Pre-S9 scoring: 4/2/1/0 in 4-player groups, 4/2/0 in 3-player. S9 onward: 7/5/3/1 and 7/4/1. When comparing across eras, we multiply old scores by 1.75x to roughly equalize them. It's not perfect (last place was 0 pre-S9 but 1 now), so cross-era stats are always approximate.

Points are cumulative across the season. The 2 lowest single-meet scores are dropped before calculating your standing. That adjusted total is your rank. The ✓ next to your name is meets you played; a ✗ count means you missed that many.

Your Rating combines 5 stats. Win Rate matters most (30%), then Division Level (25%), Consistency (20%), Machine Skill (15%), and Clutch Factor (10%). Each stat is ranked 1-99 based on where you fall vs everyone else. Your Rating is the weighted average. You need at least 8 career meets (one full season) to get a Rating.

Career win rate = % of all groups where you finished 1st. This is the one on your stat strip and Rating wheel. Machine win rate = wins on a specific machine divided by total plays. These are two different numbers, so don't confuse them.

Your last 2 seasons average points per meet played vs your career average, also per meet played. A positive delta means you're playing better lately. Negative means a recent slump. Both halves count only the nights you actually turned up: a meet you missed was never a match up, so it is never in the divide.

Special meets (1-2 per season) where a non-A player earns a +1 bonus point for beating an A-division player head-to-head. A players can also "defend" against challengers, tracked separately.

Raw scores use median, not average. Pinball scores have wild outliers (one billion-point game can skew an average badly). Median gives you a more honest picture of typical performance. Win Score on machine detail is the all-divisions Target Score: the median winning score across every division, no-contest groups excluded, recent seasons weighted 2x. See Target Score below for the full method.

On the compare page, each shared machine gets a call based on the two median raw scores. We use the gap between them as a share of the bigger one, because 40 million is a landslide on some tables and a rounding error on others. Under a quarter apart we call it an even machine. A quarter to a half, someone usually wins here. Half or more apart, they own it. Both players need at least 3 plays or we say nothing at all. Checked against every pair who have a real head-to-head record on the machine: the even calls come out near a coin flip, "usually wins" lands about 4 times in 5, and "owns this one" lands about 9 times in 10.

First place on every single game of a night. Not most of them, not top of your group on total points, every game. A tie counts against you, and you have to have played the whole night. In 16 seasons it has happened 34 times out of about 6,700 player nights, which is roughly one night in 200.

"What wins on this machine" depends entirely on who is playing, so this is a per-division number, not a league-wide one.

Step 1 — find the winners. For every group that played the machine we take the highest raw score. A score only counts if it is above zero and was not disqualified or forfeited. If a group has fewer than 2 valid scores it was no contest and we throw it out. Ties count everyone who tied.
Step 2 — sort by skill tier. Each winning score goes in the bucket for the division that player was in that season, taken from their last regular meet of the season. So a score won in Div D five seasons ago stays a Div D data point even if that player is in A now.
Step 3 — recent seasons count double. Winning scores from the last 3 seasons are counted twice, the same 2x weighting League Night uses for recent form. Machines get worn, rebuilt and re-tuned; scores from 10 seasons ago should not carry equal weight. The sample size shown next to the number is the real, unweighted count.
Step 4 — the number is the middle. "What wins here" is the median (50th percentile) winning score in your bucket. It describes where you stand against what usually wins at your level, not a score anyone plays toward mid-game. The "middle half" range under it is the 25th to 75th percentile: hit the low end and you are in the conversation, hit the high end and you almost certainly won.

When your division is thin, we say so instead of quietly using someone else's numbers. In order:
1. Your division has 5+ winning scores on the machine, we use them, labelled "Div C".
2. It does not, so we add the neighbouring division with more data, one at a time, until we reach 5. The label changes to the range we actually used, like "Div B-C".
3. Even all four divisions together cannot reach 5, so we fall back to every score we have and label it "all divisions". A machine only ends up here if it has fewer than 5 recorded group wins in total, which means it is a machine almost nobody has played to a finish.

The count on the machine detail card is exactly how many winning scores the number was built from. If it looks small, it is small. League Night rows keep just the number and the label; the receipts live here and on the card.

We compare your median raw score on each machine vs your opponents' median raw scores. Raw scores tell you how you actually play the machine, which is more useful than league points (those depend on who else was in your group). League pts still show as secondary info. H2H records need 2+ shared encounters to show up. Confidence levels: High = 8+ plays, Med = 3-7, Low = 1-2.

Point God: Total era-adjusted points across all seasons (pre-S9 scaled by 1.75x).
Veteran: Most meets attended all-time.
Machine Whisperer: Highest avg league-point percentile across 5+ machines with 3+ plays each.
Record Books: Best single score relative to that machine's median (% above median). Point Machines: Machines where the league posts its biggest raw scores.
Never Miss a Meet: Most consecutive meets without missing one (resets each season).
Best Debut: Strongest first season, measured as points per meet played. Debuts from S5 on, and you need at least 5 meets played to make the list. Some seasons have nobody who qualifies, and then nobody wins it.
Most Improved: Biggest jump in points per meet played between two seasons in a row. Needs 3+ meets played in both. The S8-to-S9 pair is skipped, see below.
Giant Killer: Most wins against A-division players in designated Beat-the-A meets.
Untouchable: A-division player with the best survival rate on BTA nights.
Widowmaker: Machine with the biggest avg raw score gap between 1st and last per group. DQ groups excluded.
Equalizer: Machine with the tightest finishes, smallest avg raw score gap between 1st and last. DQ groups excluded.
Active filter: The "Active" tab shows players who played in at least one of the last 3 seasons.

Under the old scoring (4/2/1/0), a player could score 0 in a meet. We bump those to 1 after era normalization so they don't vanish from the math. The S9+ floor is 6 (all last-place finishes), so pre-S9 floors are a bit lower. Small bias, but a 0-point meet was genuinely a bad night.

Most Improved compares your points per meet played in one season against the next. Scoring changed from max 4 to max 7 points per game between S8 and S9. Everyone's numbers went up just from the rule change, so measuring S8-to-S9 improvement would give false "most improved" results. We skip that pair entirely. If someone genuinely got better during that transition, this stat can't see it, but there's no way to separate real improvement from scoring inflation.