With FIFA set to kick off on Thursday, June 11, 2026, the opening match on the Mexico Metropolis Stadium, I believe it will be enjoyable to construct one of the best ML mannequin we are able to to foretell match outcomes. To do that, I’ve introduced collectively a number of databases—49,000 matches—with knowledge on Elo scores, match outcomes, and cup places. From FIFA to the Baltic Cup, with matches from 1872 to 2026, we are going to take a probabilistic method to the game.
We’ll evaluate the efficiency of a number of ML fashions, together with
- multinomial regression
- multinomial ridge / elastic-net mannequin
- LightGBM
We may even work to grasp the strengths and weaknesses of our fashions to create a well-calibrated mannequin that predicts dwelling wins 86% of the time. By weighing mannequin efficiency, calibration, and complexity, we are going to discover one of the best mannequin for our knowledge.
Soccer by the Numbers
Lots of people say soccer is sleep-inducing. As a soccer fan, I disagree, however to be honest, this isn’t with out purpose. Nearly all of matches finish with fewer than 5 objectives, and something above 20 is an anomaly, if not unattainable. In distinction, it’s not unusual for one participant to attain greater than 50 factors in an NBA recreation. However regardless of the tempo, pubs from England to botecos in Rio stay full.
What critics don’t perceive is that the low rating could make a recreation extra fascinating, as this makes it tougher for groups to achieve a considerable lead, maintaining followers on the sting till the tip. Sadly, this additionally means matches finish in a draw near 22% of the time—which can be infuriating. But the game stays as well-liked as ever.
The truth that so many matches finish in a draw truly turns into a modeling downside later, however earlier than we get to that lets go over how we put this knowledge togther.
Stitching the information collectively
Oftentimes one of the best ways to enhance a mannequin is to easily get extra knowledge. We might be working with international_results.csv, international_team_ratings.csv and international_goalscorers.csv
We need to matchinternational_results.csv to international_team_ratings.csv so we are able to use Elo scores. This could possibly be easy, however as you would possibly’ve guessed, the crew names don’t match up completely, so we have to flip to textual content processing until we need to verify 336 groups individually. We additionally have to be extremely cautious of when the Elo ranking was up to date. We may take the Elo on the identical day the match happens, however that might be a supply of knowledge leakage, as Elo scores are up to date solely after the match. Making use of it as a characteristic tempting however problematic.
We should take the latest Elo rating, and as an extra engineered characteristic we hold observe of the time for the reason that newest Elo replace, positing that earlier scores can be extra informative than older ones. The code for becoming a member of these tables and the complete mission is out there within the Appendix.
international_results.csv
| Area kind | Examples |
|---|---|
| Match identification | source_match_id, date, season, competitors |
| Groups | home_team, away_team |
| Last end result | home_score, away_score, match_result, result_class |
| Context | impartial, event, metropolis, nation |
international_team_ratings.csv
| Function | Which means |
|---|---|
home_rating_pre_match |
Dwelling crew Elo earlier than kickoff |
away_rating_pre_match |
Away crew Elo earlier than kickoff |
rating_diff |
Dwelling Elo minus away Elo |
rating_age_days_home |
How stale the house crew ranking is |
rating_age_days_away |
How stale the away crew ranking is |
international_goalscorers.csv
| Function thought | Which means |
|---|---|
| Distinctive scorers in current matches | Whether or not a crew depends upon one scorer or many |
| Targets by high scorer | Focus of scoring |
| Current scoring type | Attacking output earlier than this match |
As a result of we’re doing a time-series prediction, we have to guarantee our cut up respects the time order. We’ll consider our mannequin on all video games from 2018 onward, which might be roughly 8,000 matches.
| Efficient cut up | Approximate date logic |
|---|---|
| mannequin practice | earlier a part of pre-2018 knowledge |
| validation | newest ~20% of the pre-2018 coaching pool |
| take a look at | 2018 onward |
Engineered Options
We need to transfer from fundamental match-level predictors in the direction of richer pre-match options that seize: crew energy, attacking and defensive high quality, dwelling/away results, matchup stability, goalkeeper energy, historic efficiency developments.
1. Draw-modeling options
Probably the most evident failure of our baseline multinomial logistic regression mannequin was its weak efficiency at classifying attracts. Whereas the mannequin may calculate the chance of a draw as a result of we outlined the goal variable as match_result ∈ {H, D, A} (Dwelling win, Draw, Away win), Draw was merely by no means the more than likely final result. We will see this by the lacking column for Attracts within the confusion matrix.
This poor draw efficiency shouldn’t be particular to at least one mannequin household. After we isolate high-confidence errors — circumstances the place the mannequin’s predicted class was improper, and its most predicted chance was no less than 0.60 — the identical sample seems throughout fashions: they’re systematically overconfident in dwelling wins. Many matches that truly led to attracts had been assigned a assured home-win prediction, suggesting that the fashions seize team-strength course higher than match-level uncertainty or draw probability.
To deal with this ‘blindness’ to the draw choice, we are able to engineer options reminiscent of abs_rating_diff, home_draw_rate_last_5, form_draw_rate_mean_last_5, and binary context options like impartial, flag_is_world_cup, and flag_is_friendly, indicating whether or not the match is on impartial floor or on the World Cup.
| Function group | Which means | Examples |
|---|---|---|
| Elo closeness | Measures how evenly matched the groups are. Smaller ranking gaps are particularly related for draw chance. | abs_rating_diff |
| Current draw tendency | Measures how usually every crew’s prior matches led to attracts. | home_draw_rate_last_5, away_draw_rate_last_10 |
| Mixed draw tendency | Captures whether or not each groups have just lately been draw-prone. | form_draw_rate_mean_last_5, form_draw_rate_mean_last_10 |
| Match context | Event and venue indicators that will have an effect on draw frequency. | impartial, flag_is_world_cup, flag_is_friendly |
With these options, our mannequin can now higher discriminate between Dwelling/Away wins and attracts, as evidenced by a 3.3% enhance in true-positive draw predictions. That is nonetheless low, on condition that ~20% of matches finish in attracts. So our options assist however not by a lot. This implies that it could possibly be price constructing a mannequin devoted to attract modeling, with the goal variable match_result ∈ {D, ¬D}, however for now we have to engineer extra options.
¬D represents not D that means our goal variable is the match ends in draw (1), or match doesn’t finish in draw (0)
2. Elo options
The common crew has an Elo barely above 1500; that is close to Saudi Arabia, Iceland, and Haiti for FIFA 2026. After we graph the distributions of Dwelling wins, Attracts, and away wins, we are able to see that because the distinction decreases, Attracts turn into more and more probably. Our distributions are additionally barely shifted to the left, indicating a small dwelling benefit, as anticipated.
We’d be leaving LogLoss factors on the desk if we relied solely on pre-match Elo as our solely characteristic. To get essentially the most from the information, we additionally
| Function | Which means |
|---|---|
home_rating_pre_match |
Dwelling crew Elo ranking earlier than kickoff. |
away_rating_pre_match |
Away crew Elo ranking earlier than kickoff. |
rating_diff |
Dwelling crew Elo minus away crew Elo earlier than kickoff. Constructive values favor the house crew. |
rating_age_days_home |
Days for the reason that dwelling crew’s Elo ranking was final up to date. |
rating_age_days_away |
Days for the reason that away crew’s Elo ranking was final up to date. |
3. Rolling past-performance options
A critic may argue that utilizing rolling previous efficiency and Elo shouldn’t be a good suggestion, since they each mannequin crew energy, which might add redundant or extremely correlated options to the mannequin.
Rolling previous efficiency does seize crew energy, however it’s particularly there to assist the modeling of crew momentum. Profitable streaks are a really actual factor in sports activities. Actually, the present best choice by supercomputers is Spain. One purpose they’re predicted first is their historic 31-match unbeaten streak coming into FIFA 2026.
| Function group | Which means | Examples |
|---|---|---|
| Current factors per match | Common factors earned over every crew’s earlier 5 or 10 matches. | home_points_per_match_last_5, away_points_per_match_last_10 |
| Current aim distinction | Common objectives scored minus objectives conceded over prior matches. | home_goal_diff_per_match_last_5, away_goal_diff_per_match_last_10 |
| Current draw price | Share of prior matches that led to a draw. | home_draw_rate_last_5, away_draw_rate_last_10 |
| Dwelling-away type variations | Distinction between the house and away groups on the identical rolling metric. | form_points_diff_last_5, form_goal_diff_diff_last_10 |
| Prior match counts | Variety of earlier matches accessible earlier than the fixture. | home_prior_matches, away_prior_matches |
4. Assault and protection type options
Whereas our mannequin tried to seize attacking and defending crew energy via factors, that is the place our mannequin falls in need of super-computer approaches. Trendy approaches usually additionally implement participant knowledge, which is invaluable in computing a crew’s strengths. As a result of we’re working solely with game-level knowledge, our modeling of attacking and defensive options is computed from earlier match outcomes like Current scoring charges, conceding charges, Scoring-rate distinction, and Conceding-rate distinction.
| Function group | Which means | Examples |
|---|---|---|
| Current scoring price | Common objectives scored per match over the earlier 5 or 10 matches. | home_goals_for_per_match_last_5, away_goals_for_per_match_last_10 |
| Current conceding price | Common objectives conceded per match over the earlier 5 or 10 matches. | home_goals_against_per_match_last_5, away_goals_against_per_match_last_10 |
| Scoring-rate distinction | Dwelling crew’s current scoring price minus away crew’s current scoring price. | form_goals_for_diff_last_5, form_goals_for_diff_last_10 |
| Conceding-rate distinction | Dwelling crew’s current conceding price minus away crew’s current conceding price. Decrease values favor the house crew defensively. | form_goals_against_diff_last_5, form_goals_against_diff_last_10 |
Grid Search
As a result of massive search grids can overfit in cross-validation, and grid search scales multiplicatively, parameters are searched logarithmically (1e-5, 1e-4, 1e-3, 1e-2). Besides with parameters like alpha, which should exist between zero and one.
glmnet_alphaControls the elastic-net mix between ridge and lasso regression, the place zero is Pure ridge, and one is pure lasso.
multinomial_decaypenalizes massive coefficients extra. That may cut back overfitting, however extreme decay can result in underfitting.
Grid Search O(n) = quantity of configurations examined × time to practice one mannequin
| Mannequin household | Grid/configurations proven | What was tuned |
|---|---|---|
| Baselines | majority_baseline, frequency_baseline, rating_diff_multinom |
Principally not tuned; comparability baselines |
| glmnet | alpha = 0, .25, .5, .75, 1 |
Elastic-net mixing parameter |
| multinom | decay = 0, 1e-5, 1e-4, 1e-3, 1e-2 |
L2 weight decay / coefficient shrinkage |
| LightGBM | less_regular, deeper, more_regular, current_final, l2_regularized, shallower, l1_l2_regularized, compact_robust, faster_small, slower_small |
Named bundles of tree-depth, learning-rate, boosting-round, and regularization settings |
LightGBM was essentially the most advanced mannequin household within the comparability. Not like the baseline fashions, which used few or no tuning parameters, LightGBM required decisions about tree complexity, studying price, boosting rounds, and regularization. This made it extra versatile, but additionally elevated the chance of overfitting if the parameters weren’t tuned rigorously. We additionally have to take care to not use a mannequin that’s extra sophisticated than our knowledge requires, as we may lose out on interpretability.
The GBM parameters had been tuned by evaluating a compact grid of LightGBM configurations. These configurations assorted tree complexity, studying velocity, variety of boosting rounds, and regularization energy, maintaining one of the best mannequin scored on log-loss. Beneath is a listing of the LightGBM parameters.
| Parameter | Which means |
|---|---|
learning_rate |
How a lot every new tree is allowed to vary the mannequin. Decrease values study extra slowly however can generalize higher. |
num_iterations / nrounds |
Variety of boosting rounds, that means what number of bushes are added. Extra bushes can enhance efficiency however may overfit. |
num_leaves |
Controls how advanced every tree will be. Extra leaves permit extra detailed patterns however enhance overfitting danger. |
max_depth |
Most depth of every tree. Deeper bushes seize extra advanced interactions. Shallower bushes are easier and safer. |
min_data_in_leaf |
Minimal variety of observations required in a leaf. Larger values make the mannequin much less delicate to small noisy patterns. |
lambda_l1 |
L1 regularization. Pushes some results towards zero, making the mannequin easier. |
lambda_l2 |
L2 regularization. Shrinks massive results and reduces overconfidence. |
feature_fraction |
Fraction of options used for every tree. Utilizing fewer options can cut back overfitting. |
bagging_fraction |
Fraction of rows used for every tree. Utilizing fewer rows may cut back overfitting. |
bagging_freq |
How usually row subsampling is utilized. If set to 0, bagging is often off. |
Last Mannequin
The official chosen mannequin was LightGBM with the safe_plus_form_compact characteristic set, utilizing 20 pre-match options drawn from Elo scores, event context, and lagged crew summaries. It was chosen based mostly on the bottom validation-set multiclass log loss, with the take a look at set reserved for remaining reporting.
The chosen LightGBM mannequin achieved a validation log lack of 0.893 and a take a look at log loss of 0.873. Its validation end result was one of the best throughout the Mannequin comparability, however the margin over regression was small: multinomial regression trailed by solely about 0.002 log-loss factors on validation. On the held-out take a look at set, multinomial regression barely outperformed LightGBM on each log loss and macro F1.
Which means the end result needs to be interpreted cautiously. LightGBM is the formally chosen predictive mannequin, however the proof doesn’t present that gradient boosting clearly dominates easier regression fashions for the given knowledge. Regression fashions stay extremely vital as a result of they’re simpler to interpret and carry out almost in addition to, and in some take a look at metrics barely higher than, different strategies.
Function engineering produced equally modest good points. Compact lagged options improved validation log loss relative to baseline, however the take a look at enchancment was tiny. Goalscorer options didn’t meaningfully enhance log loss within the Mannequin comparability.
The clearest limitation was draw prediction. The chosen mannequin virtually by no means predicted draw as the highest class: on the take a look at set, it accurately predicted solely 2 attracts out of 1,784 precise attracts, for draw recall of 0.11%. This implies that the mannequin’s chance estimates should still include helpful info, however argmax classification stays strongly biased towards dwelling and away wins, making a separate mannequin for draw modeling an affordable subsequent step. Elo and compact pre-match type present a helpful sign stack, however the good points over robust baselines are incremental.
The mannequin is significantly better at predicting dwelling wins than away wins on the take a look at set:
- It accurately identifies about 87% of precise dwelling wins
- It accurately identifies about 63% of precise away wins
The mannequin can also be able to outputting a chance distribution over Dwelling, Draw, and Away wins, which is commonly extra helpful than only a single arduous prediction.
Calibration
The baseline-plus fashions are broadly effectively calibrated on the take a look at set. Throughout confidence bins. This implies predicted confidence tracks noticed accuracy, that means when the fashions are reasonably assured, they’re right at roughly the corresponding price, and when confidence rises, noticed accuracy rises with it. The deviations from the perfect calibration line are modest, suggesting that the fashions’ chance estimates are typically usable somewhat than only a rank-ordering of outcomes.
The plot under measures calibration of the highest predicted class—the mannequin’s confidence in whichever final result it selected—not calibration for dwelling wins, attracts, and away wins individually. A mannequin can due to this fact look effectively calibrated total whereas nonetheless misestimating one class, particularly attracts. The combination calibration plot helps the declare that the fashions’ confidence scores are broadly reliable, however it doesn’t, by itself, present that the draw chances are effectively calibrated.
The category-specific calibration plots present the place that combination image holds and the place it turns into extra sophisticated. Dwelling-win and away-win chances comply with the perfect calibration line intently throughout most bins: because the mannequin assigns larger chance to both final result, the noticed frequency rises at roughly the identical price. In sensible phrases, the mannequin’s dwelling and away chances behave like significant chances, not simply scores.
Attracts are totally different. The mannequin’s draw chances are moderately calibrated inside its vary, however that vary is slender. It hardly ever assigns draw chances a lot above the low-to-middle vary, even when the match is comparatively balanced.
That is the central distinction: the mannequin doesn’t ignore attracts; it often treats them as danger components somewhat than probably outcomes. Draw chances should still be helpful for measuring draw danger, however attracts seldom turn into the mannequin’s high prediction, which helps clarify the persistent weak point in draw recall.
Ranking Distinction Evaluation
The rating-difference evaluation reveals why attracts are structurally tough for the mannequin. Noticed draw charges are highest when the groups are intently matched and decline as absolutely the Elo ranking hole widens. All three mannequin households study this broad sample: their predicted draw chances additionally fall as matches turn into extra lopsided.
The failure shouldn’t be directional however scalar. In essentially the most evenly matched fixtures, the noticed draw price is roughly one-third, whereas the fashions assign draw chances nearer to one-quarter. They accurately determine balanced matches as extra draw-prone, however they don’t elevate the draw chance sufficient. Because of this, the mannequin can acknowledge draw danger with out usually choosing a draw because the more than likely final result. This reconciles the obvious contradiction between affordable draw calibration and weak draw recall: the chances transfer in the appropriate course, however often not far sufficient to win the argmax determination, that being to select the category with the very best predicted chance.
Function Significance
As you would possibly count on, an important characteristic for our mannequin is the ranking distinction, adopted by whether or not the match was on neural floor—a distant second. By checking the characteristic significance, we are able to see which of our engineered options supplied significant sign.
Conclusion
I believe it is a good time to debate dataset measurement and mannequin alternative. Usually, the bigger and extra advanced the dataset, the extra purpose we now have to decide on a extra sophisticated mannequin. As we noticed on this instance, the good points from switching from regression to LightGBM had been very small; it is a good signal that trying a extra advanced mannequin on this knowledge won’t yield higher predictions. Soccer forecasting is much less about discovering a magic algorithm and extra about constructing leakage-safe options, evaluating interpretable baselines, and asking whether or not the mannequin’s confidence is deserved.
For now, one factor is evident: wer’re gonna want extra knowledge if we need to get a greater prediction. Notably player-level knowledge—figuring out if Neymar is sitting out is essential. The granularity of the information can also be vital if we need to change our forecast as the sport progresses.
Apendix
The code for the entire mission will be discovered on my GitHub
The info source has a Inventive Commons CC0-1.0 license
make_team_clean <- operate(team_name) >
stringr::str_squish() - stringr::str_squish()
- stringi::stri_trans_general(“Latin-ASCII”)
- Converts accented Latin characters to plain ASCII characters.
- str_to_lower()
- stringr::str_replace_all(“[^a-z0-9]+”, “_”)
- It replaces something that’s not a lowercase letter or quantity with an underscore.
