FRITZ 20: When a Chess Engine Has to Call Itself a 'Training Revolution'

In the entire promotional release for FRITZ 20 issued by ChessBase, the detail that should have sat on the first line is entirely absent: the release date.

No date, no engine version number, no performance comparison table, no specific feature list, no price. Only one phrase repeated like a mantra — "training revolution."

Pushed out into the AFC Cup 2026 corridor, I learned to read matches from what others leave behind. In analysis, what is left behind is often more important than what is put on display. A chess software product that says nothing about its release date, engine specifications or price is a signal, not an oversight. And a signal can always be decoded.

To understand that signal, FRITZ 20 must be placed in the context of the chess software market over recent years. That market has changed fundamentally over roughly the past fifteen years, and it has changed in a direction unfavourable to every commercial engine publisher.

Stockfish — open-source, free, the strongest engine in the world — sits in the hands of anyone with an internet connection. Leela Chess Zero — a neural-network engine, also open-source — adds an entirely different school of evaluation. Lichess offers unlimited free analysis. Chess.com integrates AI across its entire learning, playing and analysis ecosystem. Raw engine strength has been fully commoditized.

In that context, where does the value of a commercial engine lie? No longer in computational strength. It must lie in experience: interface, training methodology, learning pathways, community connection.

ChessBase — the publisher of Fritz — understands this. The FRITZ 20 release is evidence that they understand. But understanding and proving are two different things. During my fourteen years commentating on chess for VTC, I learned one principle: manufacturers never lie, but they always choose to say what benefits them. The reader's task is to find what they leave unsaid.

Analysing the FRITZ 20 text layer by layer reveals more than it claims.

The first layer is positioning. The release says the product is for "ambitious players and professionals" while also being for "players taking their first steps." That is a functional contradiction, not a broad-market strategy. A genuinely professional-grade tool — say a deep opening database or a Syzygy-tablebase endgame module — cannot simultaneously be the tool that teaches a beginner how pieces move.

Successful products in this market are specialized. Stockfish dominates raw engine strength. Chessable dominates structured learning. Chess.com dominates the integrated experience. Positioning "from zero to professional" cannot be defended against any specialized rival — you cannot be stronger than Stockfish in strength, nor better structured than Chessable in pedagogical depth.

The second layer is language. "Training revolution," "personal chess trainer," "efficient, intelligent, individualized training" — not one of these terms has an operational definition. A revolution compared to what? Individualized by which algorithm? Efficiency measured by which metric — ACPL, engine move-match rate, or rating improvement after a defined number of hours?

The evidence points one way: the text provides not a single verifiable number. No ACPL, no move-match rate, no benchmark. These are metrics any serious engine developer holds from the internal testing phase. Choosing not to publish them is a decision, not an omission.

The third layer is source conflict of interest. ChessBase is the publisher of FRITZ 20 and also the only source of information about FRITZ 20 in this text. When a source both sells a product and reviews it, every descriptive claim must be treated as marketing material, not independent editorial information.

I believe in the diagram, but I believe more in the space between diagrams. The real value of FRITZ 20 does not lie in what the release says, but in the space between its sentences — that space shows what the manufacturer chose not to say.

The counter-intuitive angle: the real story of this text is not FRITZ 20. The real story is ChessBase's position.

If your engine were genuinely stronger than or equal to Stockfish, you would publish head-to-head results. If your training feature set were genuinely groundbreaking, you would list the features and offer a free trial. Shifting the language from "engine" to "training revolution" is the sign of a strategic retreat from a field where you no longer hold an advantage.

Playing with five defenders is an admission that you are not good enough to defend with the ball. Positioning a commercial engine as a "training revolution" is an admission that you are not strong enough to compete on engine strength.

In parallel, the chess industry is moving from boxed-software sales to subscription and freemium models. This pressure forces traditional publishers to increase marketing spend to hold market share. A release with no data may be a sign that the marketing budget is growing faster than the product development budget.

The blind spot in execution lies here: the general reader, unfamiliar with Stockfish or Lichess, may accept "training revolution" as fact. That is precisely the purpose of marketing language — aimed not at the informed, but at those with no point of comparison.

FRITZ 20 may be a good product. But a release that names no date, no price, no specifications and no concrete features gives the reader no basis to believe it.

The person standing at the edge of the pitch sees the whole match, not only what is inside the goal. Here, what matters is not in the advertising lines — it is in the absence of any line about numbers. Before spending money, find an independent test. The question is not how strong FRITZ 20 is, but why its manufacturer chose not to say.