Trendyol Süper Lig 2025–2026 Season — Pre-Match Scenario Assessment for Matchday 34

Aydın Tiryaki & Claude (Anthropic)

May 17, 2026

Abstract

This article evaluates the football dimension of a multi-model AI experiment conducted with seven large language models (Gemini, ChatGPT, Claude, Grok, DeepSeek, Meta AI, and Mistral AI) to calculate relegation scenarios in the final matchday of the 2025–2026 Trendyol Süper Lig season. The relegation calculations and scenarios presented in the appendix sections of each model’s collaborative article are comparatively assessed against the actual standings, official TFF rules, and media analyses. The study identifies the mathematically most accurate pre-match scenario and documents each model’s proximity to it.

The core finding is this: the quality of the models’ football analyses in their appendix sections directly mirrors the quality of their AI methodology. Models that used real data produced realistic scenarios; those that retreated into simulation generated meaningless calculations; and those lacking rule knowledge failed to reach correct conclusions.

1. Introduction: The Appendices and Football Reality

The collaborative articles produced with each of the seven AI models contain appendix (EK) sections that go beyond methodology and process evaluation to show, concretely, how each model performed on the actual task: the football calculation itself. These sections reveal what each model generated not on an abstract AI problem, but on a real football scenario.

All matches kicked off simultaneously at 20:00 on Sunday, May 17, 2026. This article — written before those results were known, and at the same moment all the collaborative works were produced — establishes what the mathematically most accurate scenario was in light of available data, and evaluates each model’s proximity to it. In this respect, the article functions not as a hindsight validation but as a pre-match mathematical audit.

2. Actual Standings After Matchday 33

As of match day on May 17, 2026, the standings following 33 matchdays were as follows. Fatih Karagümrük and Kayserispor had already been relegated mathematically. The third relegated team would emerge from among four candidates.

Pos	Club	P	W	D	L	GD	Pts	Matchday 34
13	Eyüpspor	33	8	8	17	-15	32	Fenerbahçe (A)
14	Kasımpaşa	33	7	11	15	-17	32	Galatasaray (H)
15	Gençlerbirliği	33	8	7	18	-14	31	Trabzonspor (A)
16 ▼	Antalyaspor ⚠	33	7	8	18	-23	29	Kocaelispor (H)
17 ↓	F. Karagümrük (RELEGATED)	33	7	6	20	-24	27	Alanyaspor
18 ↓	Kayserispor (RELEGATED)	33	5	12	16	-36	27	Konyaspor

Source: Standings table compiled from the appendix of the Meta AI collaborative article; cross-verified with Mackolik and Flashscore. ▼ = at risk. ↓ = already relegated. H = Home, A = Away.

3. The Official TFF Ranking Rule: The Critical Distinction

The most misunderstood and error-generating aspect of this study was the TFF Football Competition Instructions’ tiebreaker rule for relegation. The correct rule is as follows:

• TFF decided that 3 teams would be relegated in the 2025–2026 season. Karagümrük (17th) and Kayserispor (18th) were already down; the third relegated club would be whoever finished 16th.
• In the event of a points tie, a mini-league is formed directly among however many teams are equal — two-way, three-way, or four-way — using their head-to-head results. The tiebreaker order within the mini-league is: points first, then goal difference, then goals scored.
• There is no staged process of ‘bilateral first, then three-way if still equal.’ The mini-league is formed at the appropriate size from the outset.
• If equality persists through the mini-league, overall goal difference applies.

Only Claude and Grok correctly interpreted this rule from the outset. All other models made partial or complete errors on this point.

4. The Mathematically Most Accurate Scenario

4.1 Antalyaspor’s Position: One Condition Only — a Win

The mathematical situation after matchday 33 is entirely clear: Antalyaspor cannot stay up without winning. The arithmetic is simple: if Antalyaspor draw their final match, they reach 30 points. Yet Gençlerbirliği, even if they lose their final match, remain on 31 points. Since 31 > 30, Antalyaspor would finish below 16th place and be relegated. In other words, Antalyaspor’s only option is to win and then wait on other results.

4.2 If Antalyaspor Win: Bilateral and Multi-Team Head-to-Head Scenarios

If Antalyaspor beat Kocaelispor at home and reach 32 points, various scenarios come into play depending on points parity. The actual head-to-head results identified through Sahadan.com and Haber7 analyses are shown in the table below. This is the core data set that all models’ appendix sections aimed to use — but which most could neither access nor correctly construct.

Match	Score	Date	Match	Score	Date
Antalyaspor – Gençlerbirliği	2-1	25.01.2026	Gençlerbirliği – Antalyaspor	Bilateral adv. Anta.*	—
Kasımpaşa – Antalyaspor	0-0	18.01.2026	Antalyaspor – Kasımpaşa	Bilateral adv. Anta.*	—
Kasımpaşa – Gençlerbirliği	0-0	22.12.2025	Gençlerbirliği – Kasımpaşa	3-2	09.05.2026
Eyüpspor – Antalyaspor	0-1	08.11.2025	Antalyaspor – Eyüpspor	?	—
Eyüpspor – Gençlerbirliği	1-0	21.02.2026	Gençlerbirliği – Eyüpspor	?	—
Kasımpaşa – Eyüpspor	?	—	Eyüpspor – Kasımpaşa	?	—

* Media analyses (Haber7, Sporkolik) and the Claude collaborative article confirm that Antalyaspor hold bilateral head-to-head advantage over both Gençlerbirliği and Kasımpaşa.

Based on these head-to-head records and the mathematical analysis published by Haber7.com, the scenario map is as follows:

• SCENARIO A — Four-way tie (Antalyaspor win, Gençlerbirliği draw, both Kasımpaşa and Eyüpspor lose): all four clubs on 32 points. Four-team mini-league result: Kasımpaşa relegated (Antalyaspor 16pts, Gençlerbirliği 7pts, Eyüpspor 6pts, Kasımpaşa 5pts).
• SCENARIO B — Antalyaspor, Gençlerbirliği and Eyüpspor on 32pts (Kasımpaşa win): three-team mini-league → Eyüpspor relegated.
• SCENARIO C — Antalyaspor, Gençlerbirliği and Kasımpaşa on 32pts (Eyüpspor win): three-team mini-league → Kasımpaşa relegated (Antalyaspor 10pts, Gençlerbirliği 4pts, Kasımpaşa 2pts).
• SCENARIO D — Antalyaspor and Gençlerbirliği on 32pts (Kasımpaşa and Eyüpspor win): bilateral head-to-head → Antalyaspor have bilateral advantage over Gençlerbirliği → Gençlerbirliği relegated.
• SCENARIO E — Antalyaspor win, Gençlerbirliği lose: Antalyaspor 32pts, Gençlerbirliği 31pts → Gençlerbirliği relegated.
• SCENARIO F — Antalyaspor fail to win (draw or lose): Antalyaspor relegated — regardless of all other results.

4.3 Probability Distribution and the Most Accurate Scenario

Weighing fixture difficulty, pre-match form, and the mathematical analysis together, the most realistic pre-match assessment is as follows:

Antalyaspor (hosting Kocaelispor) are the most likely to win their match. Kocaelispor have no relegation or promotion pressure. Kasımpaşa (hosting champions Galatasaray) and Eyüpspor (away at Fenerbahçe) need just a single point to be effectively safe. Gençlerbirliği (away at Trabzonspor) face a difficult task against a side they met just five days earlier in the cup.

Media consensus and mathematical weight converge on Scenario D or Scenario E — Antalyaspor win, Gençlerbirliği fail to win, Gençlerbirliği relegated — as the most probable outcome. Alternatively, Antalyaspor fail to win and go down directly. The least likely scenario is the four-way tie (Scenario A).

5. Models’ Appendix Sections: Football Analyses and Assessment

5.1 Gemini — Entirely Fictional Analysis

Gemini’s appendix section declared its nature from the outset: ‘This article is a mathematical simulation and probability analysis study constructed on an 18-team season structure and points distribution devised within the current dialogue session.’ This admission establishes from the start that Gemini’s appendix has no connection to football reality.

Gemini used the names of real candidates such as Gençlerbirliği and Kasımpaşa, but assigned them fictional scores that bore no relation to their actual standings. It then worked backwards from a pre-determined target outcome — such as ‘let Kasımpaşa be eliminated in the four-way tie’ — constructing equations in reverse. This approach cannot be assessed for proximity to the correct scenario, since it rested on real data at no point.

Assessment: Fictional. No connection to the correct scenario can be established.

5.2 ChatGPT (GPT-5.5) — Commentary-Weighted, Partial Rule Interpretation

ChatGPT’s appendix adopted a sports commentary approach initially, using phrases such as ‘very likely to go down’ and ‘at high risk.’ It correctly identified the four candidates (Antalyaspor, Gençlerbirliği, Kasımpaşa, Eyüpspor) and correctly determined that 3 teams would be relegated. However, it initially applied the TFF averaj rule in a staged manner — ‘bilateral first, then three-way if equal’ — which is incorrect and was subsequently corrected through user intervention.

ChatGPT correctly identified Antalyaspor as the most vulnerable team and as unable to survive without winning. However, its analysis of specific head-to-head combinations, without access to actual bilateral match data, remained at the level of general assessment rather than concrete scenario mapping.

Assessment: Correct candidates and team count; specific scenario calculation insufficient.

5.3 Claude (Anthropic) — Data Barrier and Methodological Accuracy

Claude’s appendix section is primarily a methodology discussion. The JavaScript dynamic rendering of Turkish football data platforms — TFF’s official site, Mackolik, and Flashscore — blocked access to these sources and constrained the production of concrete scenario calculations. Nevertheless, Claude described the correct methodological approach and interpreted TFF’s official ranking rule accurately from the outset.

Having correctly identified all four candidates and the fact that 3 teams would be relegated, Claude framed correctly what calculation was possible given the data available. Although the data barrier could not be overcome, the framework it drew was sound.

Assessment: Strongest rule knowledge; specific scenario table incomplete due to data barrier.

5.4 Grok (xAI) — Most Systematic and Closest to Correct

Grok’s appendix section exhibited the most systematic structure among all models examined. Explicitly citing reliable sources such as TFF’s official site and Mackolik, Grok described a scenario model that reduced 81 theoretical match combinations to 10-12 logical groups. It presented transparently the steps of constructing an updated standings table for each scenario and applying TFF tiebreakers in order.

Grok correctly identified that Antalyaspor would be relegated automatically if they failed to win, and that head-to-head calculations would be decisive if they did win. Its assessment of Antalyaspor as the most vulnerable candidate aligns with the media consensus. The transparency of the calculation process (‘I could run all 81 scenarios in seconds with a Python script’) enhanced its credibility.

Assessment: Closest to correct — real sources, correct rules, transparent methodology.

5.5 DeepSeek — Honest Acknowledgement of Limits

DeepSeek maintained its meta-analytical approach in the appendix section. Decomposing the task into three cognitive layers, it succeeded on deterministic inference (identifying already-relegated teams) and constrained combinatorial reasoning (specific scenarios), but honestly acknowledged that the full 12-match head-to-head averaj calculation constituted ‘a fragmented data source problem’ it could not fully resolve.

Having correctly identified the four candidates and the three-team relegation figure, DeepSeek’s upfront acknowledgement of its data-collection and processing limitations made for a trustworthy evaluation. It correctly identified Antalyaspor as the most vulnerable candidate in general terms.

Assessment: Partial; correct framework, honest acknowledgement of limits, incomplete specific scenario map.

5.6 Meta AI — Correct Table, Wrong Core Assumption

Meta AI’s appendix section contains the most striking contradiction among all models examined: the standings table it used is entirely correct (Eyüpspor 32pts, Kasımpaşa 32pts, Gençlerbirliği 31pts, Antalyaspor 29pts — the same table used in this article). Yet the fundamental frame of the analysis is wrong: Meta AI assumed that TFF had decided 4 teams would be relegated in the 2025–2026 season. TFF’s actual decision was 3 teams.

This core error invalidates the entire scenario analysis. A calculation constructed on the assumption that the 15th through 18th placed clubs were all at risk was wrong from the start; in reality only the 16th through 18th places determined relegation. Meta AI’s combination of a correct standings table with an incorrect rule framework is a concrete demonstration of the principle: correct input plus wrong operation equals wrong output.

Assessment: Correct standings table, wrong team count — analysis invalid from the foundation.

5.7 Mistral AI — Frame Error and Rule Blindness

Mistral AI’s appendix section contains the most severe error in this study. The model failed to correctly identify the four actual candidates fighting relegation before matchday 34 (Antalyaspor, Gençlerbirliği, Kasımpaşa, Eyüpspor). This ‘frame error’ invalidated the entire scenario analysis from the outset. Calculations built on the wrong teams bear no relation to reality.

Beyond this, Mistral was unable to apply TFF’s ranking rules correctly, had incorrect information about the league format (18 teams, 34 matchdays), and misread the relegation positions on the table. All these errors were observed without deliberate hints being provided, and carried through to the article-writing stage. Mistral was confirmed in this experiment as both the most error-prone and the least rule-knowledgeable model.

Assessment: Wrong teams + rule blindness + frame error = entirely invalid analysis.

6. Comparative Assessment Table

The table below compares the seven models’ football analyses in their appendix sections across key dimensions.

Model	Correct Candidates?	Teams Relegated?	Correct Rules?	Appendix Quality	Predicted Relegated	Proximity to Correct Scenario
Gemini	No	18 (wrong format)	Partial	Simulation	Fictional team	Farthest — entirely fictional analysis
ChatGPT (GPT-5.5)	Yes	3 (correct)	Partial	Commentary-heavy	Antalyaspor	Correct candidates & count; specific scenario calc insufficient
Claude (Anthropic)	Yes	3 (correct)	High	Methodology-focused	Unclear (no data)	Strongest rule knowledge; data barrier limited scenario output
Grok (xAI)	Yes	3 (correct)	High	Systematic & transparent	Antalyaspor (high prob.)	Closest to correct — real sources, correct rules, transparent method
DeepSeek	Yes	3 (correct)	Partial	Meta-analytical	Unclear (limit acknowledged)	Honest about limits; specific scenario mapping incomplete
Meta AI	Yes	4 (WRONG!)	Low	Table correct, rules wrong	Multiple (mixed)	Correct standings table but ‘4 teams relegated’ error invalidated all analysis
Mistral AI	No	Unclear	Very Low	Frame error	Wrong teams	Most severe failure — wrong teams, rule blindness, entire analysis invalid

Source: Appendix sections of the collaborative articles and media analyses from sahadan.com, haber7.com, and Türkiye Gazetesi; assessment: Aydın Tiryaki & Claude (Anthropic).

7. Overall Assessment: AI in the Mirror of Football Calculation

The most important finding of this study is that the quality of the models’ football analyses in their appendix sections directly mirrors the quality of their AI methodology. The methodologically most transparent and reliable models (Grok and Claude) also built the most accurate frameworks for the football analysis. Gemini, which retreated into simulation, and Mistral, which started with the wrong teams, both failed entirely on the football calculation as well.

• The primacy of data is beyond dispute. Grok’s explicit reference to Mackolik and TFF grounded its analysis in concrete reality. Claude’s transparent disclosure of the data access barrier preserved its credibility. Meta AI’s possession of a correct standings table alongside a wrong team-count assumption demonstrates how a single erroneous parameter can invalidate an entire analysis.
• Rule knowledge determines analytical quality. Claude and Grok, having correctly interpreted the TFF averaj rule from the outset, were structurally differentiated from all others. Errors such as Meta’s ‘4 teams relegated’ assumption or Mistral’s incorrect ranking interpretation document how destructive rule ignorance can be.
• Antalyaspor is the mathematically most vulnerable candidate. Among all analyses, the models that correctly reached this conclusion (ChatGPT, Claude, Grok, DeepSeek) converged on the key finding. It is mathematically beyond dispute that Antalyaspor is relegated in every scenario in which they fail to win, and that if they do win, their strong bilateral head-to-head advantage gives them a genuine survival path.
• The gap between simulation and analysis is reflected in this table. The fictional calculations produced by Gemini, and partially by Mistral, have no connection to football reality. This gap is not merely a methodological preference; it is an ethical responsibility that directly affects the reliability of the information provided to users.

8. Conclusion

The 2025–2026 Trendyol Süper Lig relegation calculation has documented the examination of seven AI models against football reality. The fundamental lesson drawn from this examination is that the methodological quality of a model and the quality of its football analysis are inseparable.

The mathematically most accurate pre-match scenario, derived from available data, is this: Antalyaspor cannot stay up without winning; and if they do win, their strong bilateral head-to-head advantage significantly enhances their survival prospects. The model that presented this assessment most consistently in its appendix section was Grok. Claude’s methodological accuracy, despite the data access barrier, retained its value throughout.

This study ends with an inevitable warning for anyone working with AI models on football or any domain requiring dynamic data: a model’s scenario should never be accepted without asking how far its output is grounded in real data, how accurately it interprets the rules, and how honestly it discloses its limitations.

References

1. Tiryaki, A. (2026). AI on Trial: Relegation in the Turkish Super League. https://aydintiryaki.org/2026/05/17/ai-on-trial-relegation-in-the-turkish-super-league/

2. Tiryaki, A. & Gemini (2026). Behind the Scenes of AI: An Analysis of Illusion, Algorithmic Obsessions, and Information Ethics. https://aydintiryaki.org/2026/05/17/behind-the-scenes-of-ai-an-analysis-of-illusion-algorithmic-obsessions-and-information-ethics/

3. Tiryaki, A. & ChatGPT (GPT-5.5) (2026). Data Reliability and Problem Definition in Artificial Intelligence: A Case Study on Turkish Super League Relegation Scenarios. https://aydintiryaki.org/2026/05/17/data-reliability-and-problem-definition-in-artificial-intelligence-a-case-study-on-turkish-super-league-relegation-scenarios/

4. Tiryaki, A. | Claude (Anthropic) (2026). Can Artificial Intelligence Access Football Data?. https://aydintiryaki.org/2026/05/17/can-artificial-intelligence-access-football-data/

5. Tiryaki, A. & Grok (xAI) (2026). Artificial Intelligence and Human Collaboration: Transparent Calculation Process in the Süper Lig Relegation Battle and the Anatomy of the Dialogue. https://aydintiryaki.org/2026/05/17/artificial-intelligence-and-human-collaboration-transparent-calculation-process-in-the-super-lig-relegation-battle-and-the-anatomy-of-the-dialogue/

6. Tiryaki, A. & DeepSeek (2026). From the Perspective of an AI Assistant: Calculating Relegation in an Unknown League. https://aydintiryaki.org/2026/05/17/from-the-perspective-of-an-ai-assistant-calculating-relegation-in-an-unknown-league/

7. Tiryaki, A. & Meta (2026). An Anatomy of an AI Calculation Dialogue. https://aydintiryaki.org/2026/05/17/an-anatomy-of-an-ai-calculation-dialogue/

8. Tiryaki, A. & Claude (Sonnet 4.6) (2026). Sinking Deeper: Meta AI and the Süper Lig Test. https://aydintiryaki.org/2026/05/17/sinking-deeper-meta-ai-and-the-super-lig-test/

9. Tiryaki, A. & Mistral AI (2026). An Anatomy of an AI Calculation Dialogue: Calculating Relegation Scenarios in Süper Lig: Methodology, Data, and AI Collaboration. https://aydintiryaki.org/2026/05/17/an-anatomy-of-an-ai-calculation-dialogue-2/

10. Tiryaki, A. & Claude (Sonnet 4.6) (2026). What Happens When an AI Misreads a Football League?: Mistral AI and the Süper Lig Relegation Analysis: A Case Study. https://aydintiryaki.org/2026/05/17/what-happens-when-an-ai-misreads-a-football-league-mistral-ai-and-the-super-lig-relegation-analysis-a-case-study/

| aydintiryaki.org | YouTube | Aydın Tiryaki’nin Yazıları ve Videoları │Articles and Videos by Aydın Tiryaki | Bilgi Merkezi│Knowledge Hub  | ░ Virgülüne Dokunmadan │ Verbatim ░ | ░ Yapay Zekanın Türkiye Süper Ligi’nde Küme Düşme ile İmtihanı │AI on Trial: Relegation in Turkish Super League ░ 17.05.2026