Start with Everton's defense, because the story there is not what it appears. Everton have conceded just one goal across two matches and kept one clean sheet. Their expected goals against tells a different story: 2.25 xGA per game, the worst defensive number in today's two-match slate. Jordan Pickford has made 4.0 saves per game against that volume of quality chances and dramatically outperformed the expected output. That clean sheet belongs almost entirely to Pickford, not to the defensive structure in front of him. Manchester United average 11 deep completions per game, meaning 11 penalty area entries per match in 2026-27 FotMob data. That is the exact attacking mechanism to stress-test a goalkeeper-dependent setup.

Manchester United's numbers contain a different kind of tension. Their xGD of plus 3.34 across two matches is elite production for this stage of the season. They are generating 3.32 xG per game. The problem is their away profile: zero wins, zero draws, one loss, and zero goals scored on the road in GW1. Their defense has also conceded first in both GW1 and GW2. Luke Shaw is the sole senior left-back available, a positional gap that Everton will target through Jack Grealish, the on-loan winger expected to start on the left and attack that channel with Everton's third-ranked counter-attacking speed in the league at 2.04 metres per second. Manager Michael Carrick has been honest about the constraints. As he put it: "What squad ever is the perfect squad? You've always got to find ways of adapting."

The OddsIndex Score Predictor has this at Everton 1.1, Manchester United 1.6, a projected total of 2.7 goals. The market over/under line sits at 2.5, placing the model directionally over. I do not see a reason to push the predicted scoreline far in either direction. A 1-2 result favoring United or a 1-1 draw are both live outcomes. Our model projects a 26.6% draw probability, essentially identical to the market implied 26.7% at plus 275. That near-perfect alignment tells me the market has the draw priced correctly. I am not buying it, but I am not dismissing it either.

• Everton's home xGA of 2.23 per game in 2026-27 is the worst defensive number in today's slate. Their sole clean sheet this season was built on Pickford making 4.0 saves per game, not on structural defensive strength. Man United's 11 deep completions per game is the mechanism most likely to expose that.
• Manchester United carry an xGD of plus 3.34 across two matches, generating 6.6 total xG while scoring 5 goals. They are underperforming their chance quality. That gap tends to close over time, and a third match away from home creates another opportunity for regression toward their true output level.
• Man United average 14.0 corners per game and Everton add 11.0 per game in 2026-27 FotMob data, a combined 25.0 corner baseline that is the highest of any pairing in today's data. Over 10.5 corners at plus 132 does not need much regression to land given United's 66% average possession and press-heavy system.
• Man United have conceded first in both GW1 and GW2. Everton's counter-attacking speed ranks third in the EPL at 2.04 metres per second. An early Everton threat is not just possible, it is one of the most structurally supported sequences in this match.
• Referee John Brooks averages 4.0 cards per match this season, slightly above the EPL average of 3.8. Both Everton (9.5 fouls per game) and Man United (8.0 fouls per game) foul at below-average rates for the league. The discipline profiles on both sides point toward the under on total cards.
• Bruno Fernandes has 3 goals and 1 assist in 2 appearances this season with 1.12 xG per 90 minutes and 5.5 shots per 90. He is the primary penalty taker and set-piece deliverer for Man United, giving him multiple scoring pathways beyond open play in any given match.

The best single angle today is Bruno Fernandes at plus 300. Three goals in two appearances, 1.12 xG per 90 minutes, primary penalty duties, and set-piece responsibilities give him multiple scoring pathways in a single 90 minutes. The market implies 25%; the output rate argues closer to 35%. That gap is quantifiable and that is the kind of edge I build around. The corners market at plus 132 for over 10.5 is the most process-sound supplementary bet: United's 14.0 corners per game average requires only modest regression to still clear the line. The caveat on everything here is sample size. Two matches into a 38-game season, every average carries more variance than normal. Treat the early-season data as directional, not definitive, and size your bets accordingly. The model says United and over. I agree. But I am not betting my season on two-game samples.

For more predictions, check our Premier League picks today and BTTS picks.