elo_estimates.md

Elo Estimates

Search

All tests done at 8+0.08 1.3mnps on UHO_Lichess_4852_v1.epd
All features tested against commit a049bb98 or functionally equivalent commits except for correction history, singular extensions and internal iterative reductions

feature indicates that the estimate was based off of a gainer test because I'm too lazy to run a test
feature indicates that the estimate is SSS because I'm too lazy to keep running the test

In "proper" order

• Aspiration windows: ~108 elo https://mcthouacbb.pythonanywhere.com/test/358/
• TT Cutoffs: ~101 elo https://mcthouacbb.pythonanywhere.com/test/360/
• Internal iterative reductions: ~8 elo https://mcthouacbb.pythonanywhere.com/test/460/
• Improving: ~31 elo https://mcthouacbb.pythonanywhere.com/test/381/
• TT Eval adjustment: ~8 elo https://mcthouacbb.pythonanywhere.com/test/363/
• Correction history: ~91 elo https://mcthouacbb.pythonanywhere.com/test/318/
• Static Exchange Evaluation(overall): ~258 elo https://mcthouacbb.pythonanywhere.com/test/388
• Time Management: ~64 elo https://mcthouacbb.pythonanywhere.com/test/389/
• Quiescence search(overall): ~184 elo https://mcthouacbb.pythonanywhere.com/test/359/
  • QS pruning: ~55 elo https://mcthouacbb.pythonanywhere.com/test/382/
• Move ordering features
  • TT Move: ~106 elo https://mcthouacbb.pythonanywhere.com/test/364/
  • Killers: ~6 elo https://mcthouacbb.pythonanywhere.com/test/387/
  • History: ~527 elo https://mcthouacbb.pythonanywhere.com/test/375/
    • Quiet history: ~30 elo https://mcthouacbb.pythonanywhere.com/test/374/
    • Capture history: ~19 elo https://mcthouacbb.pythonanywhere.com/test/373/
    • Continuation history: ~40 elo https://mcthouacbb.pythonanywhere.com/test/372/
• Whole node pruning: ~228 elo https://mcthouacbb.pythonanywhere.com/test/367/
  • Razoring: ~6 elo https://mcthouacbb.pythonanywhere.com/test/205/
  • Null Move Pruning: ~31 elo https://mcthouacbb.pythonanywhere.com/test/370/
  • Reverse Futility Pruning: ~88 elo https://mcthouacbb.pythonanywhere.com/test/369/
  • Probcut: ~3 elo https://mcthouacbb.pythonanywhere.com/test/346/
• Move Loop Pruning: ~184 elo https://mcthouacbb.pythonanywhere.com/test/366/
  • Futility pruning: ~1 elo https://mcthouacbb.pythonanywhere.com/test/376/
  • Late move pruning: ~23 elo https://mcthouacbb.pythonanywhere.com/test/371/
  • SEE pruning: ~5 elo https://mcthouacbb.pythonanywhere.com/test/377/
  • History pruning: ~14 elo https://mcthouacbb.pythonanywhere.com/test/378/
• Extensions
  • Check extensions: ~13 elo https://mcthouacbb.pythonanywhere.com/test/392/
  • Singular extensions: ~73 elo https://mcthouacbb.pythonanywhere.com/test/350/
• Late move reductions: ~111 elo https://mcthouacbb.pythonanywhere.com/test/368/

In order of elo gain

• History: ~527 elo
• Static Exchange Evaluation(overall): ~258 elo
• Whole node pruning: ~228 elo
• Move Loop Pruning: ~184 elo
• Quiescence search(overall): ~184 elo
• Late move reductions: ~111 elo
• Aspiration windows: ~108 elo
• TT Move Ordering: ~106 elo
• TT Cutoffs: ~101 elo
• Correction history: ~91 elo
• Reverse Futility Pruning: ~88 elo
• Singular extensions: ~73 elo
• Time Management: ~64 elo
• QS pruning: ~55 elo
• Continuation history: ~40 elo
• Null Move Pruning: ~31 elo
• Improving: ~31 elo
• Quiet history: ~30 elo
• Late move pruning: ~23 elo
• Capture history: ~19 elo
• History pruning: ~14 elo
• Check extensions: ~13 elo
• Internal iterative reductions: ~8 elo
• TT Eval adjustment: ~8 elo
• Killers: ~6 elo
• Razoring: ~6 elo
• SEE pruning: ~5 elo
• Probcut: ~3 elo
• Futility pruning: ~1 elo