GW 21 – Torstensson got 30 matches. Aksum got 18.
Karl Marius Aksum was dismissed by Mjällby on Tuesday, the day after a 1–4 home defeat to IFK Göteborg. At 19.00 on Friday Mjällby open round 21 away at Häcken with Andreas Brännström on the bench, and on Wednesday they play the rearranged fixture against AIK at Strawberry Arena. Two matches in six days, a new coach, three straight defeats. That is the simple story.
The harder one is in the weekly report's coaching table. It measures two things. First, how many points per match a team takes compared with what its underlying strength justifies, calculated with an Elo model that sees only the strength gap and home advantage. Second, how many points the team takes compared with what the chances in its matches justify, in other words against expected goals. Aksum took 1.11 points per match over 18 matches. The Elo model expected 1.86, and the gap of minus 0.75 is the worst figure for any coach who still had a job on Monday. But against expected goals the same team under the same coach took 0.27 points per match more than it deserved. Mjällby have 20 points from nineteen rounds, and the model's expected points say 19.0. Aksum was not removed because the team was unlucky. He was removed because the team is no longer the team the Elo model believes it to be.
At the top of the same table stands Anders Torstensson, who left Mjällby after last season: 2.50 points per match over 30 matches, 0.63 above Elo expectation and 1.02 above what expected goals justified. No other coach in the table comes close. Last year's Mjällby took more points than their chances to a degree that does not recur, and the Elo model learned that this was a top side. That knowledge now punishes the successor. What Aksum inherited was not a top side. It was a league position.
The underlying play shows how far from that position the team has drifted. Mjällby concede 1.8 expected goals per match, with only Halmstad and Örgryte worse. Within eight seconds of losing the ball they concede 0.47 expected goals, the highest figure in the league and 0.19 more than Friday's opponents Häcken. Mjällby play more switches of play than anyone, 17.5 per match, and lose the ball in ways that cost immediately. At the same time, the players who built last year's numbers are gone. Elliot Stroud was sold to Hull City on 9 August and still sits third in the report's list for open-play threat created, 0.49 per match. Ludwig Malachowski Thorell was unveiled at Basel in July and sits seventh for shot assists. Abdoulie Manneh is on loan at Cercle Brugge for the season and remains among the 25 highest-rated players in the league. Three of Mjällby's names in the top lists no longer play for Mjällby.
Friday in Gothenburg: Häcken against a team that has just changed its brain
Häcken have 32 points on 30.8 expected and are, measured by points against underlying numbers, the second most accurately described team in the top half. Only Hammarby sit closer to their underlying level. Häcken win more free kicks than anyone in the league, 14 per match, have the league's second-best throw-in control and concede just 0.28 expected goals within eight seconds of losing the ball. Andreas Linde has prevented 52 per cent of the expected goal value on shots he has faced, Robin Wallinder at Mjällby 42 per cent of 13.45, which says something about how many shots on target Mjällby allow. Gustav Lindgren is eighth in the xG table with 0.49 per match. And Häcken too carry a ghost in the top lists: Amor Layouni, who joined AEK Larnaca when his contract expired in the summer, still tops the league for open-play threat created and sits fourth for assists.
Head to head, this is one of the few fixtures where the lower-placed team has held the recent upper hand. Mjällby won the four meetings before May this year, including 3–0 on Hisingen in April 2025, before Häcken won 1–0 at Strandvallen on 17 May. 25 meetings since 1983 read 11–6–8 to Häcken with 2.7 goals per match.
Sunday in Borås: the model sees the same team, the table sees twelve points
The most instructive match of the round kicks off at 14.00 on Sunday. Elfsborg have 27.5 expected points. Kalmar have 27.5 expected points. Elfsborg create 1.3 expected goals per match and concede 1.3. Kalmar create 1.2 and concede 1.2. Both teams generate 1.4 in open-play expected threat per match. Elfsborg have 31 points and sit sixth. Kalmar have 19 and sit thirteenth.
Twelve points separate two teams the model cannot tell apart. The points projection makes it plainest: it takes Elfsborg to 44.8 points and fifth place, Kalmar to 32.8 and twelfth, with exactly the same slope of 1.38 expected points per match. The gap is twelve points today and twelve points in round 30. The model does not expect either team to claw anything back, because there is nothing in the play to say it should.
The coaching table supplies part of the explanation. Björn Hamberg has taken 0.35 points per match more than Elfsborg's expected goals justify over 20 matches. Toni Koskela has taken 0.40 less over 19, the largest negative deviation among coaches still in post. Then there are things neither model nor table captures. Kalmar sold Anthony Olusanya to St Patrick's Athletic at the end of July and Rony Jansson to St. Pauli on 31 July. Olusanya still tops the report's Playmaker ranking and sits fourth for threat received per match. What remains is Charles Sagoe Jr, on loan from Arsenal until January, with 0.43 assists per match, and Samuel Brolin, second in the league for both claims and crosses stopped. Kalmar have the league's weakest build-up, with 31 per cent of attacks from their own half reaching the final third, but one of the league's three best corner defences at 0.01 expected goals conceded per corner. The team is winless in six since beating Mjällby 2–1 on 25 July and has lost all four matches on natural grass over the past two seasons. Borås Arena is artificial turf.
Elfsborg have scored in seven straight matches and won their last two at home. Arber Zeneli is fourth in the league for open-play threat created with 0.44 per match, Isak Pettersson second among goalkeepers for goals prevented with 5.7. Per Frick, who has scored five goals against Kalmar over the years, has been out since April, and penalty taker Simon Hedlund suffered a muscle tear at the weekend. 71 meetings since 1949 read 28–22–21.
Saturday: AIK and Västerås, two teams still praised for players who have left
The weekly report's preview of AIK against Västerås highlights two key players. Áron Csongvai, fifth in the league for open-play threat created with 0.38 per match, was sold to Saint-Étienne in July. Mikkel Ladefoged, 0.52 expected goals per match, was sold to Lincoln City on 30 August. The text is correct about the season and wrong about Saturday, and that is not an error in the report but a property of season data. AIK have also lost Bersant Celina to Al-Ettifaq in August and Zadok Yohanna to Brighton in June, and on Monday the club announced that Kevin Filling, sixth in the assist table, is out for the rest of the year with a knee injury. Four of the players who built AIK's season will not be on the pitch on Saturday.
Before Monday's 0–0 in Malmö, AIK had 31 points on 18.9 expected. The twelve-point overperformance is the largest in the league, and it has a name. Kristoffer Nordfeldt leads Allsvenskan for goals prevented with 5.9, has saved 75 per cent of the shots he has faced and prevented 65 per cent of the expected goal value on shots on target. AIK concede 1.8 expected goals per match, one of the league's four worst defensive figures, and sit fifth with 32 points. AIK play twice this week, Västerås on Saturday and Mjällby on Wednesday, with one match fewer played than most. Hervé Matthys is third in the league for progressive passing. Västerås have 29 points on 23.6 expected, Ismet Lushaku is second in the assist table with 0.56 per match and Elis Jäger has six clean sheets. The teams have met eleven times since 1955, AIK lead 6–2–3, and it finished 1–1 in Västerås in May.
Saturday in Gothenburg: fourteen goals above expectation, and the number that explains them
IFK Göteborg have conceded 40 goals. With 1.3 expected goals against per match over twenty rounds, expectation lands around 26. The deviation is the largest in the league in either direction, and the weekly report's goalkeeper pages offer an unusually concrete explanation. Elis Bishesari has saved 57 per cent of the shots he has faced and prevented 32 per cent of the expected goal value, 3.63 of 11.22. Among the sixteen first-choice goalkeepers that is the lowest share in the report. Rockson Yeboah, fourth in the league for interceptions, was sold to Osasuna in August.
Going forward, Göteborg are a different team. Sam Larsson leads the xG table with 0.75 per match and scored twice and assisted once in the 4–1 win over Mjällby. Tobias Heintz leads the league for shot assists with 3.2 per match and sits second for open-play threat created. Hjörtur Hermannsson tops progressive passing with 564 metres per match, and the team has the league's best threat from corners at 0.21 expected goals per match. Göteborg sit eleventh with 26 points on 31.4 expected and have won two in a row.
Halmstad have eleven points, thirteen goals scored and the league's worst defence by expected goals against, 2.2 per match. They are also the league's worst at creating anything directly after winning the ball, 0.16 expected goals within eight seconds, against Göteborg's 0.29. Tim Rönning leads the league for saves per match, xG saved per match and claims, and has prevented 54 per cent of 19.31 in expected goal value on shots on target. That latter figure is the highest any goalkeeper in the league has had to face. 108 meetings since 1933 read 49–33–26 to Göteborg, and it finished 1–1 at Örjans Vall in April.
Sunday: Hammarby against a team they have beaten ten times out of thirteen, and Malmö without their striker
Hammarby host Brommapojkarna at 14.00 with the league's highest expected goals per match (2.3), lowest expected goals against (0.95), goals in eleven straight matches and four straight home wins. The head-to-head stands at 10–1 from thirteen meetings since 2007. Paulos Abraham is third in the xG table with 0.68, Victor Stange Lind third for assists, and Nahir Besara has scored three goals against BP over the years. Brommapojkarna win fewer balls in the opposition half than anyone, 21 per match against Hammarby's 38, but in Leo Cavallius they have a goalkeeper who has prevented 58 per cent of the expected goal value on the shots he has faced and leads the league for crosses stopped. Brommapojkarna beat Elfsborg 2–1 at the weekend after losing 0–4 to Gais the week before. Hammarby dropped points at Örgryte and now have Djurgården a point ahead of them.
Malmö host Örgryte at 16.30 and do so without the two players the report leans on most. Erik Botheim, ninth in the xG table with 0.48 per match, was sold to Viking on 2 September. Sead Haksabanovic is, according to the card count, suspended after three yellows. Malmö still have 30 points after Monday's 0–0 against AIK, were already five points above their expected 24.1 before that match, and have the league's best first touch on attacking corners (59 per cent), the strongest free-kick threat in the fixture (0.24 expected goals per match against Örgryte's 0.11) and, in Malte Frejd Pålsson, the league leader for progressive carrying at 163 metres per match. Robin Olsen has saved 77 per cent of the shots he has faced and prevented 62 per cent of the expected goal value on shots on target. Örgryte have 15 points, 43 goals conceded, the league's lowest field tilt (0.33) and five matches without a win since beating AIK 3–0 on 2 August. Noah Christoffersson, third in the league for threat received per match, was sold to Thun on 1 September. Hampus Gustafsson makes four saves per match. 89 meetings since 1931 read 45–24–20 to Malmö with 3.1 goals per match.
Monday: four straight clean sheets against the league's best ball-winners, and a leader who won again
Djurgården have won four in a row and kept a clean sheet in all four. Jacob Rinne leads the league with ten clean sheets and the team concedes 0.15 expected goals within eight seconds of losing the ball, the lowest in the league. With 38 points on 34.8 expected Djurgården sit second and are projected to 55.4 points, a tenth behind Hammarby's 55.5. According to the model, the race for second is a dead heat. Bo Hegland leads the assist table with 0.76 per match and Kristian Lien is second in the xG table with 0.70.
Gais are the opposite of a team that wins more than it should. 27 points on 33.2 expected is one of the league's four largest underperformances, even though the team wins more balls in the opposition half than anyone in the league (39 per match) and shares the league's second-best defence by expected goals against. Mergim Krasniqi has eight clean sheets and the league's highest save percentage, 81, but went off with a head injury against Häcken. Samuel Salter, seventh in the xG table, took a blow to the head in late August. Two of Gais's three most important players are doubtful for a fixture in which 76 meetings since 1927 have produced 3.3 goals per match.
Sirius won 0–1 in Västerås, ended a run of three straight defeats and lead Djurgården by seven points. The points projection gives the leaders 63.9 points, up from 62.7 a week ago, and the projected gap to second has grown from 6.8 to 8.4 points. The preview of Monday's home match against Degerfors names Robbie Ure as the player to watch. He was sold to Sevilla on 11 August. Isak Bjerkebo, whose Sevilla move collapsed on 1 September, instead extended to 2030 and sits fifth in the xG table with 0.53. Sirius still have the league's best chance creation, 0.05 expected goals per attack reaching the final third. Degerfors have the league's worst, 0.02, and the league's lowest expected goals per match, 0.92. Juhani Pikkarainen, fourth in the league for progressive passing, tore his ACL in June and is out until summer 2027. Nahom Girmai Netabay creates 0.32 in expected threat per match, Matvei Igonen is third among goalkeepers for xG saved per match. Degerfors beat Halmstad 2–0 at the weekend. Nine meetings since 2021 read 6–2–1 to Sirius, and it finished 0–3 at Stora Valla in April.
The coaching table is a blunt instrument, and the report itself notes that Elo catches up with a good coach in 30 to 50 matches. Torstensson got 30. Aksum got 18. The difference between them is not primarily what they did with Mjällby, but which Mjällby they were given.
That is also the point of Elfsborg against Kalmar on Sunday. Two teams with identical expected points, identical expected threat, twelve points apart. Koskela is contracted until 2027. If Kalmar lose in Borås it becomes seven matches without a win, and the question asked will be whether the coach has lost the dressing room. That question is the wrong one. Kalmar and Elfsborg play football at the same level by every measure the model has. One team sits 8.5 points below its underlying numbers, the other 3.5 above, and neither coach has changed the level. The question is who takes the blame when the table and the underlying numbers finally say the same thing.
Round 21 fixtures: Häcken–Mjällby (Fri 11/9 19.00), AIK–Västerås (Sat 12/9 15.00), Göteborg–Halmstad (Sat 17.30), Hammarby–Brommapojkarna (Sun 13/9 14.00), Elfsborg–Kalmar (Sun 14.00), Malmö–Örgryte (Sun 16.30), Djurgården–Gais (Mon 14/9 19.00), Sirius–Degerfors (Mon 19.00). Rearranged fixture: AIK–Mjällby (Wed 16/9 19.00).
Source: Allsvenskan Weekly Report – Round 21, PlaymakerAI. Transfer, injury and coaching information checked against the clubs' own channels and the Swedish football press on 9 September 2026.