FRITZ 20: The "Training Revolution" Claim and the Data Gap of a Chess Engine

In 2026, while editing a small room at Hai Phong Radio and Television Station to assemble the 400m hurdles stride rhythm of Quach Thi Lan into a three-chapter film, I had a peculiar habit: I opened a chessboard on my computer in parallel, reconstructing each position from the third game of that year's national championship. A male colleague looked over my shoulder and said something I still remember verbatim: "All software is the same, just buy the latest version and you're done." I trust footage more than testimony, because footage doesn't lie. But software is different from footage. Software can lie, or at least, software can say things that cannot be verified. When ChessBase issued its claims about FRITZ 20, I read it like I read a match report with no extra time: words present, numbers absent.

Context: a sport that runs on faith in tools

Chess is the only sport where the athlete trains against an opponent who is not present at the board. That opponent is the machine. No net, no track, no tangible finish line — just the glow of a screen and an evaluation in centipawn units. Because of this, the question "which engine to use" in Vietnam is not merely a technical question; it is a question of budget, of the training ecosystem, and of a sports culture long accustomed to waiting for sponsorship.

ChessBase is the name tied to chess software history since the 1980s. Fritz, the engine bearing the name of the cheerful bald character on the old interface, was once an opponent capable of defeating Garry Kasparov in the 2026 match in Bonn — a milestone the chess world still cites the way it cites Bob Beamon's 8.95-meter long jump in Mexico 2026. But history does not defend sales figures. The modern chess market has three overlapping layers of operation. The first is pure engine strength, where Stockfish — open source, free, running on every platform from phones to supercomputers — has taken an unarguable position. The second is neural-network engines like Leela Chess Zero, which can play more "humanly," useful for studying style. The third layer — where ChessBase is trying to position FRITZ 20 — is the integrated platform, turning analysis, opening training, and tactical practice into a seamless experience.

For Vietnamese players, this structure has very concrete meaning. A child at a chess club in Hanoi or Da Nang can open Lichess for free and analyze their game with Stockfish 16 in two seconds. A coach in Vung Tau can use Chess.com on a membership of a few hundred thousand dong per month to assign tactical exercises to the entire team. So what is FRITZ 20 selling, when the buyer already has free options? ChessBase's answer is "a training revolution." But revolution is a heavy word. And in sports, heavy words usually conceal a data gap.

Core analysis: reading a press release like reading a match missing stats

When I made the SEA Games 29 film, I had one iron rule: every emotional segment must be anchored by a measurable number. The final stride rhythm was 47 per minute in the decisive lap — without that number, the segment would just be sad music. Applying the same principle to the FRITZ 20 release, the picture becomes very clear.

The release describes FRITZ 20 as a "personal chess trainer," a "training revolution," "efficient, intelligent, individualized training," and targets both "ambitious players and professionals" and "players taking their first steps." These are four positioning claims, and all four come with no numbers. No ACPL (Average Centipawn Loss) figure compared to a reference engine. No calculation speed on specific hardware. No move-match rate against a top engine. No feature list, no system requirements, no price, no release date. Even the product name — FRITZ 20 — does not indicate whether this is the 2026 version or an entirely new product in a Fritz line that has existed for a long time before it.

A "training revolution" product with no number to measure the revolution means the only thing that can be measured is the gap between claim and evidence. This is not a judgment of the product. This is a judgment of the informational quality about the product. For a screenwriter, that difference is not small. I can write a compelling film segment about an athlete who has never won a medal, but I cannot call her an Olympic champion if her score sheet does not include her name.

What is more notable is the positioning contradiction. "Players taking their first steps" and "professional players" are two user groups with nearly opposite needs. Beginners need to learn rules, get used to the board, practice one-move tactics. Professionals need deep analysis of 30-move opening lines, endgame verification with 7-man tablebases, evaluation of an entire competitive cycle. In the chess training software market, almost no product does both well at once, because the interface and interaction model pull toward two opposite poles. When a release says "both," I usually read it as "not clear which one."

If we place FRITZ 20 beside competitors, the gap becomes clearer.

Stockfish does not advertise, does not sell, does not need positioning — it simply is the strongest engine by every measurable ranking, and it is available free across dozens of platforms. Leela Chess Zero does not sell strength; it sells a different angle: playing like a human, so that the learner can learn something beyond "the best move is always the best move." Chess.com sells an ecosystem: online tournaments, an enormous exercise bank, community, coaches, and even an AI suggestion line improving month by month. Lichess sells radical freedom: a non-profit project used by amateurs and a substantial slice of professionals as a daily analysis tool.

FRITZ 20, per the release, wants to wedge itself among all of this with the phrase "training revolution." But revolution at the elite level of sport has a very specific characteristic: it changes the measurement. The Fosbury flop changed how high jump was measured. Polyurethane swimsuits were partly banned because they turned world records into meaningless columns of numbers for two years. The big-data training revolution in football at Brentford changed how scouting was done. A chess training revolution, if it truly happens, must be measured by something: learners' rate of progress, quality of decision-making in time trouble, rate of recovery after mistakes. None of these are referenced by any metric in the release.

Of course, there is a fairer reading of ChessBase. In a market saturated with free engines, the Fritz name — once the face of machine strength — is no longer a competitive advantage in playing power. If it still sold an engine as an engine, ChessBase would lose to Stockfish in round one. Pivoting to "training" is a strategically sound move: sell experience rather than strength. The problem is in the communication execution. Saying "revolution" without evidence is not like a sacrifice move; it is like a post-match statement without footage to cross-check. For someone who has spent a life cross-checking material, that is hard to ignore.

Placed in the Vietnamese context, this story has another layer. Chess clubs in Hanoi, Ho Chi Minh City, Da Nang, Can Tho — what tools are they teaching children with? Most use free Lichess, some use Chess.com because the Vietnamese interface is more accessible, a smaller number use ChessBase for deep analysis at master level. In that picture, a new product from Europe with no published price, no release date, no system requirements, struggles to find a place, unless it creates a new need that students did not know they had. That is why marketing language matters — and that is also why marketing language easily disappoints when it comes with nothing verifiable.