Survivor ELO Blog
SHALLOW & BEAST Recap, Survivor 50 Episode 7
Published: 2026-04-10
Dee's -107.3 is the largest single-episode loss this season
TL;DR: Dee Valladares plays her Shot in the Dark, fails, and is voted out 8-4-1 in a split vote targeting her and Tiffany. Her -107.3 is the largest single-episode SHALLOW loss of any player this season. Tiffany votes for Coach (the wrong target) and drops 32.3 points, extending her season-long streak to 0W-15L. Emily and Christian both reach 11W-0L in US50 votes. Ozzy wins individual immunity.
Episode Summary: SurvivorElo Episode 7
Quick refresher: SHALLOW rates strategic positioning (voting correctly + outlasting). BEAST rates challenge performance. Both use Elo, so gains and losses depend on relative ratings. Full methodology at survivorelo.com.
Episode 7: What Happened
Fourteen players entered the first post-merge tribal council. Ozzy won individual immunity. The majority executed a split vote: eight votes on Dee, four on Tiffany, with one stray vote on Coach. Dee played her Shot in the Dark. It came up "NOT SAFE." She was eliminated 8-4-1.
Dee's Collapse
Dee entered Episode 7 at 1608, A-tier, with a 4W-0L vote record across six episodes. She left at 1501, C-tier. The -107.3 is the largest single-episode SHALLOW loss of any player this season, surpassing Jenna's -59.1 from Episode 1.
| Metric | Value |
|---|---|
| Pre-Episode 7 | 1608 (A) |
| Post-Episode 7 | 1501 (C) |
| Episode Change | -107.3 |
| Vote Losses | 0W-12L |
| Outlast Losses | 0W-13L |
| Season Total | -85.1 |
The damage has two components. First, Dee played her Shot in the Dark instead of voting. The model treats this as a non-vote. She loses a matchup against every player who voted correctly; twelve players did, so twelve vote losses. Second, she was eliminated. That produces thirteen outlast losses, one against each surviving player. The combined 25 losing matchups dropped her 107.3 points.
The model does not distinguish between a player who was blindsided and a player who saw it coming and took a desperation shot. It sees 25 losses and no wins. Four correct votes across six pre-merge episodes are now buried under a single catastrophic tribal.
Dee's season trajectory is one of the sharpest reversals the model has recorded this season. She entered at 1586, climbed to 1608 through four consecutive correct votes, and fell below her entering rating in a single episode. Her final 1501 is the lowest exit rating of any A-tier player this season.
Tiffany's Freefall
Tiffany voted for Coach. No one else did. The model registered twelve vote losses, one against each player who voted for Dee or Tiffany (the split vote targets). The outlast bonus from Dee's elimination (+5.1) partially offset the damage, but she still dropped 32.3 points.
| Player | Pre-Ep7 | Post-Ep7 | Ep7 Change | Season Change | Vote Record |
|---|---|---|---|---|---|
| Tiffany | 1507 (C) | 1474 (C) | -32.3 | -38.1 | 0W-15L |
Tiffany has now attended two tribal councils in US50. She has voted incorrectly at both. Her cumulative record is 0W-15L. No other active player this season has zero vote wins. The model sees a player who has never been on the right side of any vote. Two tribals is a small sample, but the signal is unambiguous.
The Majority Gains
Twelve players voted for either Dee or Tiffany. In the model's split-vote logic, both targets count as correct votes. Every member of the majority earned 2W-0L: one win against Dee (who did not vote) and one against Tiffany (who voted incorrectly).
| Player | Pre-Ep7 | Post-Ep7 | Ep7 Change | Season Change | Season Votes |
|---|---|---|---|---|---|
| Ozzy | 1502 (C) | 1516 (C) | +14.4 | +54.7 | 7W-3L |
| Aubry | 1544 (B) | 1557 (B) | +13.1 | +21.5 | 3W-0L |
| Chrissy | 1542 (B) | 1555 (B) | +13.1 | +25.6 | 4W-0L |
| Rick | 1551 (B) | 1563 (B) | +12.8 | +29.4 | 5W-4L |
| Rizo | 1574 (B) | 1586 (B) | +12.1 | +28.6 | 5W-0L |
| Stephenie | 1581 (A) | 1592 (A) | +11.8 | +29.7 | 8W-5L |
| Joe | 1588 (B) | 1599 (A) | +11.6 | +35.9 | 5W-0L |
| Jonathan | 1593 (A) | 1604 (A) | +11.4 | +9.7 | 4W-4L |
| Emily | 1607 (A) | 1618 (A) | +10.9 | +68.7 | 11W-0L |
| Christian | 1617 (A) | 1627 (A) | +10.6 | +66.4 | 11W-0L |
| Coach | 1624 (A) | 1635 (A) | +10.3 | +16.9 | 3W-0L |
| Cirie | 1710 (S) | 1718 (S) | +7.6 | +34.5 | 8W-0L |
The pattern is clean. Lower-rated players gained more per win (Ozzy's +14.4 at the top) and higher-rated players gained less (Cirie's +7.6 at the bottom). This is the Elo tax: wins against lower-rated opponents are worth less when you are already at the top.
Ozzy's +14.4 is the largest single-episode gain this week. His season total of +54.7 is now the third-highest among active players, behind only Emily (+68.7) and Christian (+66.4). He entered the season at 1462, the lowest of any returning player, and has climbed 54 points in seven episodes.
Season-to-Date: SHALLOW Rankings
Active players:
| Player | Vote Record | Entering | Current | Season Change | Tier |
|---|---|---|---|---|---|
| Cirie | 8W-0L | 1683 | 1718 | +34.5 | S |
| Coach | 3W-0L | 1618 | 1635 | +16.9 | A |
| Christian | 11W-0L | 1561 | 1627 | +66.4 | A |
| Emily | 11W-0L | 1549 | 1618 | +68.7 | A |
| Jonathan | 4W-4L | 1595 | 1604 | +9.7 | A |
| Joe | 5W-0L | 1564 | 1599 | +35.9 | A |
| Stephenie | 8W-5L | 1563 | 1592 | +29.7 | A |
| Rizo | 5W-0L | 1558 | 1586 | +28.6 | B |
| Rick | 5W-4L | 1534 | 1563 | +29.4 | B |
| Aubry | 3W-0L | 1535 | 1557 | +21.5 | B |
| Chrissy | 4W-0L | 1529 | 1555 | +25.6 | B |
| Ozzy | 7W-3L | 1462 | 1516 | +54.7 | C |
| Tiffany | 0W-15L | 1512 | 1474 | -38.1 | C |
Eliminated players:
| Player | Vote Record | Entering | Final | Season Change |
|---|---|---|---|---|
| Dee | 4W-12L | 1586 | 1501 | -85.1 |
| Savannah | 1W-6L | 1588 | 1544 | -44.0 |
| Q | 0W-5L | 1547 | 1503 | -43.8 |
| Charlie | 0W-4L | 1567 | 1530 | -37.3 |
| Genevieve | 0W-4L | 1553 | 1522 | -30.7 |
| Colby | 0W-4L | 1554 | 1529 | -25.1 |
| Mike | 2W-3L | 1580 | 1556 | -23.9 |
| Angelina | 2W-7L | 1528 | 1505 | -23.0 |
| Kamilla | 3W-3L | 1543 | 1531 | -11.2 |
| Jenna | 0W-7L | 1582 | 1523 | -59.1 |
| Kyle | -- | 1625 | 1625 | +0.0 |
Quick Hits
Emily and Christian are pulling away. Both are now 11W-0L in US50 votes, the only players with a perfect record across seven episodes. Emily leads the season at +68.7; Christian is at +66.4. Emily's lower entering rating (1549 vs. 1561) continues to amplify each win slightly more. Eleven consecutive correct votes is not a streak; it is the defining feature of these two players' games.
Joe crosses into A-tier. His +11.6 this episode pushed him from 1588 (B) to 1599 (A). He is 5W-0L in US50 with no vote losses. The model sees a player who has been on the right side of every vote he has attended.
Tiffany is now the lowest-rated active player by a wide margin. Her 1474 is 42 points below the next-lowest active player (Ozzy at 1516). She is the only active player with a negative season change (-38.1) and the only one with zero vote wins. If she does not vote correctly next episode, the gap will widen further.
BEAST Update
No new challenge Elo data is available for Episode 7. The BEAST standings carry forward from Episode 6.
Compare: Emily vs. Ozzy
Two players with opposite profiles and converging trajectories.
| SHALLOW | Tier | Season Change | US50 Votes | |
|---|---|---|---|---|
| Emily | 1618 | A | +68.7 | 11W-0L |
| Ozzy | 1516 | C | +54.7 | 7W-3L |
Emily and Ozzy have the two largest season gains among active players. The similarity ends there. Emily's gain comes from surgical precision: eleven correct votes, zero wrong. Ozzy's comes from volume and recovery: three early vote losses in the pre-merge, followed by a run of four consecutive correct votes and a +14.4 gain this episode. Emily entered higher (1549 vs. 1462) and has stayed clean. Ozzy entered at the bottom and has climbed 54 points through a messier path. The model rates Emily 102 points higher. The gap is real, but Ozzy is the player who has moved the most from where he started.
Explore the full rankings, player histories, and season snapshots at survivorelo.com.
Thanks to Dan Oehm (@danoehm.bsky.social) for his survivoR R package, which provides the underlying data.