As the three Baltics leagues get underway, we have decided to poke our heads into a popular trend of recent years – asking artificial intelligence (AI) for end-of-season predictions. However, instead of doing this through one of the many existing websites or services, we put the three leading models to the test in their most frontier configurations – OpenAI’s ChatGPT 5.2 Pro, Anthropic’s Opus 4.6 Extended, and Google’s Gemini 3.1 Pro – to see not only what final standings they will predict and using what methodologies, but also how they will differ among each other. This is what they came up with and how our Newsroom evaluated their predictions.
Across the three leagues, three themes dominate
First, Latvia remains the most top-heavy: Riga FC and RFS absorb an outsized share of points, and the title tends to live inside a one-result margin. Second, Estonia keeps the clearest tiering: Flora and Levadia on top, Kalju and Paide behind them, followed by everyone else. Third, Lithuania produces the widest spread in possible outcomes – ranging from continuity for Kauno Žalgiris, rebound for Žalgiris Vilnius, to a steadier “second-place ceiling” for Hegelmann.
VIRSLĪGA 2026
The league’s defining feature stays in place across all three forecasts: a two-club summit and a statistical cliff beyond it. Each of the three models keeps Riga FC and RFS around the 80% points-won band (or higher), and every model leaves third place far behind that pace.
The Championship decision can be correlated with the assumptions that each of the models makes about ‘what makes Champions’. Gemini 3.1 puts RFS first on 91 points, built on a “bounce-back” logic after Riga’s narrow title, plus a heavier emphasis on scoring rates. Making an argument that “after Riga FC secured the 2025 title by a single point, RFS is statistically favored for a bounce-back” and noting RFS’s average of 2.77 goals per match versus Riga’s 2.36. Its version of an RFS title includes 102 goals and a +78 goal difference, a projection that treats attacking throughput as the separator inside an otherwise stable duopoly.
Opus 4.6 and ChatGPT 5.2 forecast keep the title pace lower – in the mid/high 80s, and treat the margin as a conversion problem rather than a structural one. Opus edges Riga FC first (87) with RFS second (85), and the reasoning is anchored in longer-horizon weighting and small trend adjustments, with European commitments explicitly treated as correlated with dropped points. ChatGPT table keeps the same “one-result” tension but flips the winner: Riga 88, RFS 87.
The top-four debate shows the largest disagreement for the Latvian league. Gemini projects a sharp “European tier” led by Auda at 61 points and Liepāja at 57. That implies a return to Auda’s higher-ceiling seasons and a more stable separation from the mid-table. Opus and ChatGPT’s projections keep Liepāja third and Auda fourth, and both hold Auda closer to the 50-point mark, reading the recent downturn as something that takes more than one season to reverse in a 36-game league.
One line is almost unnervingly consistent – Daugavpils. All three forecasts land them on 46 points in fifth place, the closest thing to a shared fixed point across the full comparison.
The relegation picture is where “newcomer penalty” starts to show. Gemini puts Ogre United ninth (29), arguing that promoted sides often scrape survival through draws and unpredictability, while Grobiņa fall into last place. Opus is harsher on a true first-timer and sends Ogre bottom (28), with Grobiņa surviving again. ChatGPT sits in the middle – Ogre remain in the pack (33), and the bottom two are split by a small points margin.
Decisive factors in Latvia
Gemini rewards scoring dominance and tier structure, and it is more willing to “restore” Auda to a high-ceiling band. Opus leans on a four-year weighted baseline plus explicit trend adjustments and risk labels, which pushes the newcomer lower and trims the title totals. ChatGPT stays closest to short-horizon regression and goal-difference sustainability, which compresses the mid-table and keeps the bottom-half arithmetic tight.
Newsroom’s verdict: “I think AI can factor in a lot, but sport presents plenty of unknowns: injuries, player psychology, senior leadership issues, finances, unknown team personalities failing to blend and weather.
It’s not the top of the table that surprises me, but the mid-table. FS Jelgava are a strong team, and BFC Daugavpils’ transfers and pre-season hasn’t looked so strong.“
PREMIUM LIIGA 2026
Estonia produces the strongest agreement on finishing order at the top. Flora first, Levadia second, Kalju third, Paide fourth appears in all three tables. The disagreements are about pace and about who gets pulled into the mid-table slog.
Gemini sets the title pace highest: Flora 86, Levadia 84. That reflects an expectation that the champion tends to live near the upper end of the 75–85% points-won band. It also factors in longitudinal performance, framing Flora as the better “late-round grinder” but Levadia as capable of more dramatic scorelines, albeit against lower-end teams. It does note, however, that Levadia historically tends to drop points against Flora in Tallinn Derbies, which automatically puts then at disadvantage in the race for the title.
Opus and ChatGPT sit lower at the top: Flora around 80–81 and Levadia 78–79, with a similar gap. Both are closer to a view that the recent champion band is “low-80s points” rather than mid-80s. Kalju and Paide remain stable in each forecast, with totals in the high-60s to low-70s that keep them clear of the mid-table.
The fifth-place zone is also stable. All three keep Narva Trans fifth, and all three keep the number in the same neighborhood (47–50). Narva come through as the league’s gatekeeper: good enough to beat most of the bottom half often but rarely consistent enough to take a full step into the top four.
The biggest disagreement is Harju. Gemini places Harju sixth on 42 points, extending the “safe mid-table” upward. Opus drops Harju to ninth on 30, viewing recent promoted-team performance as vulnerable to regression without a deeper multi-season base. ChatGPT also leaves Harju closer to the danger zone (35), with Vaprus taking the clearer mid-table advantage.
At the bottom there’s near-total convergence as Nõmme United are placed last on 22 points in all three forecasts, and Kuressaare and Tammeka hover around the survival line in the high-20s to low-30s.
Decisive factors in Estonia
Gemini pushes the top-end pace upward and treats Harju as capable of consolidating above the line. Opus reads the league through structural gaps – top four separation and the persistent 4th-to-5th drop – and then assigns high variance and lower confidence to the survival cluster. ChatGPT keeps the mid-table winners as Narva and Vaprus and leaves Harju inside the “one bad month changes everything” range.
Newsroom’s verdict: “Harju in the 9th seems… interesting. Overall the three are very last season-y; I would personally place Tammeka a little bit higher (or at least closer to the Top 5 race) and also expect a very tight fight between Nõmme United and Kuressaare for survival“
TOPLYGA 2026
Lithuania produces the widest split in league winner predictions. Kauno Žalgiris are the continuity pick in two forecasts and FK Žalgiris are the rebound pick in the third. Hegelmann sit in the contender frame in every version, though the expected ceiling differs.
Gemini’s projection is the boldest at the top: Žalgiris Vilnius win the league with 80 points, built on historical bounce-back expectations for a title-heavy club. According to the model, “after a disappointing 3rd-place finish in 2025 (62 points, 57% RPW), they have historically responded with a +15-20% increase in points won the following year through aggressive mid-winter scouting“. That being said, such optimism will hardly be shared by those of us who have actually seen their pre-season friendlies. Kauno Žalgiris then follow at 72, Hegelmann at 67. Panevėžys are moved into a much stronger fourth-place band at 61, and the lower half is harsher on Džiugas, who fall to 30 points.
Opus and ChatGPT both keep Kauno Žalgiris first, with a tighter title band. Opus has Kaunas on 74, Hegelmann 64, Žalgiris 63, describing a three-way chase inside a relatively compressed points environment. ChatGPT keeps the top three higher (Kaunas 73, Žalgiris 71, Hegelmann 67), closer to a “stronger top-end accumulation” season with less drop-off in the top two’s conversion rate.
The mid-table is also interpreted differently. Opus expects a dense block from Sūduva through Panevėžys, then a stable lower-mid tier for Džiugas and Banga. Gemini elevates Panevėžys and pushes Džiugas toward the bottom band. ChatGPT keeps Panevėžys and Šiauliai in the high-40s/low-50s zone and holds Džiugas safer.
TransINVEST are the biggest wildcard for the three models. Gemini places them seventh on 38 points, supported by a view of resource elasticity and survival capacity among returning “yo-yo” clubs. ChatGPT puts them ninth on 36 points, reflecting the usual trend toward the playoff band for promoted teams. Opus assigns them 35 points and still places them 10th, tagging the pick as very low confidence and high variance; its own notes frame that position as one that could shift several places in either direction.
Decisive factors in Lithuania
Gemini weights rebound dynamics, draw tendencies, and club resources heavily, producing the highest champion total and the sharpest Panevėžys rise. Opus leans on a four-year weighted base with explicit transfer/coaching risk factors and trend adjustments, keeping the title band tight and attaching low confidence to the bottom calls. ChatGPT holds closer to recent points share and goal-difference sustainability, which keeps the top two close and keeps mid-table compression as the dominant feature.
Newsroom’s verdict: “Not very surprising for TOP2 and last place in ChatGPT and Gemini but Hegelmann is perhaps too high (especially by Claude) whereas Panevezys not being considered enough, much like Transinvest. It’s also pretty odd for Gemini to put Vilnius first, as Kaunas still look superior on paper for a lot of people.”
A shared Baltic bottom-line
While the three leagues project differently at the top – Latvia’s duopoly, Estonia’s stability, and Lithuania’s Wild West – the survival mathematics converge. Points totals in the low-30s sit near the line in all three sets of forecasts, and totals in the 20s carry heavy relegation pressure. The most decisive “swing factor” in every league lands in the same place, on a team able to turn a small number of losses into draws, and a small number of draws into wins across a long season where the table is shaped by steady accumulation of points rather than by dramatic up’s and down’s.
This is also illustrated well by this table that Gemini had gone ahead and produced on it’s own initiative, with explanations below.
| League | Top 3 Point Spread | Standard Deviation | Competitiveness Rating |
| Virslīga | 30 Points | High | Low (Two Giants) |
| Premium liiga | 15 Points | Medium | Moderate |
| Toplyga | 13 Points | Low | High |
Latvia: The “Elite Gap” (30 Points)
Latvia operates under a Duopoly. The Index gives the Virslīga a “Low Competitiveness” rating because the top two (RFS/Riga) are effectively in a separate league. Statistically, they are expected to win 85%+ of their matches against the bottom six. This creates a massive “dead zone” in the middle of the table.
Estonia: The “Tapered Transition” (15 Points)
Estonia is the most “stable” league. The gap between the Top 2 and the chasing pack (Kalju/Paide) has narrowed in 2025. The Index shows that while Flora and Levadia remain favorites, they are no longer “immune” to losses against 3rd or 4th place teams.
Lithuania: The “Parity King” (13 Points)
The Toplyga has the highest “Entropy” (unpredictability). With a spread of only 13 points between the podium spots, it is the only Baltic league where the title race is frequently a three-way or even four-way battle deep into the fourth round of matches.
