III Ultrabullet IJCCRL Cup 2026

III Ultrabullet IJCCRL Cup 2026 Stockfish derived chess engines

III Ultrabullet IJCCRL Cup 2026 — Stockfish Derived Engines

The III Ultrabullet IJCCRL Cup 2026 returns as IJCCRL’s dedicated showcase for Stockfish-derived UCI engines under ultrabullet conditions, where stability, instant tactical accuracy, and time management are tested at the highest intensity.

This season features a carefully selected field of 13 engines, including established contenders from previous IJCCRL cycles and new challengers aimed at proving their strength in the most demanding time controls. As with all IJCCRL events, the objective is to provide a transparent, repeatable competition environment and a clear competitive snapshot for the community.

Participants (Stockfish derived):
Revolution-4.0, Dark SisTer 11_sse41, DON-SSE41, Artemis 23TR_PrO_sse41, HypnoS 1.01 popcnt, Wordfish-3.60-181225-sse41popcnt, Brainlearn31-x86-64-sse41-popcnt, Killfish PB 151025_sse41, Stockfish 17.1, Deepalienist-2.70-301225-sse41popcnt, SpecTral 12_sse41, ShashChess40-x86-64-sse41-popcnt, RapTora 6.0_ssse3.

Follow the standings and results on this page as the event progresses. Updates are published in the IJCCRL format, maintaining consistency with the rest of our rating lists and tournament coverage.

Why 10 seconds + 0.1 increment?

The selected time control, 10s + 0.1s, is a deliberate methodological choice. In contrast to rating lists and tournaments that predominantly emphasise longer time controls (where deep calculation and late-game conversion dominate), ultrabullet testing highlights a different profile of engine performance:

• Search efficiency and move ordering become critical, as engines must rapidly identify strong candidate moves.
• Evaluation stability matters more, as large swings in evaluation can lead to practical errors when time is scarce.
• Time management behaviour is exposed under pressure, helping distinguish engines that remain consistent from those that collapse tactically or strategically at speed.
• Practical robustness is tested in a setting closer to “continuous stress,” where the engine must deliver playable, reliable chess repeatedly.

The result is a rating list that is not intended to replicate other well-known leaderboards, but instead to provide additional, complementary information. In short: it is a different lens through which to measure strength—one that is highly relevant to modern engine development, regression testing, and performance benchmarking.

Equal conditions for all engines: fairness enforced by the launcher

A defining pillar of IJCCRL testing is that the rules are not merely stated—they are implemented and enforced. The tournament launcher is configured so that every engine competes under the same conditions, eliminating bias and ensuring comparability.

Key principles include:

• Identical parameters for all engines (same framework, same match logic, same adjudication philosophy where applicable).
• A controlled and transparent UHO opening suite that guarantees variety, discourages repetition, and reduces preparation bias.
• OwnBook = false, preventing any engine from gaining an advantage via private opening books or curated opening knowledge outside the defined suite.
• Consistent run conditions (threads, hash, syzygy/adjudication behaviour if enabled, and any other relevant options), applied equally.

This is a core difference between casual engine events and a scientific-style testing environment: the method is repeatable, documented, and controlled.

Statistical depth: hundreds of thousands of games across 2026

Because the Cup runs throughout the year, the tournament is expected to generate hundreds of thousands of games. This scale is essential for:

• Reducing noise and short-term variance
• Producing more reliable Elo separation between closely matched engines
• Building confidence intervals that make results more interpretable
• Identifying long-term patterns in draw rates, decisive outcomes, and stability

In practical terms, IJCCRL aims to produce a dataset where rating differences represent more than momentary form. The goal is not a snapshot, but a robust measurement.

Mission and conclusion: computer chess dissemination through a scientific method

The II Ultrabullet IJCCRL Cup 2026 is part of a broader IJCCRL mission: to promote and disseminate computer chess using a method that resembles scientific experimentation—controlled conditions, reproducible settings, transparent rules, and large samples.

The principles are intentionally simple and clear:

• same opening framework (UHO)
• same configuration for every engine
• OwnBook disabled
• long-term volume for statistical credibility

By combining transparency with scale, IJCCRL seeks to provide a serious reference point for the community—one that complements existing rating approaches and offers a specialised, modern benchmark for Stockfish-derived engines.