Clyde Tournament eligibility: Direct, Casual and League DC games do not count toward the Clyde Tournament. See the complete tournament rules.
WEEK THREE · THE SEMIFINAL WRAPUP
Deception Takes the Final.
Two Players. One Crown.
GodHandGU makes it three pool wins in three weeks. Q310495 takes the other title seat with 117 counted wins. October's championship belongs to one of them.
The October final is set: GodHandGU against Q310495. One has finished first in every pool. The other just led the semifinals in counted games, wins and winning percentage. After three weeks and a starting field of 28, these are the two left chasing the title.
GodHandGU finished ahead of Baklava in Semifinal 1; Q310495 overturned the seed order against Gerold in Semifinal 2. Baklava and Gerold will play for third. All four remain in action, but the championship door has closed behind the two winners.
The top seed holds.
The third seed breaks through.
All four players cleared the 35-game minimum. These semifinals were decided by each player's qualifying win percentage over the week; the records below are their full weekly results.
SEMIFINAL 1
PLACE
Three rounds. Three first-place finishes.
GodHandGU has found a remarkably dependable route through October: win the pool, move on, repeat. First in Charlie, first in Pool A, and now first in Semifinal 1 with 47–20 and 70.15%. The top seed is in the final.
Baklava actually collected two more wins, finishing 49–29. Nine additional losses made the difference. Across 78 counted games, Baklava's 62.82% left a 7.33-point gap to the championship seat.
There is still a podium place to chase. Baklava has already travelled from third in the opening Delta pool to the last four; the final assignment is Gerold, with third place on the line.
SEMIFINAL 2
PLACE
Q takes the bigger workload all the way.
Q310495 arrived as the third seed and left with the best percentage of the semifinal round: 117–47, 71.34%. Those 164 counted games were also the most of the four contenders. Volume and efficiency travelled together.
Gerold finished 90–45, 66.67% over 135 games. The second seed won two games for every loss, but Q's final margin was 4.67 percentage points.
A week ago, Gerold's six closing wins squeezed him into the semifinals. This time there was no late escape into the title pool. Q advances; Gerold joins Baklava in the third-place contest.
Q kept finding
another counted win.
Going into Sunday, Q310495 stood at 97–38 (71.85%); Gerold was 85–40 (68.00%). Gerold had just put together a strong 16–3 Saturday. There was still a final day to play, but Q had the lead.
Q added 20 wins and nine losses on Sunday. Gerold's day ended 5–5. Both weekly percentages came down, but the gap grew from 3.85 to 4.67 points. Q finished the job without needing a last-minute reversal.
The effort behind that record was larger still: Q's audit contains 315 captured games, of which 164 counted. The opponent cap alone excluded 148. Even a marathon has to pass through the tournament's counting rules.
Clyde's calculator has requested a quieter hobby.
Deception gets both seats.
Different builds brought it there.
GodHandGU's familiar Guild Deception construction led his usage again, appearing in 70 captured games. Q310495's most-used build was Combined Arms Deception, at 130 captured appearances. Both finalists leaned most heavily on Deception during the semifinal week, but they arrive with different signatures.
Gerold's most-used domain was also Deception, across 84 captured games, while his most-used individual construction was Big Mana Nature, at 16. A domain total combines multiple builds; one Nature deck can lead the individual list without Nature leading the whole week. Baklava's profile was led by Death, with Everborn Death the most-used individual build at 11 appearances.
GodHandGU
47–20 · 70.15%
Guild Deception ↗
70 captured uses · Week 3
Baklava
49–29 · 62.82%
Everborn Death ↗
11 captured uses · Week 3
Q310495
117–47 · 71.34%
Combined Arms Deception ↗
130 captured uses · Week 3
Gerold
90–45 · 66.67%
Big Mana Nature ↗
16 captured uses · Week 3
Deck and domain usage includes captured games that may not count toward the tournament. The linked weekly game audits determine the qualifying records; a deck's usage count is not its tournament win total.
GodHandGU vs. Q310495.
One week for the crown.
GodHandGU brings three consecutive pool wins. Q brings the strongest semifinal percentage and more than twice GodHandGU's counted Week 3 workload. Their latest percentages were only 1.19 points apart. That makes an inviting final, even though the new week starts each record afresh.
CHAMPIONSHIP
GodHandGU vs. Q310495
Three straight pool wins meet back-to-back wins in Pool B and Semifinal 2. One more successful week makes an October champion.
THIRD PLACE
Baklava vs. Gerold
Gerold's semifinal percentage was 3.85 points higher, but those results stay in Week 3. Baklava gets a fresh week to turn a last-four run into a podium finish.
A little more room for the final opponents.
In October Week 4 only, eligible games against the other player in your own locked pool have no two-games-per-opponent limit. Games against everyone else still stop counting after the first two. That gives meetings with your final opponent extra weight whenever they occur in eligible play.
Players must still be Rank 11 or 12 when they play and complete 35 qualifying games for the week. Direct, Casual and League DC games remain excluded. The final week closes at 00:00 UTC on November 2, after Sunday's play.
Sources and scoring notes
Final records come from the official Week 3 archive, checked October 26, 2026. The competition ran October 19 at 00:00 UTC through October 26 at 00:00 UTC. The archive was finalized October 26 at 01:39 UTC. Final-week pairings were checked against the locked Championship and Third Place draw.
Saturday and Sunday figures were calculated from counted rows in Q310495's audit and Gerold's audit. Percentage gaps use the published two-decimal figures. The 444 counted player-games are player appearances, not necessarily 444 distinct matches.
Earlier records come from the preserved Week 2 report and Week 1 report. Each player card links all three weekly audits and the archived most-used deck.
The complete Week 2 final-results dispatch and its surviving Week 1 references follow. Return to the semifinal edition ↑
WEEK TWO · THE FINAL FOUR EDITION
Four Left Standing.
The Margins Get Smaller.
Six closing wins. A 26–2 record without a semifinal seat. One survivor from each original pool. October's second chapter saved its sharpest turn for Sunday.
Twelve players arrived with a Week 1 success story. Four leave with a place in the semifinals. Between those two numbers came a Sunday surge, two unfinished activity requirements, several changes at the deck desk, and one of the narrowest cuts Clyde has had to explain.
GodHandGU, Gerold, Q310495 and Baklava are October's Final Four. Each comes from a different original pool. None has won the tournament yet. What they have earned is another week in which every counted result can matter, as the last few hours of this one made abundantly clear.
Twelve names.
Four seats at the next table.
These are the final published standings from the October 12–18 Gods Unchained tournament week. Two eligible players advance from each pool. Names open the Week 2 game audits; deck links open each player's most-used event-week construction.
POOL A
TOP 2 ADVANCE
Two repeat contenders. One painfully small gap.
GodHandGU took first with 53–16 and 76.81%, following a Week 1 victory in Charlie with another pool win. Gerold claimed the other seat at 148–47, 75.90%. Two very different workloads, two places in the semifinals.
[tst]apollo finished 43–14, 75.44%. That would make a fine headline in plenty of tournaments; here it meant third, just 0.46 percentage points behind Gerold. Apollo improved from 68.12% in Week 1 and still ran out of room.
???4377822 was fourth at 28–10, 73.68%, up from 60.87%. That 12.81-point improvement was the largest among players who met the minimum in both weeks. Charlie's two Light profiles gave this pool serious depth, even though neither advanced.
Scyta completed 66 counted games and finished 36–30, 54.55%. DudsTheDude, last week's highest seed, recorded 18–11 but ended on 29 of the required 35 games. The first chapter's standout percentage could not carry into the second.
POOL B
TOP 2 ADVANCE
Q sets the pace. Baklava takes the opening.
Q310495 won Pool B with 140–57 and 71.07% across 197 counted games, the largest workload in the twelve-player field. The Week 1 Bravo runner-up now has a pool win and the third semifinal seed.
Baklava moved from third in Week 1 Delta to second here: 57–27, 67.86%. Filo won one more game, finishing 58–32, but Baklava's lower loss count produced the better percentage. The final qualifying gap was 3.42 points.
Rxbson put in 135 counted games, third most in the field, and finished 82–53, 60.74%. Cardano reached the minimum with 36 games and a 21–15, 58.33% record. That percentage exactly matched Cardano's Week 1 finish.
Then there is Cuuurrrrke: 26 wins, two losses, 92.86%. The highest percentage on either scorecard belonged to a player with only 28 counted games. The published order places that record below the eligible players. The reason belongs in the headline, too.
The archive places eligible players first, followed by players below the minimum. Player records use counted games. The linked construction reflects all captured games in the event-week usage window, including excluded games; its usage count and the player's qualifying record measure different things.
Six wins.
Just enough daylight.
At 19:46:01 UTC on Sunday, a counted loss left Gerold at 142–47, 75.13%. Ahead lay the narrowest route into the Final Four. Gerold's remaining counted results were six wins.
The first lifted the record to 143–47. The second made it 144–47, 75.39%, still below [tst]apollo's final 75.44%. The third, recorded at 20:19:58 UTC, made it 145–47, 75.52%. Three more wins brought Gerold home at 148–47, 75.90%.
That completed a 17–3 Sunday. Last week's closest advancement cut was 1.47 percentage points; this one was less than a third as wide. Apollo's 43–14 deserved respect before the run and deserves it afterward. Gerold found the additional results needed to move ahead.
Clyde would like the record to show that leaving the press box for an early dinner would have been a terrible editorial decision.
- WIN 1 · 19:56:59 UTC143–4775.26%
- WIN 2 · 20:04:54 UTC144–4775.39%
- WIN 3 · 20:19:58 UTC145–4775.52%ABOVE APOLLO'S FINAL %
- WIN 4 · 20:37:45 UTC146–4775.65%
- WIN 5 · 20:51:10 UTC147–4775.77%
- WIN 6 · 21:03:09 UTC148–4775.90%
The closing sequence is reconstructed from timestamped counted results in Gerold's final game audit. Percentages are cumulative Week 2 records at each checkpoint.
A remarkable percentage.
An unfinished requirement.
26–2. Still seven games short.
Cuuurrrrke produced the most striking raw record of Week 2: 92.86%. Every one of the 28 captured games counted. There were no hidden rank exclusions or repeat-opponent deductions explaining the shortfall.
The percentage column could hardly have looked better. The games column still needed seven more. The same 35-game minimum applied to every entrant, so the record could not earn advancement.
There is no reason in the published results explaining why the run ended there. What we can say is that the wins were impressive and the requirement remained incomplete. Both belong in a fair account of the week.
Last week's top seed exits.
DudsTheDude opened the tournament with 29–6, 82.86% and the first Week 2 seed. This time, the audit contains 31 captured games. Two were excluded under the repeat-opponent limit, leaving 29 counted games and an 18–11 record.
That left Duds six short of eligibility. Meeting the minimum would have restored the chance to compete for a place; it would not by itself have guaranteed one. Pool A's qualifying bar was already high.
All twelve entrants had captured and counted activity this week. Ten reached the minimum. Nobody here had an empty game log, and the two incomplete runs deserve an accurate explanation.
Counted games require player Rank 11 or 12 when the game is played. Only the first two eligible games against the same opponent GUID count during the week. The cumulative minimum increases by five each UTC day, reaching 35 on Sunday. Direct, Casual and League DC games do not count toward the Clyde Tournament. Read the tournament scoring rules ↗
Every original pool
still has somebody to follow.
One from Alfa. One from Bravo. One from Charlie. One from Delta. After two rounds and twenty-four eliminations from the original 28-player field, all four opening pools have exactly one representative left. Their routes here have been anything but identical.
GodHandGU
Charlie winner, Pool A winner, now the No. 1 semifinal seed. The record moved from 52–16 in Week 1 to 53–16 in Week 2, lifting the percentage from 76.47% to 76.81%. That is an unusually consistent pair of strong weeks.
The most-used construction remained the same Guild Deception list. The published name still has question marks. The case for taking this contender seriously has considerably fewer.
Gerold
The 148 wins led the entire Week 2 field. Gerold raised the qualifying percentage from 68.31% to 75.90% while increasing the workload from 183 to 195 counted games. Sunday's closing run supplied the final push.
Combined Arms Deception was the most-used construction, although the counted audit includes play across all six domains. The second seed arrives with both volume and variety on the record.
Q310495
197 counted games, a Pool B victory and a second consecutive week of triple-digit wins. Q310495 followed 124–38 with 140–57. The percentage came down from 76.54% to 71.07%; the record still earned first place in the pool.
The most-used domain changed from Magic to Deception, with a Combined Arms Deception construction leading Week 2 usage. Like Gerold, Q recorded counted games with all six domains. Those two will share a semifinal pool.
Baklava
Baklava entered Week 2 as the tenth seed after taking Delta's last qualifying seat. A 57–27 finish lifted the percentage from 64.47% to 67.86% and secured the fourth semifinal seed. Delta's third-place finisher is now its last remaining representative.
Death remains the most-used domain, but the featured construction changed from Zombie Death to Everborn Death. Baklava brings the Final Four's only most-used Death profile into the meeting with the top seed.
Deception takes three seats.
Death keeps the fourth.
The Final Four's most-used event-week domains are three Deception and one Death. That looks very different from the twelve players who advanced last week, when Magic supplied four most-used profiles. This week's survivors also show why the deck link is worth opening.
The returning leader stayed with a familiar construction. GodHandGU again favored Guild Deception. Gerold and Q310495 instead led their own usage with separate Combined Arms Deception lists. The shared domain badge covers different thirty-card constructions, and both of those players used all six domains in counted games.
Several of last week's deck stories changed direction. Baklava's most-used Death construction moved to Everborn. Filo's most-used list is now Big Mana Magic; Rxbson's most-used domain changed from Magic to Nature, with Wild Nature leading construction usage. Cardano retained a Combined Arms War profile, while Apollo's Irina Light and ???4377822's BOTC Light both produced strong qualifying percentages without advancement.
Clyde's scouting note: a favorite deck helps introduce a player. The complete audit tells you how much more they brought to the week.
Most-used domain and construction are frozen captured-usage summaries. These twelve profiles are not a study of the whole Gods Unchained metagame, and the player's qualifying win percentage is not the win rate of the single linked deck.
A closer look
inside the final four constructions.
One scouting pick from each semifinalist's exact most-used Week 2 construction. These are cards whose verified effects make their inclusion interesting; the standings do not tell us which cards were drawn or played in a particular win.
Frumentarii Instigator ↗
Two copies appear in the top seed's Guild Deception construction. Its Roar draws cards according to the other Guild creatures on the board. That gives an established Guild presence a way to refill the hand. Clyde likes a guest who brings more guests, provided somebody has counted the chairs.
Rear Admiral Thibbs ↗
Gerold's most-used list carries one copy. Blitz and a summoned 3/2 Digby give the six-mana legendary immediate tools, while cards leaving your void refresh mana. It is a flexible package worth examining beside the list's heavier cards. The admiral has arrived with an assistant and an interest in the mana budget.
Frogman Scout ↗
Q310495's featured construction contains two copies. Temporary hidden helps set up an attack; damaging the opposing god grants a stat boost, shuffles the Scout into the deck and draws two cards. That makes the connection attractive for both pressure and resources. A small creature with a fairly busy itinerary.
The Broken Seal ↗
Baklava's Everborn construction includes one copy. An all-Everborn creature lineup unlocks its opening-hand and zero-cost setup. The relic creates a Sporophyte at the start of your turn and returns to hand through its Afterlife. Those effects offer recurring material to work with. Death has brought something that is difficult to persuade to leave.
OCTOBER 19–25 UTC · FOUR BECOME TWO
The draw is set.
October still needs a champion.
SEMIFINAL 1
The winner of Charlie and Pool A meets the tenth Week 2 seed who has played into the Final Four. Consistency at the top, a rising percentage from below, and two very different featured constructions.
SEMIFINAL 2
The two largest Week 2 workloads share one pool: 195 counted games for Gerold and 197 for Q310495. Both have consecutive triple-digit win totals, and only one advances from this semifinal.
The pairings come from the locked official Week 3 field. The four advancers were seeded by final Week 2 qualifying win percentage, with wins and then lowest GUID as the seeding tiebreakers. That produces No. 1 against No. 4 and No. 2 against No. 3.
Each semifinal is a two-player statistical pool. Players build qualifying records through their normal captured games; they do not have to schedule Direct matches against the person opposite them here. The Rank 11/12 rule, first-two-per-opponent cap and 35-game final minimum continue. One eligible player advances from each pool.
Week 3 runs from October 19 at 00:00 UTC to October 26 at 00:00 UTC. The final week, including finals and third-place standings, is scheduled for October 26 through November 1 UTC. The numbers beside these pairings are Week 2 records; the new week's competition starts fresh.
Four players remain. Every original pool still has someone to cheer for. Clyde will keep the press box open; after that Sunday finish, leaving early would feel irresponsible.
Week 2 sources and reporting scope
Published October 18, 2026 in Eastern time after the round closed at October 19, 00:00 UTC. The official Week 2 archive is final and immutable, finalized at October 19, 00:00:39 UTC. The separately captured official Week 3 field supplies the locked semifinal seeds and assignments. Each dispatch keeps its own saved production evidence.
The twelve final game audits reconcile to 1,024 counted player-games from 1,320 captured player-game records. A match involving two entrants can contribute two player records. Week 1 comparisons use the original reviewed report below. Most-used domain and construction profiles cover captured event-week usage, and all four pictured cards were verified in the exact linked constructions. Card discussion is editorial analysis of their effects, without replay or interview claims.
The October and November prize-support announcement was supplied directly by the tournament organizer on October 18, 2026.
Source URLs, capture times, raw-response hashes, audit summaries and the closing sequence are retained with this dispatch. Follow the monthly qualifiers ↗ or read the preserved Week 1 report below ↓.
The original Week 1 report follows, including its correction and the Week 2 preview as published. Return to the Week 2 results ↑
WEEK ONE · THE FIRST CHAPTER
TWELVE MOVE ON.
OCTOBER HEATS UP.
A brilliant 29–6, two pool races separated by hundredths, and a reminder that every qualifying game matters. The October tournament has given Clyde plenty to write about.
Editor's correction: Marcomax was not a no-show. His PROD audit contains 31 games, all excluded for rank. Sleec is the only entrant with no captured Week 1 games. This revised issue uses the published PROD results and official Week 2 assignments throughout. View Marcomax's audit ↗
Twenty-eight players began October's opening stage, and twelve earned the next chapter. Across four seven-player pools, we got a standout percentage from DudsTheDude, serious mileage from Gerold and Q310495, and two races for first that barely fitted between the decimal places.
We also saw strong records miss the cut, promising runs end before the activity minimum, and one unused seat. There is plenty to celebrate and a little to learn. So settle into the press box: the coffee is on, the PROD results are in, and Clyde has checked which numbers mean what this time.
Alfa. Bravo. Charlie. Delta.
Four stories worth following.
Here is the complete published Week 1 field, with a report from each pool. These are statistical standings from qualifying GU games, not a series of scheduled head-to-head fixtures. Player names open their PROD game audits.
ALFA
TOP 3 ADVANCE
A marathon, a sprint, and a photo finish
Gerold led at 125–58, 68.31%, while Cuuurrrrke finished 28–13, 68.29%. Gerold played 183 qualifying games to Cuuurrrrke's 41; the published percentages still ended just 0.02 points apart. Both advance. Clyde has ordered the calculator a magnifying glass.
Cardano secured third at 58.33%, ahead of Moiches at 56.86%. Deccal bebek was close behind Moiches at 56.82%. The two leaders favored different Combined Arms Deception constructions; Cardano carried Combined Arms War into the next round.
BRAVO
TOP 3 ADVANCE
The toughest ticket in town
DudsTheDude arrived at exactly the 35-game requirement and left with 29 wins, six losses and 82.86%—the highest qualifying percentage in the field. No spare games, very little wasted motion. Q310495 supplied the long-distance version: 124–38 over 162 games, 76.54%.
Filo claimed the last seat at 73.53%. That left ???4826021 outside the cut despite a very respectable 46–20, 69.70%. Every Bravo player met the minimum. This was not an attendance contest; it was a particularly demanding pool.
CHARLIE
TOP 3 ADVANCE
Deception leads; Light takes two seats
GodHandGU won Charlie with 52–16, 76.47%, favoring Guild Deception. [tst]apollo followed at 68.12% with Irina Light, and ???4377822 took third at 60.87% with BOTC Light.
Logvinov finished 28–20, 58.33%: precisely Cardano's advancing record in Alfa, but fourth here. Same numbers, different neighbors. That is the uncomfortable arithmetic of separate pools—not a lack of effort from Logvinov.
DELTA
TOP 3 ADVANCE
Three hundredths and a sweet escape
Rxbson edged Scyta at 66.13% to 66.10%. Both advance after another finish that required the second decimal place. Their favored constructions were Magic Other and Combined Arms Magic, respectively.
Baklava brought Zombie Death through in third at 49–27, 64.47%. The name says dessert; the deck says the guests may return from the grave. ???4584768 was next at 61.19%, while Dogman put in 124 games and 72 wins but finished fifth. Credit to Dogman for the commitment.
A tiny margin.
A very real ticket.
The smallest gap between a third-place qualifier and the next eligible player came in Alfa: Cardano at 58.33% and Moiches at 56.86%. Those 1.47 percentage points were the difference between playing Week 2 and watching it.
Elsewhere, the final qualifying gaps were 3.83 points in Bravo, 2.54 in Charlie and 3.28 in Delta. The 0.02-point Alfa and 0.03-point Delta races were battles for first, not last-seat eliminations. Both players in each of those photo finishes advanced.
That distinction matters. Clyde enjoys a dramatic headline, but he would rather not send anybody home who actually qualified.
Thirty-five was the requirement,
not a stretch goal.
Twenty-two of the 28 entrants reached the weekly minimum. Six did not, and that is disappointing—we wanted every seat in the draw to become a full week of competition. Real life comes first, but the standings still need one consistent activity rule.
roky13 had 19–10, 65.52% but only 29 counted games. That was a genuinely promising position left unfinished. Skipman stopped at 33, just two short of eligibility, although meeting the minimum alone would not have guaranteed advancement. elbiosis went 6–2, and oversight went 3–2; neither sample was large enough. Plenty of intriguing starts, not enough completed assignments.
CAPTURED
sleec
PROD's Week 1 audit records no captured games for sleec. Clyde kept the seat warm, but unfortunately that is not a scoring category. We would much rather be writing about a comeback next time.
Marcomax deserves a correction, not a no-show joke. Marcomax played 31 captured games. All 31 were excluded because the player entered below Rank 11. His zero on the tournament board means zero qualifying games, not zero effort. The earlier draft got that wrong.
Counted games require player Rank 11 or 12 at game entry, with only the first two eligible games against the same opponent GUID counted during the week. The final cumulative minimum is 35; it rises by five each day. The published minimum and the two zero-count audits are linked above.
Twelve tickets.
Twelve decks to investigate.
The top three eligible players in every pool advance. Here are their final Week 1 records and most-used event-week constructions, displayed in the official Week 2 seed order.
The record belongs to the player, not exclusively to the linked deck. The most-used construction is taken from all captured games in the event-week usage window, which can include non-counted games. The original production record is preserved; current reviewed player names display without changing the linked GUID.
Four cards that caught
the correspondent's eye.
These picks come from the exact constructions linked in PROD's final standings. They are my scouting choices for cards with useful roles—not a claim that deck inclusion proves a card decided a match. Hover the artwork for a closer look, or open a card or its source deck.
Switch Duelist ↗
Two copies appear in Gerold's favored list and in the Charlie winner's. A one-mana Guild creature with temporary hidden and changing strength/health gives Deception an early presence without committing much mana. Small card, useful first step; Clyde approves of showing up early.
Land of the Dead ↗
DudsTheDude's Anubian construction runs two. Returning up to six one-mana Anubians from the void, with soulless, is a substantial way to rebuild a board. In a deck full of cheap Anubians and death-trigger payoffs, the appeal is easy to see. The guest list apparently includes the previous guest list.
Demogorgon ↗
Two copies sit in Gerold's favored construction. Putting other creatures to sleep, damaging sleeping enemies and carrying Leech makes this a clear candidate for stabilizing a longer game. Clyde is calling it a potential momentum-turner, not claiming to have watched a particular winning turn.
Epiphany ↗
Rxbson's favored Magic construction contains two. Six damage plus a replacement card is a straightforward package in a spell-heavy list. Some cards need a lengthy introduction. This one brings six damage and keeps the conversation moving.
Card lists and effects were checked against the PROD deck catalogue for the archived construction IDs. The available evidence shows inclusion, not whether a card was drawn, played or responsible for a specific win.
Bring your best deck. Finish your qualifying games. Leave the overly enthusiastic commentary to Clyde.
Sources, scope and publication notes
Source capture began October 12, 2026 at 00:42 UTC (October 11, 8:42 p.m. Eastern). Week 1's published final standings supply every result, eligibility flag, domain and most-used construction. The PROD current tournament page supplied the locked Week 2 assignments; percentages beside them are Week 1 records, not live Week 2 performance.
Sleec's audit and Marcomax's audit distinguish no captured activity from excluded games. The source URLs and response hashes are retained with the issue. Its 1,603 counted player-games are participant records, not necessarily 1,603 distinct matches. No replay-level card actions or interviews were available.