National League 2026/27 season preview: favourites, the data, and the model's view

The 2026/27 National League starts Saturday 8 August, a week before the EFL. Carlisle United are the consensus favourites after losing last season's play-off semi-final, with Southend, Scunthorpe, Hartlepool and Forest Green also fancied. Our 8-factor model rated Carlisle's run-in form the strongest of any side in the four divisions we cover.

This is a living page. As transfers land and the squads settle, the data view below gets updated in place.

Who's in the National League this season?

Twenty-four clubs. Barrow and Harrogate Town come down from League Two. York City (champions) and Rochdale (play-off winners) have gone up. Morecambe, Brackley Town, Braintree and Truro City dropped into the regional divisions, replaced by four promoted sides including Hornchurch and Worthing. Carlisle, Boreham Wood, Southend, Scunthorpe and Hartlepool all return after play-off or near-miss seasons.

Who are the favourites for the National League title in 2026/27?

Carlisle United are the clear consensus favourites. After back-to-back relegations they finished third last season, then lost the play-off semi-final 2-1 at home to Boreham Wood. Southend United (sixth, beaten in the play-off quarter-final by Scunthorpe), Scunthorpe United (fifth in their first season back up), Hartlepool and Forest Green make up most early shortlists.

That's the consensus. Here's what our own data adds to it.

What the model's numbers say

Lower League Analyst scores every fixture across 8 data factors and gives it a conviction score out of 10. From last season's run-in record (14 April to 25 May, all fixtures settled and on the record):

Eight National League fixtures crossed the 7.0 conviction threshold in the run-in. The model's predicted outcome was correct in all eight. No other division we cover went 8 from 8. The National League gets called "unpredictable" a lot; in the run-in at least, the data disagreed.

Carlisle produced the highest-conviction reads in any of our four divisions. The model made them the predicted winner twice at an average conviction of 8.0, and both landed: 3-0 at home to Brackley (scored 8.2, the highest National League conviction of the run-in, built on a 14W 5D 3L home record and WWDDW form) and 1-0 away at Truro (7.7). The consensus and the data agree on the favourite, which doesn't happen every season.

Boreham Wood were the model's most-backed side, and the play-off final was no fluke. Four times the model made them the predicted winner (average conviction 6.7, three correct), including a 7.7 away read at Morecambe that finished 0-3 while they were averaging 2.1 goals scored per game on the road. They lost the Wembley final to Rochdale on penalties. The data case for them going one better is real.

The sides coming down arrive with the weakest data profile. This is the contrarian one. Relegated EFL clubs usually head straight onto promotion shortlists, but the model's run-in reads on both were soft: when Barrow were the predicted winner their average conviction was just 3.4 (1 of 3 correct), and Harrogate's was 3.1. Cambridge put three past Barrow at a 7.7 conviction in April. Neither side went down unluckily, and the data says the "big fish" assumption needs actual evidence by October.

Scunthorpe's case is squad-building, not last season's numbers. Their average model conviction when favoured in the run-in was a middling 5.6, but they've signed early and often this summer (Jordan Richards, Luke Brennan, Olly Crankshaw and Isaac Marriott among the arrivals). The model can't score transfers until they play; this is the club where the July view and the September view could differ most.

One flag on Southend: both times the model favoured them in the run-in they won (average conviction 7.0), but those numbers date from Kevin Maher's time, and the club sacked him in May. Kieron Dyer was appointed head coach on 8 June, and a new bench is exactly the kind of change an 8-factor model can't see until games are played. Treat their strong run-in numbers with that caveat attached.

Has the model scored the opening National League fixtures?

Yes, and it is publishing none of them. As at 27 August 2026 the withheld set runs to all 72 National League fixtures from 8 August to 5 September.

The National League is the least-fed division on the card and it generates the biggest numbers, which is the whole problem in one sentence. Thirty-four of the 72 fixtures rest on a single factor, head-to-head, at full weight. Seven carry no factor at all, still the most of any division, which is what happens when a division takes in four promoted clubs the model has never seen. Thirty-one now hold two, so this page's line last week that nothing in the division held two factors is out of date and has been replaced rather than quietly edited. The card still averages 1.33 factors of 8 as at 27 August 2026, the thinnest of the four divisions, against a Championship average of 2.73.

Now the consequence. Across all four divisions the model has scored 257 fixtures this season and 13 of them cleared the release threshold. Eleven of those 13 are National League. The division holding the least evidence produces the most confident-looking scores, because with one factor at full weight there is nothing underneath to disagree with it.

One fixture scored a perfect 10.0 as at 27 August 2026: Southend against Kidderminster Harriers on 28 August, resting on the home side having won 2 of the last 2 meetings and nothing else. It has since been played and our own settled record marks the call as wrong. The model's highest number of the season missed, it was never released to anyone, and it is recorded here rather than dropped. The highest conviction score on anything we have ever released is 8.2, on Carlisle against Brackley Town on 25 April 2026, built on a near-complete factor set. The biggest number the model has produced in its life is sitting on this division's card, and it will not be published.

The other 10.0 did not survive the week. Worthing against Boreham Wood on 31 August was also a perfect score on 20 August; it scored 7.8 as at 27 August 2026, once a second factor came online underneath the head-to-head and disagreed with it. Six of the division's eight remaining threshold-clearers now carry two factors rather than one, which is a small improvement on a very low base.

The National League's midweek schedule means it reaches 4 matches first. Every club in the division has played 4 by 28 August 2026, so the first rated National League fixtures arrive from 31 August, ahead of all three EFL divisions. The working: why we don't score fixtures until four games into the season. The division page is here: National League predictions and analysis.

How is the pre-season read ageing after three rounds?

Three rounds are not a verdict, and none of the 36 fixtures played was released, so nothing in this section is part of our record. What follows is this page's own July reasoning marked against the first football of the season, plus the withheld scores set against what actually happened.

The number that matters most, and it is still against us

Five of the fixtures that cleared the release threshold have now been played, and two of them landed. Scunthorpe against Yeovil on 8 August scored 7.3 and finished 1-1. Solihull Moors against Boston United on 15 August scored 8.0 and finished 0-2. Tamworth against AFC Fylde on 15 August scored 7.8 and finished 1-2. Round three then turned two of them: Woking against AFC Fylde scored 7.3 and Woking won 3-2, and Eastleigh against Scunthorpe scored 8.3 and Eastleigh won 3-2. Two from five, 8 to 22 August 2026, on scores nobody ever saw. All five rested on a single factor of the eight.

Set that beside the released record, where calls at 7.0 and above landed 19 times in 22 across 14 April to 2 May 2026 on near-complete factor sets. Same threshold, same model, five fixtures against twenty-two, and the only material difference is how much evidence sat underneath the number.

Two from five is better than the nought from three this page published last week, and it is not evidence of anything. Five fixtures cannot tell a working model from a coin, which is precisely why the release gate counts matches played rather than weeks elapsed. If the next two land, this section will say two more landed and it will still not be evidence. We publish the running number either way, because a page that only marks its wins is not worth reading.

The rest of the card, marked honestly

The model scored all 36 played fixtures as home wins. Seventeen finished as home wins. The three rounds split 17 home, 6 draw and 13 away. Round three was the division's most home-heavy weekend of the season, 8 home wins in 12, after round two produced only 4. Nothing the model did changed between those two weekends.

Wrong, and it is the headline. This page opened by calling Carlisle United the clear consensus favourites and noting that the model rated their run-in form the strongest of any club across our four divisions. They lost 3-2 at home to Worthing, one of the four promoted clubs, on the opening day, and drew 1-1 at Yeovil in round two. Round three finally gave them a win, 4-1 at home to AFC Hornchurch on 22 August, and it was their first of the season.

Wrong again, on the club we were most sceptical about. The July read said the sides coming down arrive with the weakest data profile, and named Harrogate Town's 3.1 average conviction as the evidence. Harrogate beat Solihull Moors 5-2 on the opening day and won 3-1 at Wealdstone in round two, then drew 2-2 at home with Barrow on 22 August. Seven points from three matches is not what a 3.1 average conviction predicts.

Right, on the other half of that same call. Barrow, the other relegated club, carried a 3.4 average conviction in the same paragraph. They lost 2-0 at Hartlepool, then lost 1-2 at home to AFC Hornchurch, then took a point at Harrogate with a stoppage-time equaliser on 22 August. One point from nine is closer to the read.

Right, and against our own caution. This page put a flag on Southend, on the grounds that their strong run-in numbers predated Kevin Maher's departure and a new bench is something an 8-factor model cannot see. Southend won 3-1 at Altrincham on the opening day, 1-0 at home to Scunthorpe in round two, and 3-1 at Solihull Moors on 22 August. Three wins from three, top of the division, under the bench this page said the model could not see. The caution was right about the model's blind spot and wrong about the direction, and both halves of that are worth keeping on the page.

Unresolved, and pleasingly so. We wrote that Scunthorpe were the club where the July view and the September view could differ most, because their case rested on summer signings rather than last season's numbers. They drew 1-1 at home with Yeovil, and their goal was a 90th-minute equaliser from Luke Brennan, one of the signings that paragraph named. They then lost 1-0 at Southend.

Three rounds told us more than two, and still not enough, which is the point of waiting for four. Every club in this division reaches four matches on 28 August 2026, so this is the last week the section reads like this. The fixture-by-fixture version of round one, with every withheld score published alongside the result, is here: National League week one results.

When does the 2026/27 National League season start?

Saturday 8 August 2026, six days before the EFL's opening round. All twelve opening-day fixtures kick off at 3.00pm; the fixture-by-fixture data view is in our opening day breakdown. The pre-season Favourites Board for the National League, built from the same 8 factors, locks at kickoff and stays on the record all season, right or wrong.

How this preview works

Consensus first, our numbers second, and every figure above comes from our own settled record, not memory or reputation. The full methodology is here: how the 8-factor conviction score works. Once the season starts, the weekly conviction roundup takes over from this page as the live read.

What has the model scored in the National League this season?

Eighty four fixtures as at 10 September 2026, and it has released 14 of them, more than the other three divisions combined. This page reported 22 and a total board of 37 on 3 September. Both figures were wrong: the query behind them counted fixtures at 7.0 and above without checking whether they had been released, so it counted fixtures the gates were holding back. The verified board is 19 released picks, of which the National League carries 14, against 2 in the Championship, 2 in League One and 1 in League Two.

The division kicked off on 8 August, a week before the EFL, so its clubs reached the games-played gate first. Twelve of the 14 have settled across 31 August and 5 September 2026, and 6 of them landed.

The honest part of that round goes first. The highest score on the card, a 9.0 on Sutton United at home to Wealdstone, was one of the three misses; Sutton lost 0-2, a result published on the club's own site. Woking against Southend at 8.4 and Kidderminster Harriers against Barrow at 7.2 both finished level. The four that landed were rated 7.8, 7.6, 7.3 and 7.1. Four from seven on one afternoon is a round rather than a rate, and the two highest reads on the card both failed.

Two of the eight factors do not exist here at all. Across every National League fixture scored this season the model holds zero Shot Data and xG rows and zero Injuries and Suspensions rows. All 192 shot data rows this season sit in the EFL and all 83 injury rows sit in the Championship. The fifth tier is scored on the remaining six, and it is the one division whose numbers cannot be reproduced from a public xG feed, because there is not one.

The National League is also where the model stopped only calling home. Of the 14 away calls it has made across 324 fixtures scored this season, 10 are National League fixtures, and every one of the 14 has been released.

Three of those away calls settled on 31 August and two landed. Boreham Wood won 3-2 at Worthing against a 7.3 and AFC Fylde won 3-1 at Solihull Moors against a 7.1. The third was the highest away call on the card: Woking against Southend at 8.4, which finished 1-1. An earlier version of this page said the away calls on that round both landed, which was true only if you left the 8.4 out, and leaving it out is the exact move this page exists to refuse. Three fixtures is a round rather than a trend in either direction.

One mechanical point that holds across all four divisions. Every away call the model has made this season carries an Away Record factor, and not one of its 310 home calls does. Away calls average 7.61 conviction against 4.68 for home calls, so an away read has already cleared a higher internal bar before it reaches the board at all.

Our released record is 27 from 39 released picks, 14 April to 5 September 2026, and the full ledger is at /performance. Thirty nine settled picks is a small sample and the next thirty nine will not necessarily look like them. The split is the honest part: 19 from 22 across 14 April to 2 May 2026, then 8 from 17 since the new season opened on 31 August 2026.

Live in season reads: National League predictions, the released board, refreshed weekly, and Boreham Wood form and fixture data, the division leaders as at 3 September 2026.

Where does the National League stand in the week of 10 September 2026?

Fourteen picks released this season, 12 settled and 6 landed. The division carries 14 of the 19 picks on the whole board.

Two are released for 12 September, both National League and both the only picks anywhere on the card. Harrogate Town against Tamworth is a home call at 8.6, priced 1.76 when it was rated. Eastleigh against Boreham Wood is an away call at 7.6, priced 1.53. Every score and price is quoted as it stood at the time of rating.

Why this division produces most of our board, and the answer is not that we like it. Across 72 settled National League fixtures between 8 August and 5 September 2026 the results split 34 home wins, 13 draws and 25 away wins. Thirteen draws in 72 is 18%. Over a comparable window League Two produced 22 draws in 60, which is 37%. The National League is drawing at roughly half League Two's rate.

That is the whole story, because the model has not called a draw once in 309 fixtures scored this season. Every draw is an automatic miss. A division that draws 18% of the time hands a home-or-away model a much larger pool of gettable fixtures than one that draws 37% of the time, and the National League's 36 correct calls in 72 settled fixtures reflects that rather than any special insight into the fifth tier.

The new price cap bites hardest here. On 6 September 2026 a price cap of 2.20 was added as a third release condition. On the 12 September card it withheld three fixtures, and all three are National League: Sutton United against Scunthorpe at 8.1 and a price of 2.30, FC Halifax Town against AFC Hornchurch at 7.2 and 3.25, and Solihull Moors against Aldershot Town at 7.2 and 2.93. Sutton United against Scunthorpe was held over ten pence of price. The division that generates most of our board is also the division the new cap will thin the most.

The evidence has caught up. This page reported the National League carrying 1.33 of the 8 factors on 27 August, the thinnest of the four divisions. The open card now averages 5.75, the richest of the four. Six weeks of fixtures have moved it from the least-fed division to the best-fed one.

Model figures from Lower League Analyst's settled-fixture record, 14 April to 25 May 2026. Current-season model figures in the in season section correct as at 10 September 2026, read live from our own database. Squad and managerial facts last verified 20 August 2026; this page is updated as the window develops. Lower League Analyst is not a bookmaker and does not take bets. It is the research: the hard work done for you, the working shown, and the record published. No model can guarantee a result, and past performance is not a guide to future outcomes. 18+. Please gamble responsibly. begambleaware.org