Repository navigation
What's new ShashChess 41
ShashChess 41 marks a major evolutionary step in the engine's development. This release aligns the core architecture with the powerful Stockfish 18 codebase while integrating Alexander Shashin's theoretical principles more deeply than ever before.
Beyond raw strength, Version 41 introduces specialized modes for tactical analysis, a completely overhauled evaluation output designed for human learning, and—crucially—official international validation of its originality.

ShashChess achieved a significant milestone by being accepted into the World Computer Chess Championship (WCCC) in Santiago.
Participation in this prestigious event required a rigorous code review conducted by the tournament's technical committee. This official audit definitively validated ShashChess as a legitimate derivative engine—distinct from a simple clone. This acknowledgment confirms the substantial originality of its codebase, distinguishing its unique strategic logic from the stock engine and settling any debate regarding its authenticity.
The engine has been updated to the latest Stockfish 18 standards, incorporating state-of-the-art optimizations:
- NUMA-Awareness: Improved management of Non-Uniform Memory Access for high-end multicore CPUs, utilizing system-wide shared memory logic.
- AVX-512 & SIMD: Updated vector instructions for modern processors, ensuring maximum nodes-per-second (NPS) efficiency.
- Search & SMP: Integration of the latest Lazy SMP and search heuristics from the official Stockfish development stream.
Unlike standard Stockfish releases, ShashChess 41 does not embed the Neural Networks (NNUE) inside the executable.
- Why? To drastically reduce the file size of the engine binaries and allow easier swapping of networks without recompiling.
-
How to use: Ensure the required
.nnuefiles (e.g.,nn-c288c895ea92.nnue) are present in the same folder as the ShashChess executable.
ShashChess 41 introduces a specialized compilation variant/mode called GoldDigger. While the standard version balances strength and match-play reliability, GoldDigger is designed for maximum tactical discovery.
- The Concept: GoldDigger sacrifices a small amount of match-play Elo (approx. -3 Elo vs SF18) to find deep, speculative, and brilliant moves that other engines prune away as "too risky".
- Tactical "Heat": It implements a "Paranoid Heat" heuristic. If a position is considered "Hot" (high mobility, sacrifices available, or king danger), the engine refuses to prune branches, forcing a deeper check.
Tests performed on 325 hard tactical positions.
| Engine | Solved Positions |
|---|---|
| Stockfish 18 | 183 |
| ShashChess 41 (Standard) | 200 |
| ShashChess 41 (GoldDigger) | 206 |
To support the Shashin theory of "Tal-like" dynamic play, the search algorithm has been refined with new concepts to better identify when to play aggressively.
-
The "NodeHot" Logic: A position is flagged as
NodeHotif it meets specific criteria: King Danger, Pawns near Promotion, or Sacrificial situations. - Effect: When a node is "Hot", the engine disables or restricts pruning techniques (like Null Move Pruning). It forces the engine to "look" rather than assume a position is quiet.
- "Sniper" Extension: A "True Radar" extension has been added. If the engine detects an Aggressive or Sacrificial style combined with King Danger at high depths, it extends the search depth for that specific move by roughly 25%. This allows ShashChess to find "knockout blows" that require deep calculation which standard engines often miss.
The eval command output has been completely overhauled to provide "Actionable Chess Intelligence".
- Shashin Zones: The engine now automatically classifies the position into zones based on Win Probability (e.g., High Tal for winning attacks, Capablanca for balanced strategy, Petrosian for defense).
- Makogonov Principle: The engine lists pieces ranked by "static activity" and explicitly identifies the "Worst Unit" that needs improvement, guiding the user on which piece to activate next.
- Legal Move Analysis: A sorted list of legal moves based on static activity and win probability, showing the human player the "practical" best moves, not just the mathematical ones.
- WDL Probability: Explicit display of Win/Draw/Loss percentages from the perspective of the White player for consistency.
- ShashinManager Optimization: The logic handling Shashin heuristics has been optimized with a bitmask cache system, preventing redundant calculations for Sacrificial/Fortress checks and improving speed.
-
Full Threats Feature: A new NNUE feature set
FullThreatshas been implemented to explicitly track complex threat patterns on the board.
This project and its professional validation would not have been possible without the support of:
Prof. Paolo Ciancarini (University of Bologna) for his academic guidance and contribution to the theoretical framework.
Eng. Massimo Venuto for his technical engineering contributions.
Intel Italia for the critical hardware support; validating a chess engine with this level of statistical confidence requires enterprise-grade computing power.
- The "Giants" met at the Santiago 2024 competition, who proved that the chess programming world is not just composed of sects of hypertrophic egos, but of genuine innovators and collaborators.
