ChessBase and the August Release: When the Publisher Grades Its Own Products

On September 13, 2026, ChessBase published its roundup of new products for August. A tidy, readable catalogue: two courses on positional play, a family of updated databases for the Petroff Defence, plus a set of courses on calculation and strategic decision-making. I read the piece twice, then read it a third time with a pen in hand and a blank sheet of paper beside me.

On that sheet, I recorded every number that appeared in the article.

No games. No results. Not a single Elo rating mentioned.

For someone who works by reading data as I do, the most notable number is not in the product list. It is in the gap. A product roundup about chess, published on the publisher's own platform, and utterly devoid of any verifiable quality metric for what is being sold. Let us start there, because that gap says more than the catalogue itself.

To let readers follow the rest of this piece, I need to reconstruct a few strokes of context.

ChessBase is one of the oldest publishers of chess software and training content in the world. It sells opening databases, tactical courses, and themed training sets. Its direct rival in the training-content market is Chessable, a platform specialising in opening courses written by international masters. Behind those two names sits a broader ecosystem of open platforms, self-study communities, and free data repositories.

The business model here is subscription plus packaged retail. That is, to retain paying users, a publisher must release products regularly. The August roundup I am reading is one link in that wheel: a new product line every month, presented in the publisher's own voice. The monthly cadence is not a small detail. It is structure. It tells you this business measures success by continuous user engagement, not by a single peak product per year.

This is where a data worker must separate two different questions, and I stress it because many readers merge them into one. The first question: what does this product teach? The second: who is telling us it is good? The two have different answers, and that difference matters more than the entire product list. In my trade, confusing these two questions is the origin of most analytical errors.

The technical half of the release looks entirely reasonable, and I will state exactly what it contains, no more no less.

First is the updated Petroff Defence database. The Petroff is a system opening with the moves 1.e4 e5 2.Nf3 Nf6. It is a symmetrical defence, favoured by those who want to neutralise White's initiative at minimal theoretical risk. A refreshed Petroff database has concrete meaning: in a defence where small move-order refinements can decide who holds the initiative, theory must be continuously updated by analysis engines. A move displaced by one tempo at move twelve can turn a balance into a loss. But the roundup offers no evaluation and names no novelty. For the theory reader, that is a closed door.

Second are two courses on positional play, alongside courses on calculation and strategic decision-making. This is the skills-content group, different in nature from opening databases. If opening databases sell theory to those already at competitive level, skills courses sell transferable ability to a broad amateur audience. The three authors named — Bahne Fuhrmann, Andrew Martin and Kostya Kavutskiy — serve as content presenters. Andrew Martin is a veteran ChessBase author; Kostya Kavutskiy is a US-based international master active in commentary. Assembling so many names, as I read it, is a brand-diversification strategy: each author carries a following of his own.

Now to the most important part of this analysis. The structure of the catalogue reveals the publisher's market strategy, and it reveals more than the publisher wants to say aloud.

When a publisher sets an opening-database update beside a run of skills courses, it is serving two customer groups with different wallets and different needs in a single release. The first group is club players and juniors chasing theory at a high level. The second is the self-teaching mass, those who need to improve foundational skills such as calculation and evaluation. Two positional-play courses in a single release cycle are a signal of the second group's thickness — or a signal that the catalogue itself is becoming redundant. I leave both possibilities open, because the available data is not yet enough to pick one.

The pedagogical structure described is also worth noting. The courses combine theoretical explanation with annotated games and practical exercises. That is the standard trio of the modern training-content market: theory, illustration through real games, and exercises. It is not new, but it works, and it shows the publisher is aiming at applicability rather than scholarship. One more detail I always record: the content is segmented by playing strength. Commercially this is the right call, because it maximises the addressable market. Analytically, it tells me this release was designed to cover, not to go deep.

What I did not find is also worth recording. No engine evaluation, no move-quality metric, no database statistic to verify any quality claim. For a publisher that owns one of the strongest chess-data infrastructures in the world, that absence is not because the data is unavailable. It is simply because the roundup is a commercial document, not a technical one. Based on my experience tracking release cycles, this is a recurring pattern: the technical half is described in qualitative language, while the verification half is left empty.

The split between theory content and skills content also reflects a reality of the training market: money is made at both ends. The theory end serves the elite's opening race; the skills end serves the learning needs of the many. A publisher standing firm at both ends has solved the catalogue-balance problem. But that balance has its cost, and the cost usually does not appear in the roundup.

This is where I must say plainly something no publisher wants to hear.

ChessBase has just published a product roundup, written by ChessBase itself, about products ChessBase itself sells, with imagery credited by ChessBase itself. I am not accusing anyone here. This is an observation about source structure, and to someone who works by cross-checking data as I do, source structure is everything. People call it a shock; I call it unread data. But the reverse also holds: people call it a review, I call it a sales notice dressed as an article.

The tone is supportive. It asserts that the products deliver concrete opening advice and deeper training, a quality claim accompanied by no independent evidence. In my trade, an assertion without an independent source is worth only as much as the person making it. And the person making it is selling.

I am not saying the products are bad. I am saying we have no data to know whether they are good or bad. The transfer market is the one place where people pay for numbers no one has verified — and the training-content market is the same, except that here people pay for courses no third party has verified.

One more counter-intuitive point: the monthly release wheel may itself be squeezing content quality. When revenue depends on retaining paying users, the pressure of quantity usually beats the pressure of depth. This cycle's single opening update, with the rest being generic skills content, fits that hypothesis: the publisher is filling catalogue gaps rather than opening new theoretical territory.

And there is a data discrepancy within the input itself that I am obliged to flag. The earlier-stage note assigned the article a headline about chess historians meeting in Copenhagen, while the actual content is the August product roundup. The two cannot coexist. I work from the actual content; the other headline is a transmission error at an earlier stage. Numbers are asceticism: you must give up easy comfort to see the truth — and what I take from this is that even the data about an article needs checking before use.

So what is the signal for the next cycle? I will track three things.

First, the topic structure of the September release. If the weight of skills courses keeps rising against opening databases, that is a signal of where the money actually flows. Second, the fate of the Petroff update: whether it produces a notable novelty recognised by the theory community, or is merely a silent refresh. Third, and most important, whether any third party — an independent outlet or a user community — tests the quality of these courses.

If three months from now it is still only the publisher talking about its own products, we already know the answer. Not through a statement, but through the absence of data.