ChessBase, FRITZ 20, and the Data Gap Behind a 'Training Revolution'

The second-floor room of an old publishing house in Hanoi, a December night in 2026. A wooden chessboard, a desktop computer humming, eleven people sitting in silence. On the screen was the sixth game of Deep Fritz versus Vladimir Kramnik in Bonn. When the final move was played, Kramnik buried his head in his hands. The final score was 4-2 in the machine's favor. I wrote one line in my notebook: "The machine beat the human, and nobody in the room said a word for two minutes."

Nineteen years later, I read a press release for a new product. ChessBase, the German company founded in 2026 and the maker of the Fritz line, had just unveiled FRITZ 20. They called it a "training revolution," a "personal chess trainer," aimed at both beginners taking their first steps and professional players. I read it twice and flipped back to the first page. No technical specifications, no ACPL rating, no move-match percentage against an engine, no system requirements, no price, no release date. Only adjectives. I trust footage over testimony, because footage cannot lie. A press release with no data is also telling me a story — just a different one from the story they wanted to tell.

ChessBase is no stranger to anyone who has analyzed chess seriously. From the late 1980s, they were the first to commercialize a game database for personal computers, turning millions of historical games into something searchable in seconds. Their Fritz line became the face of machine strength in the memory of an entire generation of players. In 2026, their X3D Fritz drew 2-2 with Garry Kasparov in a New York match. Three years later, Deep Fritz defeated Kramnik. At that time, a strong chess engine was a technical marvel, and people were willing to pay to own one.

Then Stockfish appeared. Open source, free, running on almost any machine, and stronger than most commercial engines from around 2026 onward. Leela Chess Zero followed with a neural network, also free. Lichess embedded Stockfish in the browser and charged nothing. Chess.com bundled analysis, training, tournaments and community into a subscription, then in 2026 absorbed the entire Play Magnus Group, Magnus Carlsen's company. By 2026, analyzing a game at the level Kramnik once had to fight against is no longer a privilege for those who can pay. It is a browser tab.

That is the arena FRITZ 20 is stepping into.

Based on my experience following matches and the preparation of Vietnamese players over many years, I have come to see that chess players do not lack tools. They lack metrics. When a young player at a club in Hanoi or Ho Chi Minh City asks me "where did I go wrong," the useful answer is not to open an engine and point to move twenty-three. The useful answer is to tell them how many times that mistake recurred over three months, in which type of position, and whether the trend is rising or falling.

In chess analysis, those measures already exist and are standardized. ACPL — Average Centipawn Loss — measures the average error per move against the engine's best recommendation. Move-match percentage is another metric, widely used to assess the quality of a game. These are figures that can be cross-checked, compared across players, across career phases, across generations.

The FRITZ 20 release mentions none of them. It says the product is for "ambitious players and professionals," and at the same time for "players taking their first steps." One product, both at once.

I have made documentaries about track and field and swimming athletes. When someone says "I improved my performance," the first thing I do is ask for the split times. When a coach says "the players' fitness is better," I ask for the GPS data from training. Not because I doubt them. But because testimony can be true, while only data tells you where it is true and by how much.

I rewind the SEA Games 29 footage to find what people left behind. This time I have no tape to rewind. I have one product page, a few loud adjectives, and a question hanging in the air: what makes a company that once defined the industry's standard choose to speak in emotion rather than in data?

There may be technical reasons for that silence. Some vendors do not publish engine strength figures for fear of being compared directly against free rivals. It is also possible they believe the average buyer does not care about ACPL, and they may be right. But the gap between a claim and a spec sheet is still a gap — and it often says more than the content that was written.

At this point I have to apply the same standard to myself. I have no evidence that ChessBase is weakening. I only have evidence that their release lacks data. Those are different things, and conflating them would be committing exactly the error I just criticized.

What I can say with confidence is this: the "for everyone" approach of FRITZ 20 runs against the segmentation trend of the chess software market over the past decade. Stockfish wins at pure engine strength. Chessable wins in structured training content. Chess.com wins in integrated experience. None of them wins by claiming to serve every audience at once.

And here is where I want to push back on the chess community's own reflex. Many will say free engines have "democratized" chess. I do not entirely believe it. Distributing tools is not the same as distributing skill. Stockfish sits in everyone's hands, but the gap between those who use it systematically and those who merely open it to see the best move keeps widening, not narrowing. What is missing is not a stronger engine. What is missing is a tool that turns thousands of raw evaluation lines into a training program with defined inputs, measurable outputs, and demonstrable effectiveness over a set period.

FRITZ 20 is waving toward that gap. The problem is it has not yet proven it can fill it.

I ran hurdles my whole life, I was just never timed. It is the same in chess: there are quiet improvements nobody measures, and loud claims nobody checks. Someone in my trade learns one thing after nearly thirty years with a pen: what deserves trust is not what is shouted loudest, but what leaves a trace you can rewind. A good chess training program, if it is truly good, will eventually leave that trace — in exactly the metrics it refuses to publish today.

For Vietnamese players, the lesson may lie on the other side. We do not lack talent — Lê Quang Liêm and Nguyễn Ngọc Trường Sơn proved that long ago. What we still lack is the habit of measuring ourselves consistently, instead of relying on feeling after every lost game. A new tool is only useful when its user knows how to ask the right question. And the right question, before any advertisement, is always: where is your data?