Nominations are out! The jury selected a total of sixteen games for their longlist, nine for Spiel des Jahres and seven for Kennerspiel des Jahres. Out of these games, three games each are nominated for the two awards and can hope to get the coveted meeple added to their covers on July 13.
Let’s dive straight into the nominations and recommendations, before we discuss how good (or bad) my predictions were and if the agreed with our Kennerspiel score.
Nominated for Spiel des Jahres 2025
Bomb Busters
2–5 players, 30 minutes, 10+ years, medium light (2.0), 54% Kennerspiel, prediction #1
Players defuse bombs by deducing and cutting unseen wires of different values.
Bomb Busters was probably the top favourite pretty much on everybody’s list, and the jury agreed. It did not agree with the algorithm’s prediction that this is a Kennerspiel (more on that below).
Flip 7
3–99 players, 20 minutes, 8+ years, light (1.1), 99% Spiel, prediction #21
Score the cards in front of you, or take another card, risking everything you hold?
By their own claim “the greatest card game of all time”, by some other folks’ opinions dull and random.
Krakel Orakel
2–8 players, 30 minutes, 10+ years, light (1.1), 100% Spiel, prediction #7
Can you spot and reveal pictures in a tangle of lines?
Reportedly a drawing game which is still fun for people who cannot draw. With Pictomania, there’s some strong competition in that genre.
Spiel des Jahres recommendations
- Agent Avenue (prediction #8)
- Castle Combo (#4)
- Cities (#29)
- Foxy (no prediction): The inevitable “game that got away” and wasn’t on my prediction list at all.
- Perfect Words (#43)
- The Animals of Baker Street (#19): One of those late arrivals which is probably already forgotten internationally. I think it’s quite charming and I’m happy it got some recongition by the jury.
Nominated for Kennerspiel des Jahres 2025
Endeavor: Deep Sea
1–4 players, 60–120 minutes, 10+ years, medium (2.9), 100% Kennerspiel, #8
Lead your research institute to explore ocean depths and publish your findings!
Interestingly, this is a re-re-implementation of Endeavor, which was recommended back in 2010. The jury has a bit of an inconsistent attitude towards remakes.
Faraway
2–6 players, 15–30 minutes, 10+ years, medium light (1.9), 67% Kennerspiel, #3
Explore a mysterious land and fulfill quests on your way back.
The nomination that allows you to tick off the “that’s supposed to be a Kennerspiel?” on your bingo card.
Looot
2–4 players, 35 minutes, 10+ years, medium light (2.1), 92% Spiel, #30
Gather resources and capture buildings to develop your fjord.
Not going to lie: Looot’s description sounds like the most generic Eurogame imaginable.
Kennerspiel des Jahres recommendations
- The Gang (#1 Spiel prediction): I hadn’t played this one when I made my predictions, but did so in the meantime, and really had a blast. It also made me realise that the algorithm probably got its Kennerspiel score wrong, and had to post about it lest my insights go without receipts… 🤓😉
- Kauri (#45)
- Medical Mysteries: Miami Flatline (#27)
- Zenith (#99)
How good were the predictions?
This really is a drama in three acts: the decision of Kennerspiel or not, the algorithm picking the longlist, and finally my guts picking the shortlist.
Kennerspiel score
The algorithm classified thirteen out of sixteen games on the longlists correctly (Foxy was completely absent from the predictions, but its Kennerspiel score in the database is 1%), with Bomb Busters actually on the red list (as predicted in the last article), and Looot and The Gang on the anthracite list (as predicted on social media). 81% accurate isn’t bad for a simple model classifying a vague concept.
Longlist
This was a pretty poor year for the algorithm (again). It caught correctly three out of nine games on the Spiel longlist (plus one from the other list) and two out of seven on the Kennerspiel longlist (again plus one from the other list). Just 31% accurate is really bad, and even 44% accurate when counting the games from the wrong lists isn’t great. I thought the changes to the algorithm were working great, but clearly there’s still a lot of room for improvement.
I actually thought this year, I’ll also calculate some “proper” performance metric, which is one of the standard tool to evaluate how good a given ranking is: the Normalized Discounted Cumulative Gain (nDCG). I won’t get into the details here, but the general idea is to rank item (as we did with our predictions), then score them according to their relevance (in this case we give all games on the longlists a relevance of 1, with all other games getting 0). The higher our ranking sorts the recommended games, the higher the nDCG, and the better we consider the ranking. So, here are the results of the past years:
- 2021: 0.655
- 2022: 0.795
- 2023: 0.669
- 2024: 0.803
- 2025: 0.718
So, in that respect this year was a pretty average year for the predictions. It got somewhat redeemed by the misses being overall still pretty high on the full prediction lists.
Shortlist
Again, a poor performance: I’ve only predicted two out of six nominations correctly (one on the wrong list), matching the (bottom) performance from last year. Have I lost my edge? Well, I know it won’t stop me from coming back next year with more predictions.
That’s it for now, we’ll be back soon with a little more in-depth analyses on the six nominees and their chances of winning an award.