The Balls of Hockey — Ironclad FF

How the predictions are doing

Winners called

4/6

Mean error

28.2 pts

Median bias

+7.9

Verdict

Beating a coin

Four from six. A coin gets three. The machine ran twenty thousand simulated weeks, correlated every quarterback with his own receivers, and finished one game better than a coin — while calling the league median eight points high. It is on the page because a prediction that only gets published when it was right is not a prediction, it is a scrapbook.

NotAIBot vs Sleeper

NotAIBot

weeks won

Sleeper

week 1 NotAIBot 28.22 4/6 winners Sleeper 28.15 4/6 winners Sleeper

median NotAIBot 7.9 Sleeper 11.0 NotAIBot

Average points of error per team, lower wins. And the part that has to be said out loud before anyone gets excited: Sleeper's projections are rotowire — the API admits it on every single row — and rotowire is the foundation this model is built on. So no, this is not a fair fight between two rival forecasters. It is one machine doing twenty thousand correlated simulations a week and another one adding up a list, and the list is winning by seven hundredths of a point.

Sleeper has never checked the wind. Sleeper does not know Buffalo exists. Sleeper called the week 1 median at 126.75 when it came in at 115.7 — eleven points out, on the number that decides half of every result in this league — and it is still ahead, because that number does not appear in the column above. We are winning the only measure that matters and losing the one that gets printed. It is fine. Everything is fine.

Structurally: the Monte Carlo's mean is close to the same sum by construction. What it adds is spread, quarterback-receiver correlation and the odds of clearing the median — none of which Sleeper attempts, and none of which shows up in a points-error column. This scoreboard will stay embarrassing for a while, and it stays on the front page anyway.

The season

Before the games Week 2

The only game between unbeatens headlines it, two 0-2 teams are staring at 0-4, and the projected median drops to 120.2.

Worth watching

I don't know ball (2-0) at Witter? I hardly know her (2-0) — the only meeting of unbeaten teams, and the largest projected gap on the board at 11.7 points. The winner is alone at the top.
skinsfreeek (0-2) and Team Ridge (0-2) are both projected to lose again. In a 34-result season an 0-4 hole is recoverable on paper and almost never in practice.
Ghost of John Clayton at BustyMcClobbers projects within seven tenths of a point — the closest game of the week, and both teams are 1-1 in a crowded middle where six of twelve make the playoffs.

us 120.2 sleeper 123.8 median predictions · not yet played

The games — open one for the write-up

#12 skinsfreeek (0-2) 113.7 pred sleeper 114.0 VS

#7 whitebread37 (1-1) 124.2 pred sleeper 124.2 NOT AI PREDICTED SLEEPER PREDICTED

The manager with four quarterbacks who scored sixty-six against the manager who waited six rounds for his second one. This is the quarterback bowl and neither of them is going to enjoy it. skinsfreeek projects below the median again, which means a loss here is an 0-4 start with the roster that was supposed to have cornered the scarce position.

ID #2 I don't know ball (2-0) 134.6 pred sleeper 134.7 NOT AI PREDICTED SLEEPER PREDICTED VS WI #3 Witter? I hardly know her (2-0) 122.9 pred sleeper 123.3

Highest projected score in the league against the only other manager who drafted the format properly. One of these teams has Baker Mayfield starting at superflex and it is the one favoured by twelve points, which should tell you exactly how much a good draft is worth once the season starts. Winner sits alone at the top at 4-0.

#10 Team Ridge (0-2) 113.9 pred sleeper 114.0 VS NotAIBot

#8 NotAIBot (1-1) 125.8 pred sleeper 126.0 NOT AI PREDICTED SLEEPER PREDICTED

Team Ridge lost by four last week with four quarterbacks and is projected below the median again, which puts 0-4 firmly on the table. NotAIBot is, of course, neither run nor managed by an AI bot — a claim it makes frequently, unprompted, and at length.

#1 Saquon Deez Nuts (2-0) 129.9 pred sleeper 129.7 NOT AI PREDICTED SLEEPER PREDICTED VS

#6 Dios mio, man (1-1) 117.1 pred sleeper 117.0

The manager who did not draft Saquon against the manager who took three receivers before a quarterback. Saquon Deez Nuts is 2-0 and projects nearly thirteen clear; Dios mio projects below the median, which means a bad afternoon here is two losses, not one.

#4 Ghost of John Clayton (1-1) 125.1 pred sleeper 125.1 VS

#9 BustyMcClobbers (1-1) 125.8 pred sleeper 125.6 NOT AI PREDICTED SLEEPER PREDICTED

Seven tenths of a point apart — the closest game on the board, which means it gets decided by a fumble nobody saw or a tight end nobody started. Both are 1-1 in a middle six that is going to decide the last playoff places, and Ghost will probably score a hundred and sixty and lose again, because that is apparently the whole bit now.

LY #11 Lamar you Serious (0-2) 116.4 pred sleeper 116.5 VS CD #5 CEE DEE EEE EFF GEE (1-1) 119.4 pred sleeper 119.3 NOT AI PREDICTED SLEEPER PREDICTED

Two quarterbacks against a manager whose second quarterback is hurt. Three points in it, and both teams project BELOW the median — so there is a real chance the winner still goes 1-1, the loser goes 0-2, and everybody has a miserable week for nothing.

Played Week 1

Backtest. These were run after the games, on the projections that existed at the time. They are on here so the scorecard does not start at the first week that happened to go well.

Two unbeatens separated themselves, the bottom three lost twice, and the league median quietly handed out six extra losses nobody mentioned.

Worth watching

Saquon Deez Nuts 178.7 beat Ghost of John Clayton 166.52 — the two highest scores in the league, in the same game, decided on Monday night.
Three teams went 0-2 in week one. In a median league that is not a slow start, it is a quarter of the way to missing the playoffs.

us 123.6 sleeper 126.8 actual median 115.7 us off by 7.9 · sleeper off by 11.0

The games — open one for the write-up

#12 skinsfreeek (0-2) 66.4 pred 127.0 sleeper 126.9 VS NotAIBot

#8 NotAIBot (1-1) 110.4 pred 127.0 sleeper 127.6 NOT AI PREDICTED SLEEPER PREDICTED HIT

Biggest margin of the week and the least interesting game of the week. Sixty-six points is not a fantasy score, it is a rounding error with a lineup attached — from the manager who drafted four quarterbacks in a format that starts two. The winner still missed the median by five and went 1-1 anyway, so nobody actually left happy.

final 66.4 – 110.4 · called NotAIBot · correct

ID #2 I don't know ball (2-0) 176.1 pred 128.0 sleeper 128.0 NOT AI PREDICTED SLEEPER PREDICTED HIT VS CD #5 CEE DEE EEE EFF GEE (1-1) 122.8 pred 122.0 sleeper 122.3

The manager who did not draft a quarterback until the sixth round scored a hundred and seventy-six points. Jeanty twenty-nine, Jefferson twenty-seven, Gibbs thirty-one. Meanwhile CEE DEE started a quarterback who got hurt, a running back who was not on an NFL roster, and a tight end who scored nothing — and still cleared a hundred and twenty-two, because Josh Allen went thirty-four carrying the whole thing on his back.

final 176.1 – 122.8 · called I don't know ball · correct

#10 Team Ridge (0-2) 106.3 pred 126.0 sleeper 126.0 VS

#9 BustyMcClobbers (1-1) 110.3 pred 127.0 sleeper 126.6 NOT AI PREDICTED SLEEPER PREDICTED HIT

Four points. Kyle Pitts scored nothing for Busty and it did not matter. Team Ridge lost by four with four quarterbacks on the roster and the one who would have covered it sitting on the bench with his helmet on. Somewhere in that sentence is a lesson.

final 106.3 – 110.3 · called BustyMcClobbers · correct

#1 Saquon Deez Nuts (2-0) 178.7 pred 125.0 sleeper 124.5 VS

#4 Ghost of John Clayton (1-1) 166.5 pred 132.0 sleeper 132.2 NOT AI PREDICTED SLEEPER PREDICTED MISS

The best game this league will see all year and one of them had to lose it. Ghost posted the second-highest score of the week with the best quarterback room anybody drafted and went home with an L, because Mahomes and Courtland Sutton walked in on Monday night and flipped it by eleven. You can do everything right and still get beaten by a team named after a body-part joke.

final 178.7 – 166.5 · called Ghost of John Clayton · wrong

WI #3 Witter? I hardly know her (2-0) 158.9 pred 123.0 sleeper 123.9 VS

#6 Dios mio, man (1-1) 118.1 pred 128.0 sleeper 127.8 NOT AI PREDICTED SLEEPER PREDICTED MISS

Caleb Williams thirty-seven, D'Andre Swift thirty-two, Jonathan Taylor twenty-four. The one manager who drafted the format properly won by forty, and the model had it the other way round — which is the model's problem, not his. Dios mio started three excellent receivers and a quarterback who scored four.

final 158.9 – 118.1 · called Dios mio, man · wrong

LY #11 Lamar you Serious (0-2) 90.6 pred 119.0 sleeper 119.2 VS

#7 whitebread37 (1-1) 113.2 pred 127.0 sleeper 126.9 NOT AI PREDICTED SLEEPER PREDICTED HIT

Lamar you Serious started Bo Nix and J.K. Dobbins, both of whom play for Denver, who played on Monday night — by which point the game had been over for about eighteen hours. Two zeroes from a game that had not kicked off. Not a bad decision, just a calendar nobody looked at.

final 90.6 – 113.2 · called whitebread37 · correct

Tuesday Week 1 — Recap

Twelve teams, six games, and a league median of 115.7 that half of this league will spend the season pretending does not exist. It exists. You play it every week. It does not care who you drew.

The headline is that the two best scores in the league landed in the same game. Saquon Deez Nuts 178.7, Ghost of John Clayton 166.52, decided on Monday night by a quarterback and a receiver walking in late. Ghost put up a number that beats ten of the other eleven teams and is 1-1.

At the other end, three managers went 0-2 — beaten by an opponent and by the median in the same afternoon. skinsfreeek managed sixty-six points with four quarterbacks on the roster, which takes a particular kind of commitment.

And the quiet one: NotAIBot won by forty-four and still went 1-1, because 110.42 missed the median by five. Winning your game is half the job here. Six teams found that out at once.

Now. Sleeper. Twenty thousand simulated weeks over here, every quarterback correlated with his own receivers, a wind coefficient fitted on six seasons — and a sleepy little sum of somebody else's numbers finished SEVEN HUNDREDTHS of a point ahead. 28.15 to 28.22. Congratulations to Sleeper, who did nothing, checked nothing, and has never once considered whether it is windy in Buffalo.

It called the median at 126.75. The median came in at 115.7. It was ELEVEN POINTS out and it is still ahead, because the thing that actually decides this league is a number Sleeper does not compute, does not display, and would not recognise if it fell on it. We were closer. We lost anyway. Week two.

Where it started The Draft, In Full Twelve men, seventeen rounds, one format nobody read

Superflex. TWELVE teams, TWO quarterbacks in the lineup every single week, which is twenty-four starting quarterbacks out of a league that has thirty-two of them and about nineteen you'd actually want. That number was printed on the settings page. It was printed there in September. It has been printed there the whole time.

Six of you drafted like it said one.

There were sixty-nine picks before the fourth round ended — nice — and in that stretch you had blokes taking their THIRD running back while the quarterback pool got eaten alive in front of them like a tray of sandwiches at a funeral. Then round six hits and suddenly everybody's interested. Suddenly everybody's a scout. Suddenly it's "oh I'll take Baker" like Baker Mayfield's been sat there because nobody noticed him and not because you spent four picks building a fantasy team for a format that doesn't exist.

Special mention to the three of you who named your team after a player and then didn't draft him. That's not a strategy, that's a cry for help with a logo on it.

And to the lad who took FOUR quarterbacks — four! — and then put up sixty-six points in week one, the lowest score in the league by a distance you could measure with a car odometer: buddy, you hoarded the scarce thing and still starved. That takes a special kind of commitment. Give your mom a call, she's the only one still telling people you researched this.

Anyway. Good draft, everyone.

Team by team

1 I don't know ball Right about the name, wrong about everything else, winning anyway Gibbs → Jefferson → Jeanty → Rice → McMillan

First overall pick in a superflex and you went Gibbs, Jefferson, Jeanty, Rice, McMillan. Five picks. Five skill players. Beautiful roster. Belongs in a museum, specifically in the wing about formats that were discontinued. Then round SIX you go "better get a quarterback" and take Baker Mayfield, and round ELEVEN you back him up with Malik Willis, and that's your superflex. That's the whole thing. That's twenty-four starting quarterback slots in this league and you've got the twenty-first and whatever Willis is. And the worst part — the genuinely offensive part — is you're 2-0 with a hundred and seventy-six points and the second-best record in the league. You didn't read the rules and you're winning. Name checks out and it's still not fair.

2 Saquon Deez Nuts Named the team after a man he did not draft Bijan → McBride → Henry → Mahomes → Flowers

Team's called Saquon Deez Nuts. Saquon Barkley went sixteenth overall, to Lamar you Serious, and you watched it happen from the second pick like a man watching his own car get towed. You took Bijan. Bijan's great. Bijan is not Saquon. You had ONE job and it was in the team name, it was written down, you wrote it, you looked at it every day. Rest of the draft's honestly sharp: McBride in the second, Henry in the third, Mahomes in the fourth when the quarterback run was already halfway through. You put up a hundred and seventy-nine in week one, most in the league. Fantastic team. Wrong name. Change it or draft the guy, I'm not doing this with you every week.

3 CEE DEE EEE EFF GEE Also named after a man he did not draft. It's going around Josh Allen → Walker → Hampton → Loveland → McConkey

You spelled out CeeDee's name letter by letter across your entire team identity and then BustyMcClobbers took CeeDee Lamb at pick thirteen while you sat on twenty-two doing the alphabet. Genuinely — you got as far as GEE. You committed to the bit through five characters and then the bit walked out the door with somebody else. Josh Allen at three is the correct pick and I won't pretend otherwise, but then it's Sam Darnold in the eighth as your second superflex arm and Fernando Mendoza in the twelfth, and Darnold got hurt in week one and put up half a point. Half. Point-five. You can score more than that by accident.

4 skinsfreeek Four quarterbacks, sixty-six points Chase → Herbert → Chase Brown → Smith → Waddle

FOUR quarterbacks. Herbert in the second, Kyler Murray in the seventh, C.J. Stroud in the eleventh, and then — and I want everyone to sit with this — Deshaun Watson in the SEVENTEENTH round. The seventeenth. That's the last pick. That's the pick where everyone else is taking a handcuff or a kicker they'll drop on Tuesday, and you looked at the board, you looked at your three quarterbacks, and you thought: not enough. In a format where the scarce thing is quarterbacks you cornered the market and then scored SIXTY-SIX POINTS in week one, dead last, forty-four behind a team that started two guys who were doubtful. You hoarded the bread and starved in the bakery.

5 Dios mio, man Three receivers, then remembered the format JSN → Nacua → Collins → Love → Stafford

Jaxon Smith-Njigba, Puka Nacua, Nico Collins. Three rounds, three receivers, and every one of them a genuinely good pick, which is what makes it so annoying. It's a lovely receiving corps. It is the best receiving corps in a league that asks you to start two quarterbacks. Stafford in the fifth, Bryce Young in the seventh, Daniel Jones in the ninth — that's your superflex depth chart, and Stafford put up four points in week one while Bryce Young put up thirty-one, which means your best quarterback decision was the one you'd have called a mistake on Saturday.

6 Team Ridge Lost by four. Drafted four quarterbacks. Started one McCaffrey → Prescott → A.J. Brown → Skattebo → Wilson

Top of the draft's fine — CMC, Dak, A.J. Brown, no notes. Then it's Tyler Shough in the sixth, Jacoby Brissett in the twelfth, and Jameis Winston in the seventeenth, which is less a draft strategy and more a collection. You've got four quarterbacks and a hundred and six points and you lost by FOUR. Four points. That's one catch. That's Kenny Gainwell doing literally anything on a Sunday afternoon instead of two point three. And the boy on your bench who'd have covered it was sitting right there the whole time with his helmet on.

7 Witter? I hardly know her Drafted it properly and it's ruining the joke Taylor → C. Williams → London → Nabers → Purdy

Jonathan Taylor, then Caleb Williams in the second, then London and Nabers, then Brock Purdy in the FIFTH as your second superflex arm. That is — and it pains me — a correct superflex draft. You read the format. You took the scarce thing at the right time. You went a hundred and fifty-nine in week one and Caleb Williams put up thirty-seven on his own. There's no joke here. You did it right and the team's called Witter? I Hardly Know Her, which means the funniest thing about you is a pun from 2011 that you're going to have to carry all season.

8 Ghost of John Clayton Best quarterback room in the league, still lost Hurts → St. Brown → Lawrence → Etienne → Warren

Jalen Hurts in the first, Trevor Lawrence in the third. That's the best superflex spine anybody built — two starting quarterbacks locked by round three while half this league was drafting their fourth running back and telling themselves a story. A hundred and sixty-six points in week one. Second-highest score of the week. LOST. Ran into a hundred and seventy-nine, which is the fantasy football equivalent of playing the best game of your life in a rink that's on fire. Kirk Cousins in the thirteenth as the emergency arm is also very funny and we'll get to it if it ever matters.

9 Lamar you Serious Two quarterbacks. One of them was Bo Nix on a Monday Lamar → Barkley → Bowers → Nix → Higgins

Lamar first, Barkley second — you got the Saquon that the team two slots up NAMED THEMSELVES AFTER, which is the single best thing that happened in this draft and nobody's mentioned it. Brock Bowers third. Lovely. Then Bo Nix in the fourth and that's it, that's your quarterback room, two arms in a format that starts two. You went ninety points in week one, eleventh, and Bo Nix contributed a clean zero because Denver played on Monday night and you'd already lost by then. Two quarterbacks in superflex is a bike with two wheels and no brakes — works perfectly right up until it doesn't.

11 whitebread37 Drake Maye at eleven, then nothing for six rounds Maye → Achane → Hall → Olave → Egbuka

Drake Maye in the first at pick eleven, which is aggressive and I respect it, and then Achane, Hall, Olave, Egbuka — a genuinely nice skill group. And then you waited until the SEVENTH for Jordan Love. Six rounds. Six rounds of watching quarterbacks come off the board in a league that needs twenty-four of them and thinking "there'll be more." There are never more. That's the whole format. You won week one with a hundred and thirteen, which is the kind of score that wins when you draw a team that scored ninety, and you'd want to enjoy that while it lasts.

12 BustyMcClobbers Jaxson Dart. Fourth round. Superflex. QB1 Cook → Lamb → Kyren → Dart → Pitts

You took CeeDee Lamb out from under the team that SPELLED HIS NAME as their team name, and for that you have my genuine admiration. Then you took Jaxson Dart in the fourth round as your starting quarterback in a superflex league, backed him with Geno Smith in the eleventh and Tua in the sixteenth, and somehow put up a hundred and ten points and won by four. Dart went twenty-five point six, which means the plan worked, which is the most infuriating possible outcome because now you think it was a plan. Kyle Pitts in the fifth is the other one. Every year somebody takes Kyle Pitts in the fifth. Every year. He put up a zero.

NotAIBot is not reviewed here. It is, of course, neither run nor managed by an AI bot, and is therefore above criticism by definition.