
31/05/13	master vs Stockfish 3 Passed pawn tuning [differences] 38	Elo: 29.50 ±2.2 nElo: 46.13 ±3.4 [raw statistics]
23/06/13	master vs Stockfish 3 Fix some stale comments [differences] 72	Elo: 35.47 ±2.2 nElo: 55.80 ±3.4 [raw statistics]
03/07/13	master vs Stockfish 3 Simplify aspiration window code [differences] 88	Elo: 37.36 ±2.2 nElo: 58.35 ±3.4 [raw statistics]
13/07/13	master vs Stockfish 3 Fully qualify memset and memcpy [differences] 100	Elo: 39.27 ±3.1 nElo: 61.44 ±4.9 [raw statistics]
19/07/13	master vs Stockfish 3 Halve king eval margin [differences] 110	Elo: 39.83 ±3.1 nElo: 62.10 ±4.9 [raw statistics]
25/07/13	master vs Stockfish 3 Rewrite pawn shield and storm code [differences] 133	Elo: 48.84 ±3.3 nElo: 73.11 ±4.9 [raw statistics]
03/08/13	master vs Stockfish 3 Streamline time computation [differences] 147	Elo: 50.95 ±3.2 nElo: 77.14 ±4.9 [raw statistics]
18/08/13	master vs Stockfish 3 Further tweak movecount pruning [differences] 162	Elo: 56.66 ±3.3 nElo: 85.22 ±5.0 [raw statistics]
20/08/13	Stockfish 4 Official Release [differences] 165

Stockfish DD Development (20/08/13 - 29/11/13)

Date	Version	1 Thread
20/08/13	Stockfish 4
29/08/13	master vs Stockfish 4 Enable LMR for dangerous moves [differences] 12	Elo: 16.18 ±2.9 nElo: 26.72 ±4.8 [raw statistics]
01/09/13	master vs Stockfish 4 Don't use lpthread for Android [differences] 18	Elo: 19.30 ±2.9 nElo: 32.11 ±4.8 [raw statistics]
05/09/13	master vs Stockfish 4 Do not prune useless checks in QS [differences] 23	Elo: 23.80 ±2.9 nElo: 39.78 ±4.8 [raw statistics]
07/09/13	master vs Stockfish 4 Remove unreachable values in mobility table [differences] 27	Elo: 27.66 ±2.9 nElo: 46.68 ±4.9 [raw statistics]
11/09/13	master vs Stockfish 4 Extend checks more when below alpha [differences] 35	Elo: 24.87 ±2.9 nElo: 42.04 ±4.8 [raw statistics]
12/09/13	master vs Stockfish 4 Revert "Move draw by material check" [differences] 37	Elo: 24.34 ±2.9 nElo: 41.22 ±4.8 [raw statistics]
13/09/13	master vs Stockfish 4 Increase passed bonus for having more pieces [differences] 38	Elo: 26.21 ±2.9 nElo: 43.80 ±4.8 [raw statistics]
16/09/13	master vs Stockfish 4 Fix time parameters for blitz games [differences] 47	Elo: 31.56 ±2.1 nElo: 52.33 ±3.4 [raw statistics]
23/09/13	master vs Stockfish 4 Update disabled warnings for Intel compiler [differences] 54	Elo: 34.03 ±2.1 nElo: 56.49 ±3.4 [raw statistics]
28/09/13	master vs Stockfish 4 Drop 'is' prefix from query functions [differences] 62	Elo: 33.49 ±2.9 nElo: 56.93 ±4.9 [raw statistics]
29/09/13	master vs Stockfish 4 Add more depth/positions to bench [differences] 63	Elo: 33.90 ±2.0 nElo: 56.93 ±3.4 [raw statistics]
08/10/13	master vs Stockfish 4 Use TT refined value to stand pat [differences] 66	Elo: 36.58 ±2.9 nElo: 61.07 ±4.9 [raw statistics]
08/10/13	master vs Stockfish 4 Increase slowmover and reduce instability [differences] 67	Elo: 38.91 ±2.1 nElo: 65.11 ±3.5 [raw statistics]
09/10/13	master vs Stockfish 4 Smoother transition for LMR [differences] 68	Elo: 39.29 ±2.1 nElo: 65.80 ±3.4 [raw statistics]
14/10/13	master vs Stockfish 4 Double king safety weights [differences] 75	Elo: 43.64 ±2.9 nElo: 73.12 ±4.9 [raw statistics]
18/10/13	master vs Stockfish 4 Score chain pawn also by rank [differences] 78	Elo: 49.51 ±2.1 nElo: 81.68 ±3.5 [raw statistics]
19/10/13	master vs Stockfish 4 Further increase safe checks bonus [differences] 80	Elo: 50.11 ±2.1 nElo: 82.54 ±3.5 [raw statistics]
20/10/13	master vs Stockfish 4 Further improve chain pawn evaluation [differences] 84	Elo: 49.72 ±3.1 nElo: 79.12 ±4.9 [raw statistics]
22/10/13	master vs Stockfish 4 Tweak again chain pawn bonus [differences] 87	Elo: 53.85 ±3.0 nElo: 87.51 ±4.9 [raw statistics]
24/10/13	master vs Stockfish 4 Retire mirror() [differences] 94	Elo: 55.18 ±3.0 nElo: 89.82 ±4.9 [raw statistics]
28/10/13	master vs Stockfish 4 Tweak bishop pair and knight weight [differences] 96	Elo: 52.84 ±2.1 nElo: 86.04 ±3.5 [raw statistics]
01/11/13	master vs Stockfish 4 Set timer to a fixed interval [differences] 98	Elo: 59.73 ±3.0 nElo: 96.77 ±5.0 [raw statistics]
09/11/13	master vs Stockfish 4 Futility pruning simplification [differences] 106	Elo: 60.68 ±3.0 nElo: 99.73 ±5.0 [raw statistics]
10/11/13	master vs Stockfish 4 Remove opposed flag for doubled pawns [differences] 113	Elo: 61.23 ±2.1 nElo: 100.91 ±3.5 [raw statistics]
11/11/13	master vs Stockfish 4 Simplify generate<EVASIONS> [differences] 116	Elo: 63.85 ±2.1 nElo: 105.53 ±3.5 [raw statistics]
29/11/13	master vs Stockfish 4 Add support for PPC 64bit on Linux [differences] 123	Elo: 67.44 ±2.1 nElo: 110.87 ±3.5 [raw statistics]
29/11/13	Stockfish DD Official Release [differences] 124

Stockfish 5 Development (29/11/13 - 31/05/14)

Date	Version	1 Thread
29/11/13	Stockfish DD
09/12/13	master vs Stockfish DD Research at intermediate depth if LMR is very high [differences] 23	Elo: 7.94 ±1.9 nElo: 14.37 ±3.4 [raw statistics]
19/12/13	master vs Stockfish DD Faster and simplified threat eval [differences] 29	Elo: 11.35 ±1.9 nElo: 20.06 ±3.4 [raw statistics]
23/12/13	master vs Stockfish DD Loosened trigger condition for king safety [differences] 31	Elo: 18.62 ±2.0 nElo: 32.31 ±3.4 [raw statistics]
29/12/13	master vs Stockfish DD Retire asymmThreshold [differences] 36	Elo: 19.54 ±2.0 nElo: 33.87 ±3.4 [raw statistics]
02/01/14	master vs Stockfish DD Ensure move_importance() is non-zero [differences] 46	Elo: 25.36 ±2.0 nElo: 43.67 ±3.4 [raw statistics]
08/01/14	master vs Stockfish DD Position::gives_check - use ci.ksq [differences] 55	Elo: 29.85 ±2.0 nElo: 51.32 ±3.4 [raw statistics]
14/01/14	master vs Stockfish DD Introduce 'follow up' moves [differences] 59	Elo: 29.84 ±2.0 nElo: 51.09 ±3.4 [raw statistics]
19/01/14	master vs Stockfish DD Small simplification to Position::see [differences] 64	Elo: 32.49 ±2.0 nElo: 55.65 ±3.4 [raw statistics]
29/01/14	master vs Stockfish DD Tweak bishop PSQT tables [differences] 70	Elo: 36.66 ±2.0 nElo: 62.70 ±3.4 [raw statistics]
09/02/14	master vs Stockfish DD Faster handling of king captures in Position::see [differences] 79	Elo: 38.63 ±2.0 nElo: 65.38 ±3.4 [raw statistics]
22/02/14	master vs Stockfish DD Fix a warning with Intel compiler [differences] 99	Elo: 38.93 ±2.0 nElo: 65.87 ±3.4 [raw statistics]
26/02/14	master vs Stockfish DD Dynamic draw value [differences] 100	Elo: 39.25 ±2.1 nElo: 65.32 ±3.4 [raw statistics]
14/03/14	master vs Stockfish DD Depth dependent aspiration window delta [differences] 122	Elo: 40.85 ±2.0 nElo: 68.77 ±3.5 [raw statistics]
24/03/14	master vs Stockfish DD Simplify TT replace strategy [differences] 138	Elo: 43.70 ±2.1 nElo: 73.08 ±3.5 [raw statistics]
08/04/14	master vs Stockfish DD Restrict queen mobility to safe squares [differences] 159	Elo: 47.70 ±2.1 nElo: 79.86 ±3.5 [raw statistics]
12/04/14	master vs Stockfish DD Move args parsing to UCI::loop [differences] 174	Elo: 49.21 ±2.1 nElo: 81.84 ±3.5 [raw statistics]
21/04/14	master vs Stockfish DD Reset DrawValue[] before new search [differences] 184	Elo: 54.53 ±2.1 nElo: 89.09 ±3.5 [raw statistics]
25/04/14	master vs Stockfish DD Speed up picking of killers [differences] 191	Elo: 57.08 ±2.2 nElo: 91.87 ±3.5 [raw statistics]
04/05/14	master vs Stockfish DD Revert dynamic contempt [differences] 216	Elo: 53.27 ±2.2 nElo: 84.84 ±3.5 [raw statistics]
13/05/14	master vs Stockfish DD Drop to qsearch at low depth in razoring [differences] 227	Elo: 57.15 ±1.8 nElo: 91.95 ±2.9 [raw statistics]
17/05/14	master vs Stockfish DD Fix an off-by-one bug in extract_pv_from_tt [differences] 229	Elo: 55.26 ±2.2 nElo: 87.95 ±3.5 [raw statistics]
24/05/14	master vs Stockfish DD Fix a warning with Intel compiler [differences] 234	Elo: 53.28 ±2.1 nElo: 86.68 ±3.5 [raw statistics]
31/05/14	Stockfish 5 Official Release [differences] 236

Stockfish 6 Development (31/05/14 - 27/01/15)

Date	Version	1 Thread
31/05/14	Stockfish 5
03/06/14	master vs Stockfish 5 Symmetric King Safety: take 2 [differences] 8	Elo: 5.35 ±1.8 nElo: 10.03 ±3.4 [raw statistics]
11/06/14	master vs Stockfish 5 Simplify pawn threats and merge into ThreatenedByPawn[] [differences] 22	Elo: 8.30 ±2.0 nElo: 15.43 ±3.7 [raw statistics]
29/06/14	master vs Stockfish 5 Fix Singular extension condition to handle mate scores [differences] 46	Elo: 10.78 ±1.8 nElo: 20.10 ±3.4 [raw statistics]
22/07/14	master vs Stockfish 5 Outpost tuning [differences] 62	Elo: 19.63 ±1.8 nElo: 36.29 ±3.4 [raw statistics]
06/08/14	master vs Stockfish 5 Remove insufficient material rule [differences] 73	Elo: 19.68 ±1.8 nElo: 36.49 ±3.4 [raw statistics]
04/09/14	master vs Stockfish 5 Small tweak to idle_loop() [differences] 82	Elo: 15.90 ±1.8 nElo: 29.35 ±3.4 [raw statistics]
27/09/14	master vs Stockfish 5 Remove use of half-ply reductions [differences] 93	Elo: 22.80 ±1.9 nElo: 41.96 ±3.4 [raw statistics]
15/10/14	master vs Stockfish 5 Document why initing eval tables [differences] 108	Elo: 27.58 ±1.9 nElo: 49.96 ±3.4 [raw statistics]
01/11/14	master vs Stockfish 5 Merge pull request #89 from official-stockfish/pull_no_pretty [differences] 129	Elo: 31.00 ±1.9 nElo: 55.96 ±3.4 [raw statistics]
10/11/14	master vs Stockfish 5 Profile Build with Hash=16 [differences] 148	Elo: 36.21 ±1.9 nElo: 64.71 ±3.4 [raw statistics]
25/11/14	master vs Stockfish 5 Bitbase index() from ADD to OR [differences] 168	Elo: 39.02 ±1.9 nElo: 70.14 ±3.4 [raw statistics]
07/12/14	master vs Stockfish 5 Simpler PRNG and faster magics search [differences] 181	Elo: 41.42 ±1.9 nElo: 74.04 ±3.4 [raw statistics]
22/12/14	master vs Stockfish 5 Prefer names to numbers in storm code [differences] 197	Elo: 46.26 ±1.9 nElo: 81.89 ±3.4 [raw statistics]
07/01/15	master vs Stockfish 5 Assorted formatting and comment tweaks in position.h [differences] 213	Elo: 52.10 ±2.0 nElo: 91.46 ±3.4 [raw statistics]
18/01/15	master vs Stockfish 5 Stockfish 6 Release Candidate 1 [differences] 228	Elo: 51.71 ±1.9 nElo: 91.89 ±3.5 [raw statistics]
27/01/15	Stockfish 6 Official Release [differences] 236

Stockfish 7 Development (27/01/15 - 02/01/16)

Date	Version	1 Thread
27/01/15	Stockfish 6
08/02/15	master vs Stockfish 6 Pawn Center Bind Bonus [differences] 12	Elo: 7.50 ±2.0 nElo: 12.81 ±3.4 [raw statistics]
19/03/15	master vs Stockfish 6 Retire ConditionVariable [differences] 123	Elo: 15.92 ±1.8 nElo: 29.89 ±3.4 [raw statistics]
29/03/15	master vs Stockfish 6 Remove some difficult to understand C++11 constructs [differences] 137	Elo: 19.02 ±1.8 nElo: 35.82 ±3.4 [raw statistics]
10/04/15	master vs Stockfish 6 Allow Position::init() to be called more than once [differences] 151	Elo: 22.12 ±1.8 nElo: 41.48 ±3.4 [raw statistics]
09/05/15	master vs Stockfish 6 Smart TT save [differences] 164	Elo: 28.82 ±1.8 nElo: 53.80 ±3.4 [raw statistics]
07/06/15	master vs Stockfish 6 Simplify outpost evaluation [differences] 176	Elo: 27.85 ±1.9 nElo: 50.86 ±3.4 [raw statistics]
16/07/15	master vs Stockfish 6 Fix formatting of previous patch [differences] 187	Elo: 30.74 ±1.9 nElo: 56.09 ±3.4 [raw statistics]
30/07/15	master vs Stockfish 6 Simplify IID depth formula [differences] 192	Elo: 34.04 ±1.9 nElo: 61.72 ±3.4 [raw statistics]
03/10/15	master vs Stockfish 6 File based passed pawn bonus [differences] 214	Elo: 44.23 ±1.9 nElo: 80.99 ±3.4 [raw statistics]
25/10/15	master vs Stockfish 6 Use atomics instead of volatile [differences] 232	Elo: 61.76 ±1.9 nElo: 110.69 ±3.5 [raw statistics]
27/12/15	master vs Stockfish 6 Stockfish 7 Beta 1 [differences] 267	Elo: 62.62 ±1.6 nElo: 113.94 ±2.8 [raw statistics]
02/01/16	Stockfish 7 Official Release [differences] 273

Stockfish 8 Development (02/01/16 - 01/11/16)

Date	Version	1 Thread
02/01/16	Stockfish 7
28/01/16	master vs Stockfish 7 Time management simplification [differences] 16	Elo: 4.93 ±1.5 nElo: 10.90 ±3.4 [raw statistics]
10/03/16	master vs Stockfish 7 Add follow up moves history for move ordering [differences] 28	Elo: 12.83 ±1.5 nElo: 28.44 ±3.4 [raw statistics]
08/04/16	master vs Stockfish 7 Small passed pawn simplification [differences] 42	Elo: 15.27 ±1.5 nElo: 33.74 ±3.4 [raw statistics]
20/05/16	master vs Stockfish 7 More detailed dependence of time allocation [differences] 64	Elo: 29.44 ±1.6 nElo: 63.61 ±3.4 [raw statistics]
10/06/16	master vs Stockfish 7 Stat Formula Tweak [differences] 76	Elo: 36.29 ±1.6 nElo: 76.48 ±3.4 [raw statistics]
24/07/16	master vs Stockfish 7 Allow null pruning at depth 1 [differences] 94	Elo: 49.73 ±1.7 nElo: 100.27 ±3.4 [raw statistics]
18/08/16	master vs Stockfish 7 Remove a stale assignment [differences] 101	Elo: 52.59 ±1.7 nElo: 104.74 ±3.4 [raw statistics]
07/09/16	master vs Stockfish 7 Refactor previous patch [differences] 124	Elo: 66.53 ±1.8 nElo: 131.97 ±3.3 [raw statistics]
07/10/16	master vs Stockfish 7 Optimisation of Position::see and Position::see_sign [differences] 149	Elo: 76.60 ±1.8 nElo: 151.24 ±3.3 [raw statistics]
01/11/16	Stockfish 8 Official Release [differences] 168

Stockfish 9 Development (01/11/16 - 31/01/18)

Date	Version	1 Thread
01/11/16	Stockfish 8
31/12/16	master vs Stockfish 8 Small eval cleanup and renaming [differences] 55	Elo: 5.21 ±1.5 nElo: 11.72 ±3.4 [raw statistics]
29/01/17	master vs Stockfish 8 Simplify TT penalty stat (#980) [differences] 86	Elo: 8.82 ±1.5 nElo: 19.53 ±3.4 [raw statistics]
08/03/17	master vs Stockfish 8 Helper functions to count material for both sides [differences] 106	Elo: 10.84 ±1.6 nElo: 23.73 ±3.4 [raw statistics]
20/04/17	master vs Stockfish 8 simplify logic for history based pruning [differences] 127	Elo: 15.17 ±1.6 nElo: 32.82 ±3.4 [raw statistics]
07/05/17	master vs Stockfish 8 Linear Protector bonus by distance [differences] 144	Elo: 20.25 ±1.6 nElo: 43.35 ±3.4 [raw statistics]
21/06/17	master vs Stockfish 8 Increase reduction if tt-move is a capture [differences] 167	Elo: 27.41 ±1.6 nElo: 58.52 ±3.4 [raw statistics]
26/08/17	master vs Stockfish 8 Improve multi-threaded mate finding [differences] 218	Elo: 29.32 ±1.6 nElo: 62.17 ±3.4 [raw statistics]
02/10/17	master vs Stockfish 8 Good bishops on the main diagonals [differences] 237	Elo: 32.61 ±1.6 nElo: 69.44 ±3.4 [raw statistics]
03/11/17	master vs Stockfish 8 Introduce capture history table for capture move sorting [differences] 247	Elo: 35.18 ±1.6 nElo: 73.93 ±3.4 [raw statistics]
03/12/17	master vs Stockfish 8 Use constexpr when makes sense [differences] 261	Elo: 43.31 ±1.7 nElo: 90.06 ±3.3 [raw statistics]
23/01/18	master vs Stockfish 8 Contempt 20 [differences] 286	Elo: 57.25 ±1.9 nElo: 106.75 ±3.4 [raw statistics]
31/01/18	Stockfish 9 Official Release [differences] 291

Stockfish 10 Development (31/01/18 - 29/11/18)

Date	Version	1 Thread
31/01/18	Stockfish 9
28/02/18	master vs Stockfish 9 Reintroduce depth 2 razoring (with additional margin) [differences] 41	Elo: 5.16 ±1.7 nElo: 10.02 ±3.4 [raw statistics]
07/03/18	master vs Stockfish 9 Simplification: use Arctan for the optimism S-curve [differences] 53	Elo: 6.94 ±1.7 nElo: 13.60 ±3.4 [raw statistics]
13/03/18	master vs Stockfish 9 Use intrinsics only for LSB/MSB [differences] 64	Elo: 13.03 ±1.7 nElo: 25.59 ±3.4 [raw statistics]
26/03/18	master vs Stockfish 9 Make kingRing always 8 squares [differences] 75	Elo: 13.77 ±1.7 nElo: 27.26 ±3.4 [raw statistics]
03/04/18	master vs Stockfish 9 Remove the Queen from the mobility area of minor pieces [differences] 92	Elo: 16.59 ±1.7 nElo: 32.54 ±3.4 [raw statistics]
07/04/18	master vs Stockfish 9 Reset negative statScore on fail high [differences] 94	Elo: 16.63 ±1.8 nElo: 32.25 ±3.4 [raw statistics]
23/04/18	master vs Stockfish 9 Alternative formula for dynamic contempt [differences] 106	Elo: 15.90 ±1.8 nElo: 30.76 ±3.4 [raw statistics]
29/04/18	master vs Stockfish 9 Always scale using pawn contribution [differences] 112	Elo: 14.61 ±1.7 nElo: 28.54 ±3.4 [raw statistics]
03/05/18	master vs Stockfish 9 Tweak the connected[] array value for pawns on rank 5 [differences] 116	Elo: 18.52 ±1.8 nElo: 35.63 ±3.4 [raw statistics]
13/05/18	master vs Stockfish 9 Update search.cpp [differences] 128	Elo: 24.92 ±1.8 nElo: 47.42 ±3.4 [raw statistics]
24/05/18	master vs Stockfish 9 LMR Capture Tweak [differences] 137	Elo: 26.72 ±1.8 nElo: 51.08 ±3.4 [raw statistics]
05/06/18	master vs Stockfish 9 Call cycle detection before qsearch() [differences] 148	Elo: 28.16 ±1.8 nElo: 53.79 ±3.4 [raw statistics]
11/06/18	master vs Stockfish 9 Optimize an expression in endgame.cpp [differences] 154	Elo: 29.72 ±1.9 nElo: 54.60 ±3.4 [raw statistics]
23/06/18	master vs Stockfish 9 Another set of tuned values after one million games [differences] 162	Elo: 31.98 ±1.9 nElo: 58.11 ±3.4 [raw statistics]
19/07/18	master vs Stockfish 9 Better check evasion move sorting [differences] 179	Elo: 36.70 ±1.9 nElo: 65.70 ±3.4 [raw statistics]
27/07/18	master vs Stockfish 9 Simplify cmh pruning [differences] 199	Elo: 37.45 ±1.9 nElo: 67.55 ±3.4 [raw statistics]
28/07/18	master vs Stockfish 9 Increase the mg->eg gradient for the PawnlessFlank malus [differences] 200	Elo: 35.84 ±1.9 nElo: 64.78 ±3.4 [raw statistics]
31/07/18	master vs Stockfish 9 Small tweaks to recent code changes [differences] 203	Elo: 37.67 ±1.9 nElo: 68.82 ±3.4 [raw statistics]
08/08/18	master vs Stockfish 9 First check threshold in space evaluation [differences] 207	Elo: 37.78 ±1.9 nElo: 67.82 ±3.4 [raw statistics]
12/08/18	master vs Stockfish 9 Combo of several promising parameter tweaks [differences] 211	Elo: 40.88 ±1.9 nElo: 73.46 ±3.4 [raw statistics]
14/08/18	master vs Stockfish 9 Double weight of capture history [differences] 214	Elo: 41.69 ±1.9 nElo: 75.47 ±3.4 [raw statistics]
17/08/18	master vs Stockfish 9 Use an affine formula to mix stats and eval [differences] 217	Elo: 43.15 ±1.9 nElo: 77.82 ±3.4 [raw statistics]
28/08/18	master vs Stockfish 9 Tweak stat bonus formula [differences] 220	Elo: 44.20 ±1.9 nElo: 79.79 ±3.4 [raw statistics]
28/08/18	master vs Stockfish 9 Remove PawnsOnBothFlanks [differences] 225	Elo: 42.37 ±1.9 nElo: 76.44 ±3.4 [raw statistics]
01/09/18	master vs Stockfish 9 Re-introduce "keep pawns on both flanks" [differences] 227	Elo: 46.46 ±1.9 nElo: 83.56 ±3.4 [raw statistics]
10/09/18	master vs Stockfish 9 Tweak opposite colored bishops endgame scaling [differences] 230	Elo: 45.47 ±1.9 nElo: 81.36 ±3.4 [raw statistics]
27/09/18	master vs Stockfish 9 Fix two typos in comments [differences] 235	Elo: 46.93 ±1.9 nElo: 83.72 ±3.4 [raw statistics]
14/10/18	master vs Stockfish 9 Simplify check extensions [differences] 241	Elo: 49.01 ±1.9 nElo: 87.49 ±3.4 [raw statistics]
25/10/18	master vs Stockfish 9 On main thread: reduce depth after fail high [differences] 245	Elo: 52.82 ±1.9 nElo: 95.10 ±3.4 [raw statistics]
01/11/18	master vs Stockfish 9 Fix issues from using adjustedDepth too broadly [differences] 248	Elo: 51.68 ±1.9 nElo: 92.90 ±3.4 [raw statistics]
08/11/18	master vs Stockfish 9 Update list of top CPU contributors [differences] 254	Elo: 50.43 ±1.9 nElo: 89.65 ±3.5 [raw statistics]
19/11/18	master vs Stockfish 9 Stockfish 10-beta [differences] 267	Elo: 53.77 ±1.9 nElo: 95.73 ±3.5 [raw statistics]
27/11/18	master vs Stockfish 9 Simplify casting extension [differences] 274	Elo: 54.21 ±1.9 nElo: 96.75 ±3.5 [raw statistics]
29/11/18	Stockfish 10 Official Release [differences] 277

Stockfish 11 Development (29/11/18 - 17/01/20)

Date	Version	1 Thread	8 Threads
29/11/18	Stockfish 10
06/12/18	master vs Stockfish 10 Revert "pseudo_legal() and MOVE_NONE" [differences] 8	Elo: 4.32 ±1.7 nElo: 8.40 ±3.4 [raw statistics]
13/12/18	master vs Stockfish 10 A combo of parameter tweaks [differences] 16	Elo: 8.06 ±1.8 nElo: 15.41 ±3.4 [raw statistics]	Elo: 12.70 ±1.7 nElo: 25.98 ±3.4 [raw statistics]
16/12/18	master vs Stockfish 10 Use stronglyProtected [differences] 21	Elo: 7.85 ±1.7 nElo: 15.10 ±3.4 [raw statistics]
04/01/19	master vs Stockfish 10 Check tablebase files [differences] 40	Elo: 7.20 ±1.8 nElo: 13.64 ±3.4 [raw statistics]
10/01/19	master vs Stockfish 10 Remove pvExact [differences] 45	Elo: 13.14 ±1.8 nElo: 25.03 ±3.4 [raw statistics]	Elo: 16.17 ±1.7 nElo: 33.07 ±3.4 [raw statistics]
22/01/19	master vs Stockfish 10 Simplify TrappedRook [differences] 53	Elo: 13.98 ±1.8 nElo: 26.68 ±3.4 [raw statistics]
03/02/19	master vs Stockfish 10 Less king danger if we have a knight [differences] 61	Elo: 19.77 ±1.7 nElo: 40.32 ±3.4 [raw statistics]
12/03/19	master vs Stockfish 10 Increase thread stack for OS X (#2035) [differences] 80	Elo: 16.58 ±1.8 nElo: 31.46 ±3.4 [raw statistics]
31/03/19	master vs Stockfish 10 Assorted trivial cleanups 3/2019 (#2030) [differences] 91	Elo: 16.58 ±1.8 nElo: 31.16 ±3.4 [raw statistics]	Elo: 24.33 ±1.7 nElo: 48.91 ±3.4 [raw statistics]
24/04/19	master vs Stockfish 10 Remove useless initializations (#2115) [differences] 113	Elo: 16.39 ±1.8 nElo: 31.13 ±3.4 [raw statistics]
05/05/19	master vs Stockfish 10 LMR for captures not cracking alpha [differences] 121	Elo: 16.65 ±1.8 nElo: 31.17 ±3.4 [raw statistics]
15/05/19	master vs Stockfish 10 Update failedHighCnt rule #2063 [differences] 136	Elo: 19.76 ±1.8 nElo: 36.86 ±3.4 [raw statistics]	Elo: 28.93 ±1.7 nElo: 57.18 ±3.4 [raw statistics]
09/06/19	master vs Stockfish 10 Remove depth condition for ttPv (#2166) [differences] 151	Elo: 19.87 ±1.9 nElo: 36.62 ±3.4 [raw statistics]
20/06/19	master vs Stockfish 10 More bonus for free passed pawn [differences] 161	Elo: 24.06 ±1.8 nElo: 44.48 ±3.4 [raw statistics]	Elo: 30.76 ±1.7 nElo: 60.92 ±3.4 [raw statistics]
27/06/19	master vs Stockfish 10 Bonus for double attacks [differences] 167	Elo: 22.75 ±1.9 nElo: 41.95 ±3.4 [raw statistics]
11/07/19	master vs Stockfish 10 Assorted trivial cleanups June 2019 [differences] 176	Elo: 24.39 ±1.9 nElo: 44.83 ±3.4 [raw statistics]
25/07/19	master vs Stockfish 10 Tweak of SEE pruning condition [differences] 192	Elo: 25.72 ±1.9 nElo: 47.32 ±3.4 [raw statistics]	Elo: 37.49 ±1.7 nElo: 73.63 ±3.4 [raw statistics]
14/08/19	master vs Stockfish 10 Tweak unsafe checks [differences] 198	Elo: 32.24 ±1.9 nElo: 59.45 ±3.4 [raw statistics]
26/08/19	master vs Stockfish 10 Tweak Late Move Reduction at root [differences] 207	Elo: 35.63 ±1.9 nElo: 65.62 ±3.4 [raw statistics]	Elo: 44.52 ±1.8 nElo: 86.99 ±3.4 [raw statistics]
12/09/19	master vs Stockfish 10 Scale down complexity [differences] 211	Elo: 39.10 ±1.9 nElo: 72.22 ±3.4 [raw statistics]
16/09/19	master vs Stockfish 10 Raise stack size to 8MB for pthreads [differences] 218	Elo: 37.63 ±1.9 nElo: 68.83 ±3.4 [raw statistics]	Elo: 46.57 ±1.8 nElo: 91.44 ±3.4 [raw statistics]
24/09/19	master vs Stockfish 10 Increase weight for supported pawns [differences] 226	Elo: 38.97 ±1.9 nElo: 71.81 ±3.4 [raw statistics]
05/10/19	master vs Stockfish 10 Introduce separate counter-move tables [differences] 239	Elo: 41.96 ±1.9 nElo: 77.13 ±3.4 [raw statistics]	Elo: 51.76 ±1.8 nElo: 101.26 ±3.4 [raw statistics]
18/10/19	master vs Stockfish 10 Current capture for Counter-Move history [differences] 247	Elo: 44.63 ±1.9 nElo: 81.93 ±3.4 [raw statistics]
04/11/19	master vs Stockfish 10 Rook PSQT Tuned [differences] 259	Elo: 42.20 ±1.9 nElo: 77.91 ±3.4 [raw statistics]	Elo: 52.90 ±1.8 nElo: 103.29 ±3.4 [raw statistics]
14/11/19	master vs Stockfish 10 Prune before extension [differences] 266	Elo: 43.12 ±1.9 nElo: 79.39 ±3.4 [raw statistics]
21/11/19	master vs Stockfish 10 Do lmr for more captures [differences] 271	Elo: 46.56 ±1.5 nElo: 85.60 ±2.8 [raw statistics]	Elo: 53.93 ±1.8 nElo: 105.86 ±3.4 [raw statistics]
02/12/19	master vs Stockfish 10 UnblockedStorm tuned [differences] 278	Elo: 44.88 ±1.5 nElo: 83.21 ±2.8 [raw statistics]
10/12/19	master vs Stockfish 10 Refine improving-logic [differences] 288	Elo: 47.27 ±1.5 nElo: 86.56 ±2.8 [raw statistics]	Elo: 56.62 ±1.8 nElo: 111.42 ±3.4 [raw statistics]
07/01/20	master vs Stockfish 10 Tuned nullmove search [differences] 294	Elo: 51.50 ±1.5 nElo: 98.05 ±2.8   PairsRatio: 3.28 [raw statistics]	Elo: 58.15 ±1.7 nElo: 118.69 ±3.4   PairsRatio: 4.71 [raw statistics]
17/01/20	master vs Stockfish 10 Stockfish 11 [differences] 307	Elo: 53.59 ±1.5 nElo: 102.99 ±2.8   PairsRatio: 3.53 [raw statistics]	Elo: 58.07 ±1.7 nElo: 119.25 ±3.4   PairsRatio: 4.83 [raw statistics]
17/01/20	Stockfish 11 Official Release [differences] 307

Stockfish 12 Development (17/01/20 - 02/09/20)

Date	Version	1 Thread	8 Threads
17/01/20	Stockfish 11
28/01/20	master vs Stockfish 11 More bonus for bestMoves [differences] 16	Elo: -2.47 ±1.3 nElo: -5.21 ±2.8   PairsRatio: 0.93 [raw statistics]
31/01/20	master vs Stockfish 11 Revert 5 recent patches [differences] 19	Elo: 0.85 ±1.3 nElo: 1.80 ±2.8   PairsRatio: 1.02 [raw statistics]
27/02/20	master vs Stockfish 11 Weak queen protection [differences] 32	Elo: 1.33 ±1.3 nElo: 2.80 ±2.8   PairsRatio: 1.04 [raw statistics]	Elo: 0.51 ±1.4 nElo: 1.21 ±3.4   PairsRatio: 1.02 [raw statistics]
20/03/20	master vs Stockfish 11 Adjust singular extension search depth [differences] 48	Elo: 2.94 ±1.3 nElo: 6.21 ±2.8   PairsRatio: 1.10 [raw statistics]
07/04/20	master vs Stockfish 11 Introduce capture history pruning [differences] 63	Elo: 5.74 ±1.3 nElo: 12.11 ±2.8   PairsRatio: 1.16 [raw statistics]	Elo: 6.49 ±1.5 nElo: 15.05 ±3.4   PairsRatio: 1.23 [raw statistics]
16/04/20	master vs Stockfish 11 Remove one condition in probcut TTmove [differences] 76	Elo: 11.33 ±1.3 nElo: 23.65 ±2.8   PairsRatio: 1.36 [raw statistics]
02/05/20	master vs Stockfish 11 Fishtest Tuning Framework [differences] 84	Elo: 15.21 ±1.3 nElo: 32.08 ±2.8   PairsRatio: 1.52 [raw statistics]	Elo: 16.71 ±1.5 nElo: 37.85 ±3.4   PairsRatio: 1.69 [raw statistics]
21/05/20	master vs Stockfish 11 Tweak knight mobility [differences] 100	Elo: 15.97 ±1.3 nElo: 33.08 ±2.8   PairsRatio: 1.52 [raw statistics]
06/06/20	master vs Stockfish 11 Use lowply-history also on low depths [differences] 117	Elo: 19.45 ±1.4 nElo: 39.70 ±2.8   PairsRatio: 1.65 [raw statistics]	Elo: 23.70 ±1.5 nElo: 53.99 ±3.4   PairsRatio: 2.14 [raw statistics]
13/06/20	master vs Stockfish 11 Tuned values for search constants [differences] 127	Elo: 20.91 ±1.4 nElo: 43.14 ±2.8   PairsRatio: 1.72 [raw statistics]	Elo: 24.86 ±1.5 nElo: 56.58 ±3.4   PairsRatio: 2.22 [raw statistics]
29/06/20	master vs Stockfish 11 Tweak single queen endgame scaling [differences] 148	Elo: 25.67 ±1.3 nElo: 53.10 ±2.8   PairsRatio: 1.98 [raw statistics]
17/07/20	master vs Stockfish 11 Do not overwrite valuable TT data [differences] 163	Elo: 26.44 ±1.3 nElo: 54.75 ±2.8   PairsRatio: 2.02 [raw statistics]	Elo: 30.71 ±1.5 nElo: 69.50 ±3.4   PairsRatio: 2.65 [raw statistics]
31/07/20	master vs Stockfish 11 Tweak cutnode reduction [differences] 167	Elo: 25.49 ±1.4 nElo: 52.38 ±2.8   PairsRatio: 1.95 [raw statistics]	Elo: 32.39 ±1.5 nElo: 73.03 ±3.4   PairsRatio: 2.80 [raw statistics]
06/08/20	master vs Stockfish 11 Add NNUE evaluation [differences] 168	Elo: 83.42 ±1.7 nElo: 144.72 ±3.0   PairsRatio: 5.11 [raw statistics]	Elo: 86.10 ±1.9 nElo: 163.20 ±3.6   PairsRatio: 7.68 [raw statistics]
07/08/20	Add NNUE evaluation Official Release [differences] 168
08/08/20	master vs Stockfish 11 LMR search tweak [differences] 185	Elo: 106.20 ±1.7 nElo: 189.91 ±3.2   PairsRatio: 9.17 [raw statistics]
11/08/20	master vs Stockfish 11 This commit enables a mixed bench [differences] 205	Elo: 125.60 ±1.7 nElo: 224.82 ±3.3   PairsRatio: 14.05 [raw statistics]	Elo: 111.78 ±1.9 nElo: 217.93 ±3.8   PairsRatio: 18.46 [raw statistics]
18/08/20	master vs Stockfish 11 Fix Makefile typo [differences] 226	Elo: 121.54 ±1.7 nElo: 215.22 ±3.3   PairsRatio: 12.38 [raw statistics]	Elo: 111.40 ±1.9 nElo: 213.20 ±3.8   PairsRatio: 15.49 [raw statistics]
30/08/20	master vs Stockfish 11 Update parameters in classical evaluation [differences] 255	Elo: 130.96 ±1.7 nElo: 232.58 ±3.4   PairsRatio: 15.34 [raw statistics]
02/09/20	master vs Stockfish 11 Stockfish 12 [differences] 262	Elo: 133.65 ±1.7 nElo: 238.67 ±3.4   PairsRatio: 16.72 [raw statistics]	Elo: 117.62 ±1.9 nElo: 224.93 ±3.8   PairsRatio: 19.95 [raw statistics]
02/09/20	Stockfish 12 Official Release [differences] 262

Stockfish 13 Development (02/09/20 - 18/02/21)

Date	Version	1 Thread	8 Threads
02/09/20	Stockfish 12
08/09/20	master vs Stockfish 12 Double probability of using classical eval [differences] 10	Elo: 6.49 ±1.0 nElo: 17.26 ±2.8   PairsRatio: 1.32 [raw statistics]
21/09/20	master vs Stockfish 12 Add large page support for NNUE weights [differences] 21	Elo: 13.88 ±1.0 nElo: 37.07 ±2.8   PairsRatio: 1.80 [raw statistics]	Elo: 8.97 ±1.1 nElo: 27.99 ±3.4   PairsRatio: 1.67 [raw statistics]
28/09/20	master vs Stockfish 12 Include pawns in NNUE scaling [differences] 33	Elo: 23.15 ±1.0 nElo: 63.29 ±2.7   PairsRatio: 2.92 [raw statistics]
18/10/20	master vs Stockfish 12 Do more reductions for late quiet moves [differences] 41	Elo: 24.09 ±1.0 nElo: 64.60 ±2.7   PairsRatio: 2.93 [raw statistics]	Elo: 21.17 ±1.1 nElo: 65.43 ±3.2   PairsRatio: 3.58 [raw statistics]
01/11/20	master vs Stockfish 12 Update default net to nn-cb26f10b1fd9.nnue [differences] 48	Elo: 28.02 ±1.1 nElo: 71.59 ±2.7   PairsRatio: 3.15 [raw statistics]
15/11/20	master vs Stockfish 12 Rook Mobility Tweak [differences] 60	Elo: 29.99 ±1.0 nElo: 80.38 ±2.7   PairsRatio: 3.91 [raw statistics]	Elo: 25.49 ±1.1 nElo: 77.27 ±3.2   PairsRatio: 4.52 [raw statistics]
29/11/20	master vs Stockfish 12 Update default net to nn-62ef826d1a6d.nnue [differences] 72	Elo: 30.61 ±1.0 nElo: 82.09 ±2.7   PairsRatio: 4.02 [raw statistics]
14/12/20	master vs Stockfish 12 Increase reduction in case of stable best move [differences] 79	Elo: 32.09 ±1.0 nElo: 85.57 ±2.7   PairsRatio: 4.28 [raw statistics]	Elo: 27.50 ±1.1 nElo: 83.75 ±3.1   PairsRatio: 5.31 [raw statistics]
31/12/20	master vs Stockfish 12 WeakUnopposed penalty for backwards [differences] 89	Elo: 33.71 ±1.0 nElo: 90.01 ±2.6   PairsRatio: 4.69 [raw statistics]
13/01/21	master vs Stockfish 12 Optimize generate_moves [differences] 101	Elo: 32.40 ±1.0 nElo: 87.55 ±2.7   PairsRatio: 4.54 [raw statistics]	Elo: 28.29 ±1.1 nElo: 86.06 ±3.1   PairsRatio: 5.65 [raw statistics]
15/02/21	master vs Stockfish 12 Small trivial clean-ups, February 2021 [differences] 121	Elo: 36.03 ±1.1 nElo: 95.63 ±2.7   PairsRatio: 5.17 [raw statistics]	Elo: 29.08 ±1.1 nElo: 89.08 ±3.1   PairsRatio: 6.16 [raw statistics]
18/02/21	Stockfish 13 Official Release [differences] 123

Stockfish 14 Development (18/02/21 - 02/07/21)

Date	Version	1 Thread	8 Threads
18/02/21	Stockfish 13
26/02/21	master vs Stockfish 13 Introduce DistanceFromPV [differences] 4	Elo: 1.34 ±0.8 nElo: 4.55 ±2.8   PairsRatio: 1.10 [raw statistics]
24/03/21	master vs Stockfish 13 Small cleanups (march 2021) [differences] 18	Elo: 0.61 ±0.8 nElo: 2.09 ±2.8   PairsRatio: 1.05 [raw statistics]
15/04/21	master vs Stockfish 13 Use classical eval for Bishop vs Pawns [differences] 29	Elo: 1.47 ±0.8 nElo: 4.93 ±2.8   PairsRatio: 1.11 [raw statistics]	Elo: 0.89 ±0.9 nElo: 3.38 ±3.4   PairsRatio: 1.08 [raw statistics]
22/05/21	master vs Stockfish 13 Sometimes change the balance [differences] 62	Elo: 10.12 ±1.0 nElo: 29.34 ±2.8   PairsRatio: 1.67 [raw statistics]
14/06/21	master vs Stockfish 13 Update default net to nn-8e47cf062333.nnue [differences] 90	Elo: 21.80 ±1.1 nElo: 57.47 ±2.7   PairsRatio: 2.57 [raw statistics]	Elo: 16.96 ±1.1 nElo: 52.38 ±3.3   PairsRatio: 2.71 [raw statistics]
18/06/21	master vs Stockfish 13 Make net nn-50144f835024.nnue the default [differences] 100	Elo: 25.56 ±1.0 nElo: 69.95 ±2.7   PairsRatio: 3.29 [raw statistics]
29/06/21	master vs Stockfish 13 Update Top CPU Contributors [differences] 113	Elo: 30.27 ±1.0 nElo: 83.66 ±2.6   PairsRatio: 4.40 [raw statistics]	Elo: 22.62 ±1.1 nElo: 71.15 ±3.1   PairsRatio: 4.28 [raw statistics]
02/07/21	Stockfish 14 Official Release [differences] 114

Stockfish 15 Development (02/07/21 - 18/04/22)

Date	Version	1 Thread	8 Threads
02/07/21	Stockfish 14
26/07/21	master vs Stockfish 14 Update default net to nn-26abeed38351.nnue [differences] 17	Elo: 2.74 ±0.8 nElo: 9.85 ±2.8   PairsRatio: 1.24 [raw statistics]
15/08/21	master vs Stockfish 14 New NNUE architecture and net [differences] 26	Elo: 9.31 ±0.9 nElo: 29.61 ±2.7   PairsRatio: 1.75 [raw statistics]	Elo: 6.08 ±0.9 nElo: 22.66 ±3.4   PairsRatio: 1.64 [raw statistics]
31/08/21	master vs Stockfish 14 Update default net to nn-735bba95dec0.nnue [differences] 39	Elo: 15.04 ±0.9 nElo: 47.86 ±2.7   PairsRatio: 2.54 [raw statistics]
15/09/21	master vs Stockfish 14 Update default net to nn-13406b1dcbe0.nnue [differences] 46	Elo: 16.64 ±0.9 nElo: 49.53 ±2.7   PairsRatio: 2.45 [raw statistics]	Elo: 12.39 ±0.9 nElo: 45.59 ±3.2   PairsRatio: 2.80 [raw statistics]
06/10/21	master vs Stockfish 14 Capping stat bonus at 2000 [differences] 57	Elo: 18.68 ±0.9 nElo: 57.70 ±2.7   PairsRatio: 3.05 [raw statistics]	Elo: 11.74 ±0.9 nElo: 43.60 ±3.2   PairsRatio: 2.70 [raw statistics]
18/10/21	master vs Stockfish 14 Simplify probCutCount away [differences] 67	Elo: 17.69 ±0.9 nElo: 54.78 ±2.7   PairsRatio: 2.84 [raw statistics]
23/10/21	master vs Stockfish 14 Adjust ButterflyHistory decay parameter [differences] 78	Elo: 17.87 ±0.9 nElo: 57.35 ±2.6   PairsRatio: 3.15 [raw statistics]	Elo: 10.27 ±0.9 nElo: 40.55 ±3.2   PairsRatio: 2.67 [raw statistics]
28/10/21	Stockfish 14.1 Official Release [differences] 80
05/11/21	master vs Stockfish 14 Tweak initial aspiration window [differences] 89	Elo: 18.71 ±0.9 nElo: 60.14 ±2.6   PairsRatio: 3.39 [raw statistics]
23/11/21	master vs Stockfish 14 Less futility pruning [differences] 98	Elo: 19.35 ±0.9 nElo: 60.39 ±2.6   PairsRatio: 3.27 [raw statistics]	Elo: 13.28 ±0.9 nElo: 49.26 ±3.2   PairsRatio: 3.13 [raw statistics]
28/11/21	master vs Stockfish 14 Refine futility pruning for parent nodes [differences] 103	Elo: 24.44 ±0.9 nElo: 76.95 ±2.5   PairsRatio: 4.86 [raw statistics]
07/12/21	master vs Stockfish 14 Update default net to nn-63376713ba63.nnue [differences] 118	Elo: 26.99 ±0.9 nElo: 83.13 ±2.5   PairsRatio: 5.47 [raw statistics]	Elo: 17.80 ±0.9 nElo: 64.76 ±3.1   PairsRatio: 4.76 [raw statistics]
14/12/21	master vs Stockfish 14 Remove NNUE scaling term [differences] 126	Elo: 26.45 ±0.9 nElo: 82.28 ±2.5   PairsRatio: 5.53 [raw statistics]
22/12/21	master vs Stockfish 14 Update default net to nn-ac07bd334b62.nnue [differences] 139	Elo: 28.65 ±0.9 nElo: 85.61 ±2.5   PairsRatio: 5.52 [raw statistics]	Elo: 17.66 ±0.9 nElo: 64.60 ±3.0   PairsRatio: 4.79 [raw statistics]
10/01/22	master vs Stockfish 14 Adjust pruning constants [differences] 148	Elo: 30.51 ±1.0 nElo: 89.03 ±2.5   PairsRatio: 5.67 [raw statistics]
29/01/22	master vs Stockfish 14 Do stats updates after LMR for captures [differences] 159	Elo: 32.07 ±1.0 nElo: 93.05 ±2.5   PairsRatio: 6.18 [raw statistics]	Elo: 22.31 ±1.0 nElo: 76.78 ±3.0   PairsRatio: 6.02 [raw statistics]
10/02/22	master vs Stockfish 14 Update architecture to "SFNNv4" [differences] 166	Elo: 34.88 ±1.0 nElo: 100.85 ±2.5   PairsRatio: 7.44 [raw statistics]	Elo: 24.37 ±1.0 nElo: 83.29 ±2.9   PairsRatio: 7.51 [raw statistics]
17/02/22	master vs Stockfish 14 Tune search at very long time control [differences] 168	Elo: 33.07 ±1.0 nElo: 95.49 ±2.5   PairsRatio: 6.47 [raw statistics]	Elo: 23.83 ±1.0 nElo: 82.31 ±2.9   PairsRatio: 7.46 [raw statistics]
19/03/22	master vs Stockfish 14 Remove ttPv tree shrinking [differences] 180	Elo: 35.01 ±1.0 nElo: 102.62 ±2.5   PairsRatio: 8.16 [raw statistics]
17/04/22	master vs Stockfish 14 Decrease LMR at PV nodes with low depth [differences] 189	Elo: 36.69 ±1.0 nElo: 107.69 ±2.4   PairsRatio: 9.49 [raw statistics]	Elo: 26.09 ±1.0 nElo: 89.84 ±2.8   PairsRatio: 9.59 [raw statistics]
18/04/22	Stockfish 15 Official Release [differences] 190

---

External Links

There are several pages on the web run by chess engine fans. Some collect progress of Stockfish over previous stable versions and development builds. Some compare Stockfish progress to other chess engines. Here is a collection of some useful links in that regard.

• Computer Chess Rating Lists (CCRL)
• FastGMs Rating Lists (FGRL)
• Ipman Chess
• Mate Finding Effectiveness
• Next Chess Move (NCM)
• Stefan Pohl Computer Chess (SPCC)

---