Skip to main content
Educational analyticsNot betting advice · research use only.
GameTime Picks

Soccer · model-only forecasts

Ligue 1 match forecasts

Win, draw or win, expected goals, over 2.5, both teams to score and the likeliest scorelines for every Ligue 1 fixture in the next eight days — from the same model as our Premier League page.

Updated Oct 30, 11:58 AM ET · fit on 1,342 Ligue 1 matches

Model-only forecasts for research and entertainment — not betting advice. On a season it had never seen, the sportsbook closing line was more accurate than this model.

Friday, October 30

  • Le Havre ACvParis Saint-Germain
    2:00 PM ET
    6% home12.8% draw81.2% away
    Expected goals
    3.25
    Over 2.5
    63%
    Both score
    41.5%
    Likeliest score
    0–2 (13.7%)
  • NicevStade Rennais
    2:00 PM ET
    46.9% home25.7% draw27.4% away
    Expected goals
    2.61
    Over 2.5
    48.3%
    Both score
    51.8%
    Likeliest score
    1–1 (12.2%)
  • LillevLens
    4:05 PM ET
    40.6% home27.1% draw32.3% away
    Expected goals
    2.47
    Over 2.5
    44.9%
    Both score
    50.1%
    Likeliest score
    1–1 (12.8%)

Saturday, October 31

  • TroyesvLe Mans
    3:45 PM ET
    25.1% home21.8% draw53.2% away
    Expected goals
    3.29
    Over 2.5
    63.8%
    Both score
    62.9%
    Likeliest score
    1–1 (9.7%)

    Small-sample input: Le Mans's away rates come from 2 matches. The model treats these tiny splits as fully informative, so this forecast is unusually sensitive to them.

Sunday, November 1

  • StrasbourgvAJ Auxerre
    9:00 AM ET
    58% home21.9% draw20.1% away
    Expected goals
    2.99
    Over 2.5
    57.5%
    Both score
    55.7%
    Likeliest score
    1–1 (10.3%)
  • LyonvAngers
    11:15 AM ET
    69.2% home19.3% draw11.5% away
    Expected goals
    2.76
    Over 2.5
    52.2%
    Both score
    44%
    Likeliest score
    2–0 (13.4%)
  • Paris FCvAS Monaco
    11:15 AM ET
    34.1% home22.1% draw43.9% away
    Expected goals
    3.52
    Over 2.5
    68.2%
    Both score
    68.3%
    Likeliest score
    1–1 (9.1%)
  • MarseillevToulouse
    2:45 PM ET
    54% home22.2% draw23.8% away
    Expected goals
    3.12
    Over 2.5
    60.2%
    Both score
    59.6%
    Likeliest score
    1–1 (10.2%)

Monday, November 2

  • LorientvBrest
    2:45 PM ET
    45.9% home23.5% draw30.6% away
    Expected goals
    3.09
    Over 2.5
    59.6%
    Both score
    61.2%
    Likeliest score
    1–1 (10.7%)

Friday, November 6

  • ToulousevStrasbourg
    2:45 PM ET
    35.9% home24.7% draw39.4% away
    Expected goals
    2.90
    Over 2.5
    55.5%
    Both score
    58.6%
    Likeliest score
    1–1 (11.6%)

Saturday, November 7

  • AJ AuxerrevParis FC
    11:15 AM ET
    32.9% home30.5% draw36.6% away
    Expected goals
    2.03
    Over 2.5
    33.3%
    Both score
    40.7%
    Likeliest score
    0–1 (13.8%)
  • Paris Saint-GermainvTroyes
    2:45 PM ET
    87.3% home7.7% draw5% away
    Expected goals
    5.11
    Over 2.5
    88.6%
    Both score
    64.4%
    Likeliest score
    4–1 (7.2%)

How it’s doing now

45 matches graded · average log loss 0.8829 (a one-in-three guess scores 1.0986) — accumulating — still a small sample for a three-way market.

Only the win/draw/win forecast is graded so far. Over 2.5, both teams to score and the likeliest score are published here but not yet graded.

Every graded Ligue 1 match →

  • Paris Saint-Germain 2–0 Lyon: the forecast gave the home win 60%.
  • Toulouse 0–0 Troyes: the forecast gave the draw 20.4%.
  • AJ Auxerre 1–1 Le Mans: the forecast gave the draw 24.9%.
  • Angers 0–0 Lorient: the forecast gave the draw 26.2%.
  • Marseille 4–3 Le Havre AC: the forecast gave the home win 66.8%.
How it was testedIts result on a full season it had never seen, against a no-information guess, a plain rating and the sportsbook closing line.

How this model did on a season it had never seen

  • Across 306 matches of the 2025-26 season it was more accurate than knowing nothing about the clubs (log loss 1.0127 against 1.0652).
  • Its draw probabilities were calibrated: stated and observed draw rates differed by 2.7 percentage points on average.
  • The sportsbook closing line was more accurate than this model, by 0.0372 log loss, and so was a plain Elo rating (by 0.0108).

Model-only forecasts for research and entertainment — not betting advice.

Fixtures: ESPN public scoreboard. Results history: openfootball football.json (results, public domain). Kick-off times in Eastern Time.