খালি ডেটার ফাঁদ: ফুটবল বিশ্লেষণে কৃত্রিম বুদ্ধিমত্তার 'শূন্য ইনপুট' সংকট

গত সপ্তাহে লন্ডনের একটি কফি হাউসে বসে আমি একটা জিনিস লক্ষ্য করলাম, যেটা আমাকে রাতের ঘুম কেড়ে নিয়েছে। একজন তরুণ ফুটবল সাংবাদিক তার ল্যাপটপে একটা বিশাল টেবিল দেখাচ্ছিলেন — সাতটি কলাম, প্রতিটিতে 'N/A' লেখা। টেবিলটার নাম ছিল 'ট্যাকটিক্যাল অ্যান্ড টেকনিক্যাল অ্যাসেসমেন্ট'। আমি জিজ্ঞেস করলাম, এটা কার বিশ্লেষণ? তিনি বললেন, কোনো এক দলের, কিন্তু কোন দল সেটা কেউ জানে না। এটাই আজকের ফুটবল-প্রযুক্তি জগতের সবচেয়ে বড় অস্বস্তিকর সত্য: আমরা বিশ্লেষণের যন্ত্র তৈরি করেছি, কিন্তু সেই যন্ত্রকে খাওয়ানোর ডেটা দিতে ভুলে গেছি।

The consensus across the football-tech industry right now is simple and, on the surface, entirely reasonable. Every major broadcaster, every club's analytics department, every betting syndicate has spent the last three years bolting large-language models onto their pipelines. The promise was seductive: feed in a match report, a post-match interview, a set of Opta event logs, and out comes a nine-dimensional tactical breakdown complete with xG models, wage-structure projections, and managerial sack-pressure indices. The vendor decks all say the same thing — automate Stage-1 deconstruction, then scale Stage-2 deep analysis. That is the mainstream consensus I want to dismantle today.

The core problem is not the model. The core problem is what happens when the pipeline's first stage returns nothing. I have spent twenty-nine years in this industry, and I have watched a very specific pathology repeat itself three times now: first with expected goals, then with automated scouting databases, and now with AI-driven narrative analysis. Every time, the same failure mode appears — when the input is empty, the system still produces output. It produces a beautiful, structurally perfect, cross-referenced template full of nulls. And because that template looks like an analysis, because it has tables and flags and a glossary, nobody upstream notices that the machine was fed a blank page.

What happened in the Stage-1 failure is the football-analytics equivalent of a set-piece routine that nobody checked. England scored twelve goals at the 2026 World Cup and nine of them came from dead balls. I flew to Nizhny Novgorod and watched five of them go in against Panama, and I wrote then that the set piece had stopped being a phase and had become a personality. But here is the part I should have flagged harder at the time: a set-piece routine only works if the ball actually reaches the box. If the delivery is empty, the entire choreography is theatre. That is exactly what a null Stage-1 output is — a beautifully rehearsed corner routine with nobody taking the kick.

Look at the financial dimension, because this is where the football industry has the least tolerance for fantasy. A transfer-market analysis without an identified club is not a neutral placeholder. In a live pipeline, that null template can be auto-filled downstream by a model that sees 'Broadcasting revenue: N/A' and guesses a number. I have watched a €222 million transfer get repriced by the market within one season because the consensus was wrong about what the fee actually represented — it was a broadcast-rights purchase disguised as a player purchase. The dangerous thing about an empty field is not that it is empty. The dangerous thing is that somebody downstream will fill it.

And it is not only money. The governance layer is equally exposed. Financial fair play and profitability-and-sustainability checks depend on named clubs with disclosed balance sheets. If a Stage-2 engine is fed an empty Stage-1, the compliance checklist comes back with every box marked 'N/A' — and a club official reading that output could reasonably conclude there is no compliance risk, when the truth is that no compliance test was ever run. In my 2026 empty-stadium piece I called home advantage permanently dead within seventy-two hours of the Bundesliga restart, and by early 2026 I had to walk that back on camera. The lesson I took from that failure was not 'be more careful.' The lesson was: a claim without an expiry date is a claim without a spine. The same rule applies to a null analysis. It needs to carry a hard label saying 'this is not a finding,' not a soft one implying 'this is a clean bill of health.'

Now let me be honest about where I could be wrong here, because the contrarian version of my own argument deserves a hearing. It is entirely possible that the industry is not actually deploying these empty pipelines at scale, and that the Stage-1 failure I was shown is the work of one careless engineer, not a systemic threat. Maybe the nine-dimension framework with all-null handling is exactly what a well-disciplined system is supposed to produce — a loud, structural red flag that stops the pipeline cold rather than letting it continue with invented content. In that reading, the 'empty analysis' is not a bug but the safety valve working precisely as designed. I will grant that. But the safety valve only matters if somebody is reading the pressure gauge. The vendor demos I have sat in on this year have all shown the success path, and none have shown the null path, which suggests the valve exists in code but not in the operational culture.

There is also a tactical counter-argument worth taking seriously. If the Stage-2 engine cannot determine even the analysis subject — whether it is looking at team tactics, an individual player, a coaching duel, or a single-match review — then it genuinely has no basis for any of the nine dimensions. Pressing intensity, possession share, pass completion, expected goals against: none of these mean anything without a subject. A human analyst in that position would do exactly what the engine did, which is refuse to guess. So perhaps the correct reading is that the model behaved more professionally than many of the humans who built the pipeline around it.

That is where I land, but with one amendment. The null template is correct as diagnosis. It is unacceptable as a deliverable, unless it is paired with a gate that physically blocks the next stage. In football terms: the assistant referee raising the flag is not the same thing as the match being abandoned. Somebody still has to walk over and make the decision. The flag tells you the offside happened. It does not tell you what the score is.

So here is my testable call, with a date on it. By the end of the 2026-27 season, I expect at least one major European club or broadcaster to be publicly embarrassed by an AI-generated analysis that contained fabricated squad financials traced back to exactly this kind of empty-input propagation. Watch the compliance filings and the transfer-rumour cycles — not the press releases. If the pipeline cannot prove it ingested a non-empty source before Stage-2 fired, the entire output is theatre, no matter how many tables it produced. The teams that fix this first will not look smarter in their highlight reels. They will look smarter in the wage bill, and that is the only scoreboard that has ever mattered in this business.