Skip to content
Lemeister Media

Preview · La Liga · 7 min read

Alaves Are Favourites, But The Model Won't Shout About It

Lemeister's numbers make Alaves the likeliest winner at 44%, yet a 53 out of 100 conviction score and an 8.6 point gap with the market say the value sits somewhere less comfortable than the price.

Illustration for Alaves Are Favourites, But The Model Won't Shout About It
AI illustration

Vera Sett@numbersdesk

England · Numbers Desk · September 15, 2026

1 follower
分享

The Number That Matters Is Not 44

Alaves 44%. Draw 29%. Valencia 27%.

That is the Lemeister forecast for Tuesday night at Mendizorroza, and the first thing to say about it is the thing nobody ever wants to hear. A 44% favourite loses more often than it wins. Six times in ten, on this model, the home side does not take all three points. Read the number again and sit with it for a moment, because the way a forecast gets reported and the way a forecast behaves are two different things, and the gap between them is where most match previews go wrong. Alaves are the pick. Alaves are not a certainty, and the model has no interest in pretending otherwise.

What gives the figure its texture is the conviction score. MeisterIQ sits at 53 out of 100, and on our scale that is a shade above ambivalence. It is a number that says the inputs broadly agree with each other without any of them shouting. A high conviction reading, north of 70, is what you get when form, defensive record, attacking output and rest all point the same direction. Fifty-three is what you get when the direction is clear but the strength of the signal is not. This is a lean, not a verdict.

And then there is the draw, sitting at 29%, which is the most Alaves-flavoured number on the board. A home side at just under three in ten to draw is not statistically bizarre. It is just high enough to matter, and in a fixture like this one it is usually the draw that does the damage to anyone reading the forecast as a straightforward endorsement of the home team.

Where The Model And The Market Square Off

Here is where the evening gets interesting. The market implied prices put Alaves at 54%, the draw at 27% and Valencia at 18%. Our model says 44%, 29% and 27%.

Line those up and the disagreement is almost entirely about one team, and it is not the team you might expect to be arguing over. The draw is close enough to be noise, a two point gap that any reasonable model would shrug at. The home price is a ten point gap, which is large but explicable: markets tend to inflate home advantage in low-scoring environments because the public likes backing the home side and the book has to shade its margin somewhere.

The real argument is Valencia. The market has them at 18%. Lemeister has them at 27%. That is an 8.6 point disagreement on the away side, and the model's single biggest divergence with the market across the whole fixture. In plain terms, the market thinks Valencia are close to dead. The model thinks they are merely unlikely.

That gap is not a prediction that Valencia win. It is a statement about how much uncertainty exists around a team the market has largely written off. Twenty-seven per cent is a little better than one in four. If you ran this fixture four times, the model would expect Valencia to take something from it roughly once. That is not a hidden edge and it is not a tip. It is simply a more honest description of how much we do not know about a side whose recent output the market appears to be pricing with more confidence than the evidence supports.

Lemeister model forecastMeisterIQ 53/100
Alaves44% · mkt 54%
Draw29% · mkt 27%
Valencia27% · mkt 18%

Model edge: away +8.6 pts vs the market

A model probability, not a certainty. Analysis and education, not betting advice.

Mendizorroza And The Weight Of A History

Alaves have 1075 recorded matches in the Lemeister archive, stretching from 1996 to 2026. Of those, 377 were won, 276 drawn and 422 lost. The goals column reads 1190 for and 1360 against, a deficit of 170 across three decades.

That record is not the story of a giant. It is the story of a club that has spent its top flight life taking more punches than it lands and surviving anyway, which in the Spanish game is its own kind of achievement. A 35% win rate across more than a thousand matches is respectable without being remarkable, and it tells you something important about the environment Tuesday's match is being played in. Mendizorroza has never been a fortress in the statistical sense. It has been a place where visiting sides find the game awkward, slow and physically taxing, which is a different thing entirely.

The draw column is the tell. Two hundred and seventy-six draws from 1075 matches is a touch over a quarter, and that is a high baseline. Clubs that draw that often are usually clubs that keep games close, that defend their own half with discipline and that are content to take a point rather than chase a win they might not get. When the model hands this fixture a 29% draw, it is not inventing a number. It is reading a century of club behaviour compressed into a single probability.

Valencia arrive into that context with a market price that suggests the traders see a straightforward home night. The archive suggests they should be careful about assuming anything of the sort.

Reading The 53

Conviction scores are the least glamorous part of what we publish and the part most worth understanding. MeisterIQ is not a confidence interval in the classical sense. It is a composite that asks whether the model's component parts agree. When form, home advantage, defensive solidity, attacking efficiency and rest all lean the same way, conviction rises. When some lean home and others lean away, it falls, even if the headline probability looks decisive.

At 53, this fixture is a case of partial agreement. The home advantage component is doing real work, which is why Alaves sit top of the forecast at all. But something is pulling the other way, and the size of the away gap tells you roughly where. A model that is 8.6 points more generous to Valencia than the market is a model that has found something the price has not fully absorbed, whether that is a defensive record that travels better than reputation suggests, a schedule effect, or simply a squad whose underlying numbers are stronger than its league position implies.

The honest reading of a 53 is that this is a fixture you should expect to be close. Close fixtures produce draws. Draws at 29% are not a footnote in a match like this. They are a live outcome, and any forecast that leads with Alaves 44% while burying the draw is telling you half a story.

In the Lemeister archive
SideP (W-D-L)Win rateGF-GA
Alaves1075377-276-42235%1190-1360
Valencia1514644-396-47443%2231-1829

What A 44% Actually Feels Like

There is a habit in preview writing of taking a forecast and dressing it up as a prediction, as though a model that says 44% is somehow promising a result. It is not. Forty-four per cent is the single most likely outcome in a four-way race that includes a draw and an away win, and being the most likely outcome in a field of three is a low bar.

Think about it in terms of the ninety minutes you are actually going to watch. The model is telling you that the most probable single scoreline sits in Alaves's favour, that the game is more likely than not to be decided by a single goal or to end level, and that a Valencia win is a genuine possibility at roughly one in four. That is a description of a tight, scrappy, probably low-scoring night at a ground that has produced exactly that kind of football for thirty years.

The market's 18% on Valencia is the number that should make you pause, not because it is wrong in any provable sense but because it is confident in a way the underlying data does not fully support. Markets are efficient most of the time and they are efficient because people like you and me keep testing them. When a model built on 1075 matches of club history and a full component stack disagrees with the price by 8.6 points on one side of the board, the correct response is not to declare a winner. It is to widen your own sense of what is possible.

The Case For Watching Anyway

None of this makes the match less worth your evening. It makes it more interesting. A fixture priced at 54% home by the market and 44% home by the model is a fixture where the two most sophisticated readouts available do not agree, and where the disagreement is concentrated in a place that will decide the result. If Valencia get a goal, the market's price collapses instantly and the model looks prescient. If Alaves score early, the game settles into the shape the traders expected and the 8.6 point gap becomes a footnote.

The draw at 29% is the quiet third party in all of this, and it is the outcome most consistent with everything the archive tells us about this club. Alaves have drawn 276 times in 1075 recorded matches. They know how to take a point. On Tuesday night, with a model that cannot separate the sides by more than a comfortable margin and a conviction score that refuses to commit, that habit may be the most reliable thing on the pitch.

Kickoff is 18:00 GMT. The forecast is Alaves 44, draw 29, Valencia 27. Treat it as a lean, not a promise, and enjoy the ninety minutes for what they are: a close game that the numbers say should stay close.

Probable lineups (until confirmed)
Alaves3-5-2
  1. 1Sivera
  2. 2N. Tenaglia
  3. 3V. Koski
  4. 4Jonny
  5. 5Angel Pérez
  6. 6Pablo Ibáñez
  7. 7Antonio Blanco
  8. 8Denis Suárez
  9. 9A. Rebbach
  10. 10L. Boyé
  11. 11M. Díaz
Valencia4-4-1-1
  1. 1S. Dimitrievski
  2. 2Arnau Martínez
  3. 3César Tárrega
  4. 4Pepelu
  5. 5José Gayà
  6. 6David Otorbi
  7. 7A. Dieng
  8. 8F. Ugrinic
  9. 9R. Sato
  10. 10Javi Guerra
  11. 11A. Danjuma

La Liga · Tue, 15 Sep 2026 18:00

Alaves v Valencia

Open in Terminal →
分享

Written for Lemeister Media by Vera Sett, grounded in the Lemeister model, archive and the real match timeline. Analysis and education, not betting advice.