Eight Matchweek 3 fixtures are on our board, running from Crystal Palace v Manchester City on Friday night to Aston Villa v Arsenal on Monday. Only one of the eight carries a Strong rating: Liverpool at 74% against Nottingham Forest. Five of the eight are rated under 55%, which is us saying plainly that this is a hard matchweek to call. Our Premier League record so far is four wins and two losses from six graded picks, part of an overall 25–14 from 39 graded picks, 64%.
Every Matchweek 3 pick with its confidence score
All times GMT. Each pick is published before kick-off and graded against the official full-time result afterwards. A draw counts as a loss, so a Close call rating is a real warning rather than decoration.
| Kick-off | Fixture | Our pick | Conf. | Rating |
|---|---|---|---|---|
| Fri 28, 19:00 | Crystal Palace v Manchester City | Manchester City win | 58% | Good |
| Sat 29, 11:30 | Liverpool v Nottingham Forest | Liverpool win | 74% | Strong |
| Sat 29, 14:00 | AFC Bournemouth v Everton | Bournemouth win | 45% | Close call |
| Sat 29, 14:00 | Coventry City v Hull City | Coventry City win | 51% | Close call |
| Sat 29, 16:30 | Tottenham Hotspur v Newcastle United | Tottenham win | 44% | Close call |
| Sun 30, 13:00 | Chelsea v Brighton | Chelsea win | 63% | Good |
| Sun 30, 13:00 | Leeds United v Brentford | Brentford win | 42% | Close call |
| Mon 31, 19:00 | Aston Villa v Arsenal | Arsenal win | 56% | Good |
Liverpool v Nottingham Forest is the only Strong pick
Liverpool at 74% is the single pick this matchweek above our 70% threshold, and it comes straight after they cost us. We backed Liverpool at 49% away at Newcastle in Matchweek 2 and the match finished 2–2, which grades as a loss under our rules. The number is far higher this time for the obvious reason: home, 11:30 kick-off, and a Forest side that has to defend for long spells at Anfield.
The context worth knowing is that our Strong band has been graded seven times this season for six wins and one loss, an 86% strike rate. That one loss is the reason the next section exists.
Coventry City v Hull City is the fixture we owe an explanation on
On 22 August we rated Manchester United at 71% away at Hull City. Hull won 2–0. It remains the only Strong pick we have lost all season, and the reason was that our model was still pricing United on last season’s squad rather than on current evidence, while promoted sides in August have almost no current-season data to price at all.
Coventry v Hull puts that problem back in front of us in its purest form, with two recently promoted sides facing each other and Hull arriving with a genuine result behind them. We have rated it 51%, effectively a coin flip with a slight lean to the home side. After what happened at Hull, a confident number here would not be honest.
Chelsea backed again after winning at Fulham
Chelsea at 63% against Brighton follows a 49% call on them winning at Fulham on 24 August, which came in 3–2. Backing them again at a higher number is not momentum chasing: the fixture is easier, it is at home, and Brighton have been the more inconsistent of the two.
Crystal Palace v Manchester City, 58%
City won 2–1 against Bournemouth in Matchweek 2 off a 66% call from us. Palace at home on a Friday night is a harder assignment, which is why the number is eight points lower. Palace also lost 2–0 at Everton in Matchweek 2, a result that won us a 44% pick on Everton and says something about how Palace have travelled.
Tottenham v Newcastle, 44%
The lowest-rated pick of the weekend on a fixture between two sides who both drew their most recent match we graded. Newcastle held Liverpool 2–2. We lean Tottenham at home and we are telling you openly that we expect to lose this one more often than win it.
Aston Villa v Arsenal, 56%
Monday night, and the pick most likely to be revised if team news changes. Arsenal beat Coventry 3–0 in a match we rated at 82%, our highest confidence score of the season and a winner. Villa away is a different proposition entirely, hence 56% rather than anything close to that.
What our Premier League record actually looks like
| Matchweek 2 pick | Result | Conf. | Graded |
|---|---|---|---|
| Manchester United at Hull City | Hull 2–0 | 71% | LOSS |
| Manchester City v Bournemouth | City 2–1 | 66% | WIN |
| Chelsea at Fulham | Chelsea 3–2 | 49% | WIN |
| Liverpool at Newcastle | 2–2 draw | 49% | LOSS |
| Everton v Crystal Palace | Everton 2–0 | 44% | WIN |
| Premier League, season to date | 4 won, 2 lost | 6 picks | 67% |
Two things stand out. Our one Premier League loss that was not a draw came on our highest-rated pick of the round, and our two wins at 49% and 44% were both matches we rated as near coin flips. Six picks is not enough to draw conclusions from, which is exactly why we publish the number rather than a marketing claim.
Also on the board this weekend
- Bundesliga: Bayern Munich v VfB Stuttgart on Friday at 18:30 is our highest-confidence pick of the entire week at 75%, followed by seven Saturday fixtures at 18:00.
- La Liga Matchday 3: Real Madrid at 84% against Malaga and Barcelona at 78% against Rayo Vallecano are the two biggest numbers on the board.
- Serie A Matchday 2: four fixtures spread from Friday to Monday, with AC Milan at 76% the standout.
- Betway Premiership: a midweek round on Wednesday, covered in full in our Richards Bay v Kaizer Chiefs preview.
Follow the picks this weekend
- Today’s football betting tips, refreshed each morning
- Bet of the day, one pick only, and no-bet days are declared
- Correct score predictions, with the honest hit rate attached
- Midweek results: six wins from six graded picks
- Draw no bet meaning, and why draws cost us more than wrong winners
How these picks get graded
- A pick wins only if the named side wins the match. A draw counts as a loss.
- Picks are published before kick-off with the confidence score attached and are never edited afterwards.
- Wins and losses both go on the public record. Eleven picks have been voided this season where a fixture could not be verified as played.
- The full running record sits on the home page results board and is updated after every round.