Dan Oehm's Avatar

Dan Oehm

@danoehm.bsky.social

πŸ”” Statistician 🫢 #Rstats πŸ“ˆ #Dataviz πŸƒβ€β™‚οΈ Trail runner 🀘 Metalhead ❀️ Dad πŸ”₯ #survivor 🌏 Canberran 🌿 Vego Survivor Stats db: https://survivorstatsdb.com/ Ramblings: https://gradientdescending.com/

662 Followers  |  1,204 Following  |  662 Posts  |  Joined: 01.01.2024  |  2.0957

Latest posts by danoehm.bsky.social on Bluesky

Post image

The way I read this is, if Steven and Kristina are on the jury, Sage will get their votes.

Savannah, Rizo, and Sohpi shouldn't be trying to sit next to each other at the end, but rather get their vote at the FTC.

#Survivor49

05.12.2025 10:09 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Post image

As an aside I've been looking into if a lull in confessionals is attributed to a chances of NOT winning. So far, there's not much of a relationship there. Higher CPH is what matters the most. Higher CPH = more winning.

05.12.2025 09:56 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Post image

#Survivor confessionals per hour

Despite having a few quieter episodes, Savannah still has the largest difference (+1.5), followed by Steven (+0.7) and Sage (+0.7).

Steven picked up heaps due to going on the journey and winning the advantage.

05.12.2025 09:56 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0
Post image Post image Post image Post image

This is how I concentrate 🀘

At the start of the year I wouldn't have thought Cryptopsy would be my most listened to band, but there's something I love about that album.

Although, my number 1 song has to be Rivers of Nihil - House of light. So good.

#YouTube #heavymetal

05.12.2025 08:59 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Post image

This model includes much more data than just vote connections. What it means is, considering all data points, Savannah is in a more similar position to other winners of the past. It means all the other data points are pulling up the odds.

More on this in another post

04.12.2025 22:32 β€” πŸ‘ 3    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Post image

I also simulate all possible jury configurations (20 configurations at F6). π‘†π‘Žπ‘”π‘’ is expected 7 out of 20 scenarios, the most out of the F6.

One of her best scenarios is with Rizo and Savannah.

Savannah only has one winning scenario, which I know conflicts with the winner predictions!

04.12.2025 22:32 β€” πŸ‘ 2    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0
Post image Post image

#Survivor49 votes with the Jury

At this point, π‘†π‘Žπ‘”π‘’ is in the most favourable position with the Jury with the highest number of expected votes from the jury (white stars, according to the numbers).

If she gets to the FTC with Rizo and Savannah, I think π‘†π‘Žπ‘”π‘’ wins.

#Survivor

04.12.2025 22:32 β€” πŸ‘ 7    πŸ” 1    πŸ’¬ 1    πŸ“Œ 0

Both Rachel and Gabler had slightly fewer votes with the jury (-2) than Sam and Sue, and Cassidy, respectively. A more complex model picks up Rachel and Gabler but misses a couple of others. I find it interesting how well a simple model does.

30.11.2025 20:02 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Post image

Tree is up. Countdown begins πŸŽ„

30.11.2025 10:00 β€” πŸ‘ 3    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Post image

I also have a very simple model to predict how many jury votes the finalist will receive based on votes with and votes for the juror. It has been correct for 6 out of 8 New Era seasons.

I'll run this just before the S49 finale, simulating all jury configurations.

#Survivor

30.11.2025 09:04 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

Yeah it didn't seem like he spent much time gauging the reactions, but maybe he didn't have to! It will be interesting to see if he sowed some doubt into Kristina and Steven.

28.11.2025 03:24 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

I'm at the point where I can think of the most likely scenario but also know that's not happening this season!

28.11.2025 02:52 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Is it possible Rizo made a fake idol and intended to play it like it was his real one, act surprised when it's called out as fake to try to fool people into thinking he actually doesn't have a real idol? If so, he did a terrible job of it! #survivor

28.11.2025 02:27 β€” πŸ‘ 2    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0
Post image

#Survivor49 network

That trio definitely looks powerful! I'm looking forward to Sophi playing her KIP, taking Rizo's idol, playing for Sage, and sending Savannah or Rizo to the jury.

It's in the trio's best interest not to sit next to each other at the end, so let's see what happens.

27.11.2025 19:07 β€” πŸ‘ 3    πŸ” 1    πŸ’¬ 1    πŸ“Œ 0

Everyone loves an underdog story so maybe they're leading up to a Sage-Savannah clash and Sage comes out on top πŸ€”. Looking forward to ep11!

27.11.2025 11:33 β€” πŸ‘ 2    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Post image

#Survivor49 confessionals per hour

What's interesting is Sage still has the 2nd highest difference between observed and expected after Savannah, and that's more important the Sophi and Rizo having a slightly higher cph in terms of chances of winning

27.11.2025 11:33 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0
Preview
Do Jurors Vote for their Alliance at the Final Tribal Council? - Dan Oehm | Gradient Descending At the end of every season of Survivor the jury needs to make a choice, who gets the $1M, which makes jury management very important.

For those who want to dive further into how I measured this relationship you can read this post.

gradientdescending.com/do-jurors-vo...

27.11.2025 11:24 β€” πŸ‘ 2    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Post image

#Survivor49 votes with the jury

This is one of my favourite charts. A juror is 1.6x more likely to vote for the finalist they voted with the most - their main alliance. This shows who has voted with whom and who may get their vote. Another data point to infer the position of the remaining players

27.11.2025 11:24 β€” πŸ‘ 8    πŸ” 2    πŸ’¬ 1    πŸ“Œ 0
Post image

#Survivor49 Predictions

With Jawan gone, it's really put Sage in a less favourable position. The top 3 are now:
πŸ”΄ Savannah
πŸ”΅ Sophi and
πŸ”΄ Rizo

with Savannah increasing her chances to 1-in-3! If Savannah doesn't make it to the end, eyes on Sophi.

#Survivor

27.11.2025 11:14 β€” πŸ‘ 4    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0

My favourite part was how Andy remembered that he forgot to reply to my email!

I also defended his 6.25%!

25.11.2025 10:40 β€” πŸ‘ 3    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

Winning 3 consecutive IICs makes a player's chances even worse:
- 70% chance of surviving the next TC (vs 86%)
- 45% chance of surviving the 2nd (vs 71%)

Blog post coming soon with the gorey stats, model, and code.

6/

25.11.2025 10:37 β€” πŸ‘ 2    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

In summary, given where Savannah is in the game, she has a:
- 75% chance of surviving the next TC (vs 86%)
- 65% chance of surviving the 2nd (vs 71%)
- 44% chance of surviving the 3rd (vs 57%)

Definitely not quite as big as I thought, but definitely lower.

5/

25.11.2025 10:37 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

The analysis used all 48 seasons. Given the game can differ a lot e.g. when they merge, number of challenges, when they win 2 back-to-back challenges, I've taken all the different probabilities into account to estimate this effect.

4/

25.11.2025 10:37 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

Answer: Yes!

Intuitively, it makes sense. The effect isn't quite as big as I would have thought, but it is statistically significant.

3/

25.11.2025 10:37 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

Inspired by the challenge stats @joshkettles.bsky.social shared, I wanted to look into it a bit deeper, but reframing the problem slightly:

Are the chances of surviving subsequent Tribal Councils lower after winning back-to-back individual immunity challenges?

2/

bsky.app/profile/josh...

25.11.2025 10:37 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0
Post image

The chances of surviving subsequent Tribal Councils decrease after winning back-to-back IICs.

This means Savannah has a 75%, 65%, and 44% chance of surviving the next 1, 2, and 3 tribal councils unprotected - ~10% points lower than equal chance, a clear decrease.

1/

#Survivor

25.11.2025 10:37 β€” πŸ‘ 6    πŸ” 1    πŸ’¬ 1    πŸ“Œ 0
Creating beautiful tables in R with {gt}

While the gt() website has a fantastic documentation, just a reminder that @albertrapp.bsky.social ebook on gt() is the best collection of tools and trainings to help you master this awesome package

gt.albert-rapp.de

24.11.2025 00:52 β€” πŸ‘ 5    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0
Post image

Code for those inclined

22.11.2025 04:42 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Rizo may need to rely on winning an IIC after playing his idol and I don't like his chances. He hasn't performed particularly well.

In which case, he may as well hold onto it for as long as he can. If he gets voted out with it, well that's not too dissimilar from being voted out at the next tribal.

22.11.2025 04:42 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0
Post image

- 23 players have played an idol successfully for themselves after merge (changing the outcome of the tribal)
- 6 went on to win the game
- 4 of those players won at least one IIC after playing the idol
- 2 won all IICs after playing the idol (Mike S30, Rachel S47)

#Survivor

22.11.2025 04:42 β€” πŸ‘ 5    πŸ” 1    πŸ’¬ 1    πŸ“Œ 0

@danoehm is following 20 prominent accounts