Model

All leagues Premier League LaLiga Bundesliga

Target: home possession %. Possession is zero-sum inside a match, so this one number defines the whole result — the away side is 100 minus it. This is a regression problem, not a win/draw/lose one. No model is trained yet; the figures below are what the modelling layer has to work with.

Training rows

Usable features

Fit split

Holdout split

332

Fixtures to score

What a model must beat — mean absolute error, percentage points

Home side's own recent average Ignores who they are playing		10.04
Always predict 50% The know-nothing floor		9.94
Naive midpoint Home form vs what the away side concedes — the one that matters		7.87

A model is only earning its keep below 7.87 points of error. Beating "always 50" is not an achievement.

Training set

Rows Played matches where both sides had enough prior matches to form an average	18
Home possession seen	25.1% – 66.7%
Mean / spread	49.0% ± 12.6
Earliest match	03 Sep 2026
Latest match	13 Sep 2026
Split Chronological, never random	14 / 4

43 features against 18 rows. More columns than examples means any model will fit the training data perfectly and learn nothing that generalises. Until the season fills out, prefer a handful of possession columns and strong regularisation over the full set — or widen training.season in config.yaml.

Feature inventory — 43 columns present in both training and fixtures

Minutes spent ahead, level and behind are deliberately not here. They are only known after the whistle, so using them to predict possession would be leakage. They are stored for explaining errors instead: a match settled early is inherently less predictable, because the leading side stops trying to keep the ball.

Possession (29)

home_poss_l5away_poss_l5home_poss_l10away_poss_l10home_poss_l20away_poss_l20home_poss_home_l5away_poss_away_l5home_pass_acc_l5away_pass_acc_l5home_passes_l5away_passes_l5home_touches_l5away_touches_l5home_long_balls_l5away_long_balls_l5home_poss_lost_l5away_poss_lost_l5home_dispossessed_l5away_dispossessed_l5home_recoveries_l5away_recoveries_l5home_poss_won_mid_l5away_poss_won_mid_l5home_passes_f3_l5away_passes_f3_l5poss_gap_l5poss_gap_l10poss_naive_l5

Points and results (5)

home_ppg_l5home_ppg_l10away_ppg_l5away_ppg_l10ppg_edge_l5

Volume (6)

home_shots_l5home_sot_l5home_corners_l5away_shots_l5away_sot_l5away_corners_l5

Head to head (1)

h2h_meetings

Other (2)

home_matches_beforeaway_matches_before

Stake guidance — open lines, rated out of 10

Two separate things have to be true before staking much: the edge must be large, and the model must have shown it can find real edges. The first is arithmetic; the second is a track record. The rating is the smaller of the two, so a big edge from a model with no record scores low — correctly. It is 5 settled bets in, so the ceiling is low for a while yet.

Kickoff	Match	Line is about	Line	Under	Over	Model	Market	Bet	Edge	Stake	Why
13 Sep 19:00	RSO v ATM	Atlético de Madrid	52.6	1.84	1.84	52.6%^*	50% over	no bet	-4.3%	0/10	best side is OVER at 1.84, still 4.3% short of its 54.3% breakeven

^* 1 row here did not come from a fitted model. This league has too little history to fit one, so the figure is the naive midpoint: the side's own recent possession averaged against what its opponent concedes — on as few as two matches each. That formula shrinks towards 50 by construction, so it understates a mismatch. A strong side away at a weak one reads too low, and the edge it implies against a bookmaker's line can point the wrong way entirely. The Bet, Edge and Stake columns on these rows are arithmetic, not a tested prediction.

0/10 means the price is against you — not a close call, a bet to skip. Anything above that is a fraction of what you would stake at full confidence, and quarter-Kelly is treated as the ceiling even at 10/10, because Kelly assumes the probability is exactly right and this one is an estimate.

Next fixtures — with the naive prediction to beat

Kickoff	Match	Home form	Away form	Gap	Naive home poss
13 Sep, 16:30	Getafe CF v RC Deportivo	44.0%	41.4%	+2.6	51.3%
13 Sep, 19:00	Real Sociedad v Atlético de Madrid	48.9%	56.1%	-7.2	46.4%
14 Sep, 19:00	Villarreal CF v Real Betis	61.4%	52.1%	+9.3	54.6%
15 Sep, 17:00	Rayo Vallecano v RCD Espanyol de Barcelona	44.7%	46.9%	-2.2	48.9%
15 Sep, 18:00	Deportivo Alavés v Valencia CF	44.0%	44.5%	-0.5	49.8%
15 Sep, 19:30	Elche CF v Real Madrid	51.8%	56.8%	-5.0	47.5%
16 Sep, 17:00	Atlético de Madrid v CA Osasuna	56.1%	50.0%	+6.1	53.1%
16 Sep, 17:00	RC Deportivo v Sevilla FC	41.4%	43.3%	-1.9	49.0%
16 Sep, 19:30	Levante UD v Athletic Club	40.6%	49.3%	-8.7	45.7%
16 Sep, 19:30	FC Barcelona v R. Racing Club	68.3%	49.3%	+19.0	59.5%