When Stage-1 Returns Empty: Lessons on Transparency in Football Analysis

In early August 2026, a two-stage football analysis pipeline received an input containing only empty fields. Stage-1 — supposed to decode raw text into structured information points — returned no title, no source, no player list, no tactical data. Stage-2, rather than fabricating an analysis from nothing, output a report with dozens of fields all marked N/A alongside a clear warning: 'Insufficient information, cannot assess.'

I have been tracking the football analysis industry for eighteen years. This is a rare moment where I see a system self-constrain rather than hallucinate — generating content from void to fill gaps. And that is what deserves attention.

The two-stage analysis framework: How Stage-1 and Stage-2 operate

According to the structure described in this document, the pipeline begins with Stage-1: a deconstruction step aimed at transforming raw text into structured information points. These serve as the foundation for Stage-2's in-depth analysis across nine dimensions: tactical-technical, club finance, match results, league positioning, rules compliance, dressing-room analysis, risk profile, media expectations, and industry transmission chain.

This model appears rigorous at first glance. But the critical weakness lies in the dependency chain: if Stage-1 returns empty, all downstream analysis dimensions collapse in a domino effect. Similar to a VAR referee with no footage to review — the process remains, but without input data, no decision can be made.

In my actual tracking work, this is an issue I encounter frequently when working with reports from the Vietnamese market. Many articles have catchy headlines but lack basic parameters: xG for the team in the last three matches, wage structure, contract duration of key players. When I attempt analysis, I often have to gather additional data myself — or admit that the article lacks sufficient foundation for responsible conclusions.

Core issue: Empty input and the hallucination phenomenon

The most notable aspect of this Stage-2 document is not the N/A fields — but the clear, responsible null-handling mechanism. The system does not attempt to fill in the blanks with speculation. Instead, it marks each analysis dimension as 'insufficient information, cannot assess' with confidence level: Low.

This is something I consistently emphasize in young journalist training: numbers do not lie, but those who record them can. And when there are no numbers to verify, the right thing to do is stay silent rather than fabricate an engaging story.

However, I also recognize a contradiction within this pipeline itself. The 'Entities Involved' field — the list of relevant entities — not only remains empty but also self-references back to 'identify from the information points above'. This is a dependency cycle: field A needs field B, but field B points back to field A. When information points are empty, both fields have nothing to display. In refereeing discipline analysis, this is called a circular reference — and it prevents the entire system from recovering gracefully.

Contrarian angle: Why empty output is actually a good sign

Most readers would assume that an analysis system returning all N/A is a failure. I think the opposite. This signals that the system has self-audit and self-constraint mechanisms — two attributes the current football analysis industry severely lacks.

Take an example from the Vietnamese market itself. The 2026-24 season, a V-League club announced a foreign player signing with a reportedly 'record-breaking fee'. Immediately, numerous analysis pieces appeared about the player's 'tactical impact' and 'development roadmap'. But when I verified the original contract — through an internal source — the actual fee was only one-third of the announced figure. All those meticulously written analyses were based on incorrect data.

A system with strong null-handling would never fall into that situation. When input is insufficient, it stops and warns. No fill-in, no extrapolation, no 'creative interpretation'.

Looking more carefully, I see Stage-2 even provides clear follow-up actions for each null case: 'Re-run Stage-1', 'Fix entity extraction', 'Confirm source timestamp capture'. This is systematic working methodology — reflecting an important philosophy: analysis is not storytelling, but structured reasoning based on evidence.

Vietnamese context: Data deficiency and the habit of filling gaps

Vietnam's football analysis industry is in a rapid development phase but lacks foundation. While top European leagues have detailed statistical systems from Opta, Statsbomb, Second Spectrum, the Vietnamese market still relies heavily on aggregate data with low resolution.

The consequence is that when writing about V-League clubs, I often have to build datasets from scratch. I once spent an entire night cross-referencing footage of forty-seven foul situations in a single match — work that a good statistical system could provide in seconds.

The problem is not just technology. It is journalism culture. Many Vietnamese journalists — and this is my observation after years working in Madrid and tracking Southeast Asian markets — tend to write 'sufficient' articles rather than 'accurate' ones. An article lacking xG data gets replaced with subjective descriptions of 'impressive form' or 'controlling style'. Readers consume and believe — but have nothing to verify.

This Stage-1/Stage-2 pipeline, though imperfect, is attempting to establish an opposite standard: no information means no statement. This is a high standard — and it will not fit the current rapid content production model. But long-term, it is the right direction.

Framework limitations and hidden blind spots

However, I also see two blind spots in this very framework.

First, it has no mechanism to distinguish between 'no information available' and 'information being withheld'. In the context of Vietnamese club finances, this is a major issue. Many clubs do not disclose contract details not because they lack them — but because they choose not to. A pipeline that only detects empty fields will not differentiate these two cases, potentially issuing 'insufficient information' assessments when the reality is 'withheld information'.

Second, the nine-dimension structure, while comprehensive, has a risk of parallelism — dimensions processed independently without cross-referencing mechanisms. For example, Dimension 1 (tactics) and Dimension 6 (dressing room) might produce two contradictory pictures without a horizontal verification process. In actual matches, tactical failure usually reflects dressing room issues — and vice versa.

Lessons learned and development directions

Returning to the opening moment: an analysis pipeline receives empty input and returns all N/A with warnings. Should this be evaluated as success or failure?

My answer: this is a significant step forward. In an industry where output pressure often distorts analysis, having a system that dares to say 'insufficient information' is valuable. It establishes a new standard for what constitutes a valid analysis.

For the Vietnamese market, I recommend three development directions. First, invest in match data collection infrastructure — not just results, but PPDA, xG, pass maps. Second, build cross-verification culture within journalist teams — more sources, fewer assertions. Third, develop case studies on null-handling from major leagues to learn how systems self-constrain.

The season continues. Every week, I still sit down with footage — looking for moments when players stop at the right time, rather than just watching goals. That is how I avoid empty analysis.

As for this pipeline: it has passed its first test — not generating content from void. That is the foundation to build upon.