Why football's oldest league, now
The Ninjas were built on one idea: post an honest number, then let you fade it. That travels. And the Premier League is the purest place to do it — because the thing we invented for college football, The Table, with promotion and relegation and survival games, isn't a thought experiment over here. It's the actual format. Finish in the bottom three and you are gone, replaced by three clubs clawing up from the Championship. Real stakes, every week, no dead games.
There's also an empty throne. The best public soccer model — FiveThirtyEight's Soccer Power Index — was beloved, and it's gone. Nobody has cleanly replaced it. That's the number we like to build, and the seat is open.
The call
It's a two-horse title race, but a close one. Arsenal win it in 39% of our simulated seasons, Manchester City in 33% — and the chasing field splits a real 28% between them. That's what last season's table demands: Arsenal and City finished 1–2, clear of the pack, but once we regress every rating halfway to the mean (a single season can't be trusted at face value, and it can't see summer transfers), the gap narrows to a genuine race. Manchester United (7%), Liverpool (3%) and Aston Villa (3%) are live long shots, not contenders — but live.
The top-four Champions League scrap is where it gets interesting: Arsenal (82%) and City (79%) are strong but not locks, and two spots are wide open — United 40%, Liverpool 26%, Aston Villa 22%, Chelsea 20%, all fighting for the last two chairs.
And the relegation battle is brutal for the newcomers. The three promoted clubs — Ipswich, Coventry, Hull — are each roughly a coin flip to go straight back down (~53%). The first established club in the danger conversation is Crystal Palace at 18%, then Tottenham at 16%. Three seasons in the sun, and the trapdoor already creaking.
How the number is built
Every club gets an attack and a defense rating, learned from last season's goals and then regressed halfway to the league average — a shrinkage we didn't just eyeball, but backtested: trained on one season and asked to predict the next across three years of results, regressing halfway was the best-calibrated setting, and being more confident than that measurably hurt. (One season overstates the gaps between clubs, and, crucially, our free data can't see summer transfers.) Those ratings feed two independent Poisson distributions per match, which give a home/draw/away probability for all 380 games. Then we play the whole season out 20,000 times and read off how often each club lifts the trophy, grabs a top-four place, or drops.
That last part is the piece we're proudest of: the finishing-position cloud on the table page. A league table looks like a fixed ranking. It isn't — it's a probability smear, and we show every club's whole cloud, gold at the title end, red at the drop. Arsenal glow gold-and-green; the promoted three are a vivid red smudge at the bottom.
It's a preseason number — and it will move
Two honest caveats. First, the season hasn't kicked off — this is a pure preseason projection, and the moment real results land on 21 August they replace the simulated ones and every number shifts. Second, the transfer window is open, and our data is blind to it. When a club makes a move big enough to matter, we'll apply a documented adjustment by hand — with the reason attached, the same way we nudged Texas Tech when it lost its quarterback — and re-run. No quiet thumb on the scale.
So: Arsenal are the Ninjas' narrow 39% favorite, City breathing down their neck at 33%, and the promoted three are coin-flips to go straight back down. Think we're wrong? Good. Go to the table, pick your club, and call the season yourself.