THE RANGE

WHAT THIS IS

A grader. You bring the lineups; nothing here builds one. Each lineup is scored on 2,500 simulated Sundays and dropped into a fake Milly on each of them. Where it lands pays what the real ladder pays at that place. The average over the Sundays is its theoretical ROI.

THE SUNDAYS

TwoGunGPT simulates every game 10,000 times. THE GUNSMITH gets 5,000 of them: half to build, half to grade. This page reads only the grading half, the 2,500 Sundays no lineup on this site was built on. A lineup you built elsewhere never saw any of them.

THE FIELD

A fake Milly with as many entries as the newest real $20 Milly on file. Its lineups are drawn from our Milly ownership and stack and spend the way this season’s real Millys did. The page names its size and its ladder’s week. Our fake field has fewer twin lineups than a real one (under 1% of entries against about 10%), so a chalk lineup keeps a little more of its prize here than it would on DraftKings.

THE MILLY NUMBER

Jackpot in, smoothed. A Sunday in the top 1% pays the average prize of the top 1% places (about $900 on the $20 Milly, with $1,000,000 to first), not the exact place’s prize. Why: 2,500 Sundays can’t tell 1st from 500th for one lineup. Graded by the exact place, two independent sims of the same lineup agreed at a correlation of 0.00 (one win read +4,000%, no win read at the field); graded this way they agree at 0.6 to 0.9, and a set’s number comes out the same either way within its ±. A lineup’s place inside the top 1% is close to uniform, measured on 630,000 real Milly lineups. The 1ST column still counts the exact wins. Read the set’s number first; the ± is two standard errors.

THE POWER SWEEP NUMBER

The $150 Power Sweep ($50K to first, 5 entries max), graded in a second fake field: the sharp kind, about 1,500 entries drawn from our small-field ownership and stacked the way this season’s $555 Millys, Wildcats and Power Sweeps stacked (fewer naked QBs, more two-catcher stacks, more bring-backs than a Milly), paid by the Sweep’s own, flatter ladder. First place there is 333 times the entry, so the number is steady, and it is the gauge: if it ever reads +80% for a set, the sim is being too kind. Lineups built on these same sims, like the Gunsmith’s, lean their way here.

THE RAKE

RAKE OUT (the default): the rake is paid whatever you enter, so it isn’t a decision. The field’s average lineup sits at 0%, and beating 0 means beating the field. RAKE IN is the real-dollar number: the field’s average lineup sits at the rake. The two differ by a fixed factor, the contest’s net payout.

THE OTHER COLUMNS

1ST counts the Sundays the lineup lands first. TOP 1% and CASH are the share of Sundays it finishes in the top 1% and above the cash line; a tie at a line counts as paid. FINISH is its median place. MED is its median score. OWN adds up our projected ownership of its players.

THE ±

2,500 Sundays is a lot, not infinite. Every number carries a ±: two standard errors, about a 95% range. The set’s ± counts lineups that win together once, so it can be smaller than any one lineup’s. A gap inside its ± is noise.

THE DUEL

Your set against THE GUNSMITH’s default 20 on this board: its opening dials, no locks, a fixed seed, built once per field so everyone meets the same 20. Each Sunday, the average score of your lineups against the average of theirs. “Yours outscore them on 44 of 100 Sundays ±2” is that count, with its ±. A tie isn’t a win. The Gunsmith’s 20 were built on these same sims, so their numbers lean their way.

THE REPORT CARD

Under the tool. Every real $20 Milly lineup of the weeks on file, graded by this sim before lock the way it grades yours (the newest board and ownership posted before kickoff, the fake Milly on that week’s own ladder), then what each one did: cashed or not, top 1% or not, its real prize. By bucket of what the sim said, the misses first. Lineups with a player our board had out, unpriced or didn’t carry are counted and left out; when the week’s winner is one of them, the card says so. Bucket totals only; no lineup and no name leaves this PC.

THE CHECK

Every time a field is built, lineups drawn like the field itself but not in it get graded the same way: 300 for the cash line, and a second draw of 1,000 for the ROI, jackpot in, for the Milly and for the Sweep. They should cash about as often as the ladder pays and read near 0 rake out. The page prints those lines under the slate stamp. When one drifts, that field is off that week, and so is every number it feeds.

THE FILE FORMATS

DraftKings’ upload file: a header of QB,RB,RB,WR,WR,WR,TE,FLEX,DST, then one lineup a row, IDs or Name (ID). Your DKEntries.csv: the same columns after Entry ID, Contest Name, Contest ID and Entry Fee; the player pool under it is ignored, and the entry IDs ride along into the CSV out. Or type: one lineup a line, nine names split by commas or tabs, or DraftKings’ standings style with the slot before each name.

THE CSV OUT

Every lineup with its names, IDs, salary, MED, both rakes’ ROI, 1ST, TOP 1%, CASH and FINISH, the set’s row on top, and a note naming the board, the ownership and the ladder the grades came from.

Sims, not results. No number here is a prediction, and nothing here enters a contest for you.