Premier League predictions: every game this weekend, plus the best of Europe

Arsenal have won four of their last five and the club model makes the home win 72%, a touch above the 68% the price implies. The last five meetings between these two all went Arsenal's way. If you want a price in this game, the club model makes both teams scoring 56% against the 45% that the 2.08 implies.

Our numbers split here. The club model has Fulham at 52% and the Intel page agrees, but the match model gives them only 33% and prefers Ipswich at 2.75. Fulham arrive on DDLLL, Ipswich on LWLLW, and the last two meetings were draws. The tiebreaker is the run-in: Fulham have the kindest five games in the league and this is the first of them.

Brighton's rating is 113 points clear of Sunderland's, and the form reads WWDLW against LLDWL. Our models put the away win anywhere from 38% to 55%, against a price that implies 41%. Goals could be scarce: the club model makes under 2.5 at 2.02 a 55% shot, and the Intel page has over 2.5 at just 45%.

The result is a true three-way: Villa 2.60, Brentford 2.59, the draw 3.50, and our models cannot agree on it either (the match model leans Brentford at 41%, the club model leans Villa at 43%). What they agree on is goals. The club model makes both teams scoring 63% against the 60% in the price, and over 2.5 at 1.70 comes out at 59%. Villa are WLDLL, Brentford WDDDW.

The club model says 60% and the Intel page says 60%, but the match model gives Chelsea only 49%, so the price is on the short side. Chelsea's form is LDLWW, Bournemouth's LDDDL, and three of the last five meetings were draws. Chelsea win is still the call, just not one to lean on hard.

United at 1.70 is the lean on the result: the Elo model has them at 64%, but the match model only 51% against a price implying 56%, so we are not calling it. The firmer number is goals. Both teams scoring sits at 66% in the club model against 62% in the price, and the Intel page has over 2.5 at 63%, even with Spurs scoring just twice in their last five.

Four of the last five meetings were draws, and the Intel page flags the draw at 27%. Palace edge it on every model we run, 38% to 41% against a price implying 36%. Forest come in on LWDDL and start the second-toughest five-game run in the league here. The tightest call on the card.

The result is a coin flip: Everton 2.05, Hull 3.60, with our models putting Everton between 37% and 44%. The goals line is clearer. The club model makes under 2.5 a 57% shot against the 52% in the price, the Intel page has over 2.5 at only 43%, and both sides have averaged 1.2 goals a game over their last five.

City come in on five straight wins, Liverpool on WDWDD. Every model we run has City ahead, 41% to 46%, against a price implying 39%. Expect goals: the club model has both teams scoring at 65% and over 2.5 at 61%. Liverpool also start the toughest five-game run in the league here.

The club model has Newcastle at 60% and the Intel page agrees, against a price implying 45%. Coventry have lost four of their last five and scored once in those games. The one dissenter is the match model, which makes Coventry at 3.25 a 36% shot and calls that the value. We side with the ratings.

Sporting's rating is 120 points clear of Braga's and they arrive on three straight wins. The Intel page has the away win at 47% against a price implying 41%. Goals are the other angle: both teams have scored in four of the last five meetings, and the club model makes both teams scoring 61% at 1.70. That leg sits in the Accumulator King's Friday ticket.

Bayern carry the second-highest rating of any club we track and have won four of the last five meetings. The Intel page has over 2.5 at 75%. One caution: the match model only makes Bayern 68% against the 80% the price implies, so 1.18 is short even for them.

Five wins from five, 4.8 goals a game in those five, and the top rating of any club in our nine leagues. The 1.07 is no price at all. The club model's angle is both teams to score at 2.34, which it makes 47% against the 40% implied, though Getafe have scored in only one of the last five meetings.

Inter are DWWWW with three goals a game over that run, Parma WLDLL. Inter have won three of the last four meetings and drawn the other. The match model makes it 77% against the 83% implied, so the price is a shade short, but there is no case for anything else.

Madrid at 65% on the match model and 67% on the Intel page against 69% implied: a fair price. Villarreal arrive on three straight defeats, and Madrid have won four of the last five meetings with a 4-4 draw the other. Both teams scoring runs at 62% on the club model at 1.44.

The Intel page has PSG at 82% and over 2.5 at 78%. Our read on Le Mans is thin, with only five rated games in the model, which is why the match model lands lower at 71%. The club model's price angle is both teams to score at 2.07, which it makes 54% against 45% implied.

The biggest edge the match model finds all weekend: Como at 46% against the 37% in the price. Como are LWWWD and the last four meetings split two each. Against that, Roma arrive on DWWWW and are rated 49 points higher, so the Intel page keeps its confidence low at 36%. A call with the model, not a sure thing.

Milan at 1.80 is the lean on the result (WDDWW, 54% on the Intel page), but the match model gives it only 45% against 53% implied. The stronger number is goals: both teams scoring is 65% on the club model and on the Intel page against 55% implied, and it has landed in three of the last five meetings, including a 3-3 and a 2-5.