FRITZ 20, ChessBase, and the Gap Between Marketing Language and Verifiable Data

This December, a press release about FRITZ 20 landed in my inbox at eleven o'clock at night, Shenzhen time. Outside my window the city was still lit like a giant chessboard no one had cleared. Inside, I read a six-line paragraph three times, a paragraph made entirely of adjectives: "training revolution," "personalized," "intelligent," "efficient." I folded the laptop, poured a cup of tea, and opened a blank file. The first thing I do with any piece of advertising is not decide whether to believe it. It is to list what can be verified. That night, the list was empty. Not one specification. Not one comparison. Not one release date.

I mention this detail because it is the common denominator of nearly every chess-software release I have received in seven years. And because it opens a question far larger than FRITZ 20: when the chess-tool market has entered a phase where engine strength is no longer a competitive advantage, what exactly are manufacturers selling — and what exactly are players buying?

Context: from the software box to the synchronized dance of platforms

To understand why a press release like this gets written, we need to step back. ChessBase, the German company founded in 2026 by Frederic Friedel and Matthias Wüllenweber, was a near-monopoly in chess software for two decades. When I began working with chess data, the Fritz box was what every serious player had to have installed. To my generation, it was the standard. Nobody analyzed a game without opening Fritz.

Then three storms hit at once, and each struck the boxed-software model.

The first was named Stockfish. An open-source, free engine that, from roughly 2026 onward, became stronger than every commercial engine on the market. When a free tool consistently tops the engine rankings, the entire logic of "pay for stronger calculation" collapses. You are no longer buying strength. You are buying something else.

The second was named Leela Chess Zero, also free, a neural-network engine that introduced a more human-like evaluation. Many professionals found in it a different lens from traditional engines. Its arrival proved that even at the computational-strength layer, innovation keeps coming from the open community, not from boxed software.

The third, and perhaps the most frightening for a company selling a box, was integrated platforms. Chess.com, Lichess, Chessable. They do not sell a tool. They sell an ecosystem: play, analysis, opening study, tactics training, progress tracking, all in one browser tab, on a monthly subscription. A boxed-software company must compete with an ecosystem the user pays ten dollars a month for and never has to install.

I call this the transfer market of tools. There, value no longer lies in raw strength but in adaptability. An engine that is strong but that nobody opens is like a striker who scores twenty-two goals while the whole team depends on him. Over the long run, the system that absorbs him survives.

Core: reading a press release as a data table with missing columns

When I opened the FRITZ 20 announcement, I read it with the same eye I once used to dissect the xG of a Brazilian striker in the Chinese Super League. Not to find what was wrong, but to find what was measurable.

I found four claims, and not one of them came with numbers.

The first claim: FRITZ 20 is "a chess engine." True, but meaningless — every engine is an engine. What matters is how strong, measured by what. The industry benchmark is ACPL, average centipawn loss per move, and the match rate against the engine's top recommendation. The release offers not a single figure. An engine that does not publish its ACPL is like a player with no match data: all you have is self-declaration.

The second claim: it is "more than an engine — a training revolution." I filed "training revolution" under marketing language. A revolution is a claim about method, and method must have a structure. Without a feature list, an algorithm, or a description of how the system evaluates a game and prescribes exercises, it is a metaphor, not a product.

The third and fourth claims are where one should linger longest. The product is advertised for "ambitious players and professionals," and also for "players taking their first steps." Read that slowly. These two groups have nearly opposite needs. A beginner needs a guided path, concept explanations, basic error correction. A professional needs deep analysis tools, the ability to dissect variations, a vast opening library, and an engine strong enough to challenge their own intuition. A product that wants to serve both usually means it was not truly designed for either.

In transfer data, I once named this phenomenon the sign of the empty all-purpose box. When someone advertises a player who can play every position, it is usually because he excels at none. Mature markets always move toward specialization. Stockfish owns raw engine strength. Chessable owns structured learning. Chess.com owns the integrated experience. When a product will not choose a home for itself, the market will choose one for it — and usually not in its favor.

But wait. I must apply exactly the discipline I apply to myself: treat my own data as harshly as a rival's. A press release's lack of numbers says nothing yet about product quality. It says something only about the quality of the press release. This is the line many analysts cross wrongly. They turn marketing's silence into a verdict on the product. I do not. I say only this: until independent testing exists, every FRITZ 20 claim sits in the unverified drawer — and that drawer grows thicker with every year I work.

There is something subtler inside the structure of the release: ChessBase's choice not to mention ACPL. For a company whose product history is bound to engine strength, that silence is itself a signal. When a brand stops advertising what it was once strongest at, it is often because it is no longer strongest at it. Silence in data must also be fed into the model, like an empty cell in a transfer table.

Placing FRITZ 20 in the comparison table the release dared not build

One way to test positioning quality is to build the comparison table the manufacturer did not. I do this with every player file before a decision. Here, the table has four columns: engine strength, training features, audience breadth, and price.

Engine strength: undisclosed. The main rival, Stockfish, has already passed every threshold a commercial engine once reached. With no figures, no score is possible. This is the largest empty cell.

Training features: asserted but unspecified. The rivals are integrated platforms where every feature can be trialed before payment. With no feature list, no comparison is possible.

Audience breadth: too broad. This is a weakness, because it dilutes specialized value. The rivals each chose a niche and took it.

Price: undisclosed. Competitiveness cannot be judged.

Four columns, three completely empty, one showing weakness. That is everything an independent analyst can extract from this release. And that is what is most noteworthy: a company with a long tradition chose to announce a product using language that cannot be verified, in an era when users can verify almost everything themselves within minutes.

I once witnessed something similar in a Chinese Super League transfer. A club announced a blockbuster signing with an abundance of flashy language, but when I dissected the contract structure, I found most of its value in performance clauses that were unlikely ever to trigger. The announcement was beautiful. The structure was hollow. A Chinese club taught me that data is not the destination, but a walking stick. You use it to walk steadily, not to replace your own legs. But when someone deliberately hides the stick, the reader must know what ground they are standing on.

The contrarian angle: marketing is also a kind of data

This is the part I love most in any analysis — the part where I question my own prettiest assumption.

My assumption: FRITZ 20 is vaguely positioned, and that vagueness is a weakness. But let us flip the board. What if the vagueness is not a weakness but a carefully calculated strategy?

There is a very specific group of users the language of "training revolution" targets: those unfamiliar with the free alternatives. A fan attached to the Fritz brand since the CD-box era, someone looking to relive the feel of desktop software, a player who grips the mouse and wants to believe in a name. For that group, "revolution" needs no numbers to sell. It needs only belief. And here is the crux: vagueness in product positioning is often designed to serve the loyal group, not to persuade the informed one.

This is when I remember the lesson from a Chinese club in 2026. When I presented data and proposed changing their attacking system, not everyone wanted to listen. Data does not lie — but how people receive data reflects what they wish to believe more than what is proven. It took me time to understand: when data does not lie, we ourselves are the ones deceiving ourselves. With FRITZ 20, both the writer of the release and its intended reader share one belief — and that belief sells better than numbers.

Flip once more. Is there a chance ChessBase's silence is because the product is not ready, and this release is only a prelude? Entirely possible. In software, feature-blurring announcements appear thickly before a real launch. But again, I have no release date to place beside it. Without a date, every judgment about recency is a guess. A product bearing the number 20 could be this year's release or an old item warmed over. For a data worker, this is a dark zone that no intuition can fill.

Three flips, and I still reach only one defensible conclusion: I do not know what FRITZ 20 is. And honesty about not knowing is the most valuable asset of a data worker. After 2026, I no longer believe in predictions. I believe only in early-warning systems.

Long-term valuation: the life cycle, not the moment

A common mistake in analyzing chess products is to judge them within a single season. In reality, the lifespan of a training tool lies across many years. Users do not leave after one update. They leave after a platform changes their habits.

Apply the long-term valuation principle I use for the transfer market. A contract is valuable only if the player stays and adapts for two or three seasons. A training tool is the same. The right question is not "is FRITZ 20 strong," but "three years from now, will users still open it every day?" And the answer lies in adaptability, not peak strength.

I once built a small valuation model based on a cultural-adaptation index — originally to filter Brazilian players moving to new leagues. A player who had played in Portugal had an integration rate about forty-two percent higher than one who had never touched a European environment. It sounds foreign to chess, but the logic is identical. It does not measure talent. It measures the prior collision between a person and a new system.

Applied here: a user already used to the Chess.com workflow — where everything runs in the browser, all data syncs, all exercises sit in one timeline — will find it hard to return to desktop-installed software. Their adaptation index, in that sense, runs against FRITZ 20. This is the project I began in 2026, when COVID wiped out the calendar and left empty stadiums. COVID did not destroy football; it exposed who was living on illusion. I sat at home, analyzed ten years of Premier League transfer data, and realized that what decides success is not peak talent but the fit between a person and a system.

With FRITZ 20, I measure three columns. First: the cost of changing habits. Second: network value — where your friends play, where your community is. Third: engine update speed versus free versions. In all three columns, a traditional boxed product starts behind the line. That does not mean it will lose. But it means that to win, it must win by something else — and this release does not tell me what that something else is.

A beautiful chart is no substitute for a correct process

It once took me three months to understand that a beautiful chart is no substitute for a correct process. The lesson came from the 2026 World Cup, when I predicted Germany would defend their title based on possession and pass-completion data from qualifying. They were eliminated in the group stage. The numbers I clung to completely missed pressure-conversion and the speed of wide attacks. I spent three weeks rewatching all forty-eight group-stage matches, recalculating field tilt and high turnovers by hand, and building my own data table for the weaker teams.

What I drew from it was not "don't trust numbers." It was: numbers only have value when the process that produces them is trustworthy. A release that says "revolution" without a process behind it is like a beautiful chart with no source-data column. You cannot check it, cannot reproduce it, cannot use it to decide. And a claim that cannot be checked is not information — it is the signal of a different intention.

In my trade, people confuse information with signals. Information lets you act. Signals let you prepare. The FRITZ 20 release gives me no information to evaluate the product. It gives me a signal: that ChessBase is under pressure to reposition itself, that they know engine strength is no longer the battlefield, and that they chose to sell belief instead of numbers. Data is a mirror; but only those who dare face themselves see truly.

An open ending: when free tools become the default

What gives the FRITZ 20 story meaning beyond a single product is that it exposes a structural shift across the whole chess industry. For nearly two decades, engine strength was the competitive weapon. Now it has become public infrastructure — free, stronger than anything ever sold, and present on every free platform from Lichess to open tools. When something that was once an advantage becomes infrastructure, the entire business model of those who sell it must change.

I once wrote that the transfer market is not a chess game but a synchronized dance of thousands of algorithms. The chess-tool market is the same. No lone competitor can win on raw strength anymore. The winner is whoever reads the pace of changing user habits faster than rivals, and accepts that what they sell is not strength but fluency.

At the same time, this story poses a problem for people in media like me. When a marketing release is written in unverifiable language, the responsibility to translate it into useful information belongs to the reporter, not the seller. If I republish the advertisement without dissecting it, I become a link in the chain of illusion. If I dissect it and show what is empty, I hand readers a stick to walk with themselves.

With FRITZ 20, I will not settle the bet while the game is unfinished. I will wait for independent data. But if forced to pose one question for the near future, I would ask this: can a brand bound to memory survive by selling memory, in a market where new users have no memory to sell? The answer will not lie in a press release. It will lie in the login logs of millions of accounts, three years from now, when the advertising lights have gone out and only what people actually open each morning remains.