Thursday, May 30, 2019 Game Recap and Discussion (SPOILERS)

This is where all of the games are discussed.

Moderators: alietr, trainman, econgator, dhkendall

User avatar
gnash
Jeopardy! Champion
Posts: 1678
Joined: Fri Jul 01, 2011 9:24 am

Re: Thursday, May 30, 2019 Game Recap and Discussion (SPOILERS)

Post by gnash »

John Boy wrote: Fri May 31, 2019 12:05 pm
Leander wrote: Thu May 30, 2019 7:56 pm
TenPoundHammer wrote: Thu May 30, 2019 7:29 pm

Tainted by association with Nazis -> "Deutschland über Alles" -> Germany. What? It's not Germany?! But... but Deutschland! That's Germany! HOW IS THAT WRONG?!? I still don't see how that ends up being Austria instead and I've read the Wikipedia article on it.
Two clues to me that it was not Germany:

In 1947, it would have been West Germany, which technically is not the same as Germany today, and I thought they would be asking about a current country.

Haydn was Austrian, and why even mention him in the clue unless you’re pointing to Austria.
Now THAT'S the way to read a clue, follow the breadcrumbs, rule out the screamingly obvious negbait, and come up with a sole get FJ.

Well done.
No, that's the way to get the facts wrong and stumble to the right answer for all the wrong reasons by sheer luck.

There was no West, East, or any Germany in 1947.

"West Germany" is the colloquial name that was used between 1949 and 1990 for Federal Republic of Germany, which is very much a country today, and the same country as it was then - just as the USA is the same country it was before it had all of its current 50 states.

Recognizing Haydn as a hint was the only part that was correctly reasoned.
User avatar
alietr
Site Admin
Posts: 8980
Joined: Thu Jun 30, 2011 1:20 pm
Location: Bethesda, MD

Re: Thursday, May 30, 2019 Game Recap and Discussion (SPOILERS)

Post by alietr »

opusthepenguin wrote: Fri May 31, 2019 12:32 pm
alietr wrote: Fri May 31, 2019 10:02 am
opusthepenguin wrote: Thu May 30, 2019 8:16 pm Fun fact: Shakespeare died on April 23, 1616. Miguel de Cervantes died 11 days earlier on April 22, 1616.
Speaking of temps perdu ...
Maybe so, but I'm not the one who perdued it. My fun fact is accurate.
Wanna read that one again?
User avatar
This Is Kirk!
Jeopardy! Champion
Posts: 6562
Joined: Mon Dec 17, 2012 1:35 am
Location: Seattle

Re: Thursday, May 30, 2019 Game Recap and Discussion (SPOILERS)

Post by This Is Kirk! »

alietr wrote: Fri May 31, 2019 12:45 pm
opusthepenguin wrote: Fri May 31, 2019 12:32 pm
alietr wrote: Fri May 31, 2019 10:02 am
opusthepenguin wrote: Thu May 30, 2019 8:16 pm Fun fact: Shakespeare died on April 23, 1616. Miguel de Cervantes died 11 days earlier on April 22, 1616.
Speaking of temps perdu ...
Maybe so, but I'm not the one who perdued it. My fun fact is accurate.
Wanna read that one again?
They didn't have the internet and video games back then so it felt like 11 days...
User avatar
opusthepenguin
The Best Darn Penguin on the Whole JBoard
Posts: 10319
Joined: Thu Aug 11, 2011 2:33 pm
Location: Shawnee, KS
Contact:

Re: Thursday, May 30, 2019 Game Recap and Discussion (SPOILERS)

Post by opusthepenguin »

If anyone wants to do the math, here are the scores when James uncovers DD3, says "Oh, this category seemed hard," and bets "only" $5,000

Image

And here are the clues left on the board:

Image

So... looks like $9600 left on the board? And if James had lost that $5,000, he'd be down to $26,412 or $7,612 more than Megan's doubled score. So if he'd missed, Megan had a chance to prevent a runaway but not a very good one. She'd have to pick up at least 8 of the remaining 12 clues--more if she doesn't get one of the high-value clues. And if she manages that, she's still got to get FJ right while James gets it wrong. The risk may have been unnecessary on James' part, but it was also very small.
User avatar
opusthepenguin
The Best Darn Penguin on the Whole JBoard
Posts: 10319
Joined: Thu Aug 11, 2011 2:33 pm
Location: Shawnee, KS
Contact:

Re: Thursday, May 30, 2019 Game Recap and Discussion (SPOILERS)

Post by opusthepenguin »

This Is Kirk! wrote: Fri May 31, 2019 12:48 pm
alietr wrote: Fri May 31, 2019 12:45 pm
opusthepenguin wrote: Fri May 31, 2019 12:32 pm
alietr wrote: Fri May 31, 2019 10:02 am
opusthepenguin wrote: Thu May 30, 2019 8:16 pm Fun fact: Shakespeare died on April 23, 1616. Miguel de Cervantes died 11 days earlier on April 22, 1616.
Speaking of temps perdu ...
Maybe so, but I'm not the one who perdued it. My fun fact is accurate.
Wanna read that one again?
They didn't have the internet and video games back then so it felt like 11 days...
I stand by my numbers. Anyone want to bet me $1000 that I'm wrong? (Oh please, oh please....)
User avatar
opusthepenguin
The Best Darn Penguin on the Whole JBoard
Posts: 10319
Joined: Thu Aug 11, 2011 2:33 pm
Location: Shawnee, KS
Contact:

Re: Thursday, May 30, 2019 Game Recap and Discussion (SPOILERS)

Post by opusthepenguin »

Ironhorse wrote: Thu May 30, 2019 10:08 pm I'm curious how many here came up with Elizabeth I's death year. I certainly did not.
Not me. I guessed 1606 but wouldn't have rung in with it. I knew she made it to the 17th century, but not very far in.
User avatar
This Is Kirk!
Jeopardy! Champion
Posts: 6562
Joined: Mon Dec 17, 2012 1:35 am
Location: Seattle

Re: Thursday, May 30, 2019 Game Recap and Discussion (SPOILERS)

Post by This Is Kirk! »

opusthepenguin wrote: Fri May 31, 2019 12:53 pm
This Is Kirk! wrote: Fri May 31, 2019 12:48 pm
alietr wrote: Fri May 31, 2019 12:45 pm
opusthepenguin wrote: Fri May 31, 2019 12:32 pm
alietr wrote: Fri May 31, 2019 10:02 am

Speaking of temps perdu ...
Maybe so, but I'm not the one who perdued it. My fun fact is accurate.
Wanna read that one again?
They didn't have the internet and video games back then so it felt like 11 days...
I stand by my numbers. Anyone want to bet me $1000 that I'm wrong? (Oh please, oh please....)
Calendar switch? You got us.
User avatar
opusthepenguin
The Best Darn Penguin on the Whole JBoard
Posts: 10319
Joined: Thu Aug 11, 2011 2:33 pm
Location: Shawnee, KS
Contact:

Re: Thursday, May 30, 2019 Game Recap and Discussion (SPOILERS)

Post by opusthepenguin »

This Is Kirk! wrote: Fri May 31, 2019 1:00 pm
opusthepenguin wrote: Fri May 31, 2019 12:53 pm
This Is Kirk! wrote: Fri May 31, 2019 12:48 pm
alietr wrote: Fri May 31, 2019 12:45 pm
opusthepenguin wrote: Fri May 31, 2019 12:32 pm
Maybe so, but I'm not the one who perdued it. My fun fact is accurate.
Wanna read that one again?
They didn't have the internet and video games back then so it felt like 11 days...
I stand by my numbers. Anyone want to bet me $1000 that I'm wrong? (Oh please, oh please....)
Calendar switch? You got us.
Ding ding ding! Cervantes was in Catholic Spain which had already switched to Gregory's calendar. England was still on the Julian calendar since the Gregorian version was pure papist meddling. So April 22 in Spain was April 12 in England and Shakespeare had 11 days left to live. This used to be more fun as a bar bet back when Cervantes' death date was given as April 23--same as Shakespeare's but 10 days earlier. Then some smartypants had to do some research and notice that Cervantes actually died a day earlier than we'd thought.
User avatar
gnash
Jeopardy! Champion
Posts: 1678
Joined: Fri Jul 01, 2011 9:24 am

Re: Thursday, May 30, 2019 Game Recap and Discussion (SPOILERS)

Post by gnash »

jeff6286 wrote: Fri May 31, 2019 12:12 pm To me it's not about guaranteeing the lock at that point of the game, more like ensuring the lock in all likelihood. With 12 or 16 or so clues still on the board, to lock the game mathematically you'd have to imagine either of the other two players getting all of the remaining clues, and particularly if you're James, that's just not going to happen. So it's not so much doing massive amounts of calculations in 5 seconds before betting, more just making some broad estimates based on what's on the board and the way the game has gone thus far. Sure he could bet $5 if he's really concerned about the category and get a 99.9% chance of winning, but by betting $5,000 he probably still has a 95-97% chance of winning, and essentially puts an extra $10k in his pocket at the end.
You are making good points, but let me take a step back and outline the fundamental premises of DD wagering theory. Based on historical data, the DD get rate is significantly higher than the FJ get rate for pretty much everybody. A good player should get more than 75% of DDs, and 90% get rate is totally realistic for a well-prepared contestant. By contrast, the average FJ get rate is about 50% (which I believe is the producers' target) and elite players like Jennings, Rutter and Madden got 70%. There have been exceptional DD get rates (Rogers 16/17, Ingram 13/14, Kelly 12/14 - I may be slightly wrong on some of those, but that's the range) but those were still small enough samples that they weren't statistically inconsistent with 70% long-run averages. James's 30/31 get rate is completely off the charts. Not only is it extremely improbable that his "true" average is anywhere close to 70%, but his FJ get rate is also even higher than his DD get rate (68/72). He actually trails David Madden in DD percentage.

So, the standard thinking goes that a lock is extremely valuable and high DD wagers are good because, if your choice is between deciding the game on that DD or on the FJ, the DD offers you a better chance. Also, a crush is significantly more valuable than a 2/3 game and being behind in 2/3 is worth risking if the upside is moving from 2/3 lead to crush or from crush to lock. This all follows very straightforwardly from typical DD/FJ get rates.

But much of that is in doubt with James's statistics. If we took his 97% FJ average as a "true" parameter, he has over 98% chance of winning if he merely leads going into FJ. Then again, reducing the win probability by, say, 0.5%, something a "normal" player might justifiably consider negligible, has a potentially large expected dollar value in James's case.

Bottom line: I admit I don't have a %$#@! idea what the optimal wagering strategy for James is. It probably isn't the same as for other people.
User avatar
jeff6286
Watches Jeopardy! Way Too Much
Posts: 5228
Joined: Fri Jul 01, 2011 7:34 pm
Location: Indianapolis, IN

Re: Thursday, May 30, 2019 Game Recap and Discussion (SPOILERS)

Post by jeff6286 »

gnash wrote: Fri May 31, 2019 1:08 pm
jeff6286 wrote: Fri May 31, 2019 12:12 pm To me it's not about guaranteeing the lock at that point of the game, more like ensuring the lock in all likelihood. With 12 or 16 or so clues still on the board, to lock the game mathematically you'd have to imagine either of the other two players getting all of the remaining clues, and particularly if you're James, that's just not going to happen. So it's not so much doing massive amounts of calculations in 5 seconds before betting, more just making some broad estimates based on what's on the board and the way the game has gone thus far. Sure he could bet $5 if he's really concerned about the category and get a 99.9% chance of winning, but by betting $5,000 he probably still has a 95-97% chance of winning, and essentially puts an extra $10k in his pocket at the end.
You are making good points, but let me take a step back and outline the fundamental premises of DD wagering theory. Based on historical data, the DD get rate is significantly higher than the FJ get rate for pretty much everybody. A good player should get more than 75% of DDs, and 90% get rate is totally realistic for a well-prepared contestant. By contrast, the average FJ get rate is about 50% (which I believe is the producers' target) and elite players like Jennings, Rutter and Madden got 70%. There have been exceptional DD get rates (Rogers 16/17, Ingram 13/14, Kelly 12/14 - I may be slightly wrong on some of those, but that's the range) but those were still small enough samples that they weren't statistically inconsistent with 70% long-run averages. James's 30/31 get rate is completely off the charts. Not only is it extremely improbable that his "true" average is anywhere close to 70%, but his FJ get rate is also even higher than his DD get rate (68/72). He actually trails David Madden in DD percentage.

So, the standard thinking goes that a lock is extremely valuable and high DD wagers are good because, if your choice is between deciding the game on that DD or on the FJ, the DD offers you a better chance. Also, a crush is significantly more valuable than a 2/3 game and being behind in 2/3 is worth risking if the upside is moving from 2/3 lead to crush or from crush to lock. This all follows very straightforwardly from typical DD/FJ get rates.

But much of that is in doubt with James's statistics. If we took his 97% FJ average as a "true" parameter, he has over 98% chance of winning if he merely leads going into FJ. Then again, reducing the win probability by, say, 0.5%, something a "normal" player might justifiably consider negligible, has a potentially large expected dollar value in James's case.

Bottom line: I admit I don't have a %$#@! idea what the optimal wagering strategy for James is. It probably isn't the same as for other people.
I think it's very hard to quantify both his DD and FJ get rates. Many of his DDs have seemed absurdly easy, but it's hard to attach any kind of figure to it, and it's highly possible that even if the DDs were harder he'd have no problem handling most of them. The FJs we do have a way to quantify, thanks to the jboard polls, but I don't know if anyone has done any kind of stat-compiling. It has also seemed like a historically easy stretch in FJ, and it feels impossilbe that he'd be able to go 30/31 over most stretches in the show's history. I'd love for someone to ask him if he's ever tracked FJ get rates over time and if he's shown 85% or 90%, I"d be stunned if he said 97% or anything in that range, but who knows, maybe his brain is just wired perfectly for FJ solving, which along with knowing all the things he knows is a deadly combo.
seaborgium
Undefeated in Reruns
Posts: 8941
Joined: Fri Jul 01, 2011 9:31 am

Re: Thursday, May 30, 2019 Game Recap and Discussion (SPOILERS)

Post by seaborgium »

John Boy wrote: Fri May 31, 2019 12:05 pm I for one will be just as happy to never again see that "dates" category come up. The easy ones were grade school level: In what month was MLK born? Duh, how about the month in which the entire US celebrates MLK's birthday. The three on the bottom rows were ridiculous, stuff no one would know and that can't be figured out from anything in the clue. Boo, hiss.
I was off by a day on Shakespeare and by a year on Elizabeth I, and had nothing but vague recollections to go on with them. But I knew Anne Frank and her family went a long time before being found, and that she died months before the end of the war.
User avatar
hbomb1947
Still hoping to get on Jeopardy! while my age is in double digits
Posts: 2434
Joined: Fri Jul 01, 2011 8:31 am

Re: Thursday, May 30, 2019 Game Recap and Discussion (SPOILERS)

Post by hbomb1947 »

Just to be pedantic, no one knows for certain that Shakespeare was born on April 23; his date of birth wasn't recorded. April 23 is generally accepted as his birthday because he was baptized on April 26; but the April 23 date is just an assumption.
Follow me on twitter, even though I rarely tweet! https://twitter.com/hbomb_worldwide
User avatar
gnash
Jeopardy! Champion
Posts: 1678
Joined: Fri Jul 01, 2011 9:24 am

Re: Thursday, May 30, 2019 Game Recap and Discussion (SPOILERS)

Post by gnash »

opusthepenguin wrote: Fri May 31, 2019 12:52 pm If anyone wants to do the math, here are the scores when James uncovers DD3, says "Oh, this category seemed hard," and bets "only" $5,000
(...)
So... looks like $9600 left on the board? And if James had lost that $5,000, he'd be down to $26,412 or $7,612 more than Megan's doubled score. So if he'd missed, Megan had a chance to prevent a runaway but not a very good one. She'd have to pick up at least 8 of the remaining 12 clues--more if she doesn't get one of the high-value clues. And if she manages that, she's still got to get FJ right while James gets it wrong. The risk may have been unnecessary on James' part, but it was also very small.
Thanks. I don't think it's helpful to use Megan's current doubled score in calculations, as what matters is her final doubled score - effectively, all she gets of the remaining clues needs to be doubled also. What matters is that James has 31,412, Megan has 9,400, and there's 9,600 left on the board.

If James wagers $5, he has either 31,407 or 31,417. If he gets nothing more (I know...), Megan needs to get 6,400 more to break the lock. That's very hard. Not only does she have to get 2/3 of the remaining values, but she absolutely has to get at least one of the bottom two Kowalski clues (otherwise her goal is mathematically impossible). But it is possible. Also, she has no possibility of getting to a 2/3 game, so the best theoretical case for her is to have to get the FJ while James misses. Now how much does James need to add to his score to ensure a lock? Assuming no rebound opportunities, James needs 2,400. Then he would have 33,807 (or 33,817) and Megan could only get up to 16,600 (= 9,400 + 9,600 - 2,400) if she sweeps everything James doesn't get. (Note that you can only get a multiple of 400 from the remaining clues. While it is true that James would have a lock with 2,200, that's not an available amount.) This means if James gets the 2000 clue and one other clue, he has a lock. (Also, if James gets the 2000 clue, Megan has to get literally every other clue. If he gets the 1600 clue, Megan can leave at most 800 to Rob or a TS.) What would you say the probability of a lock is in this case? I'd say it's well above 99%.

If James makes his usual wager, say $9,812, he has either 21,600 or 41,224. Obviously, it's a lock with a get. With a miss, what does he need to secure a lock? Again assuming no rebound opportunities, the equation is 21,600 + x > 2 * (9,400 + 9,600 - x), or x > 16,400/3. In terms of feasible values, he needs at least 5,600, or over half of the remaining values. That's actually more than he got in the game (5,200), though of course Megan didn't sweep all the remaining clues. In this situation, the probability of a lock is considerably diminished, though remember that it applies only if he misses. (Also note that Megan can get within 2/3 in this scenario. James needs to pick up 2,800 to prevent that.)

What wager ensures a lock with a get? Since Megan can at most get 19,000, he needs to have 38,001, for which he needs to wager 6,589. If he does that and misses, he has 24,823 and needs to pick up 4,400 to ensure a lock and 1,600 to ensure a crush.

With James's actual wager (5,000), he still needed 800 to ensure a lock after the get. That's not exactly 100% sure, but is pretty damn close. With a miss, he would've had 26,412 and would have needed 4,000 to ensure a lock and 1,200 to ensure a crush. While pretty close, I'd say this makes the wager clearly inferior to the (calculable in real time only if you're Watson) $6,589 one.

What is the maximum wager that doesn't risk a crush? One that leaves him with at least 28,501, i.e., $2,911. If he misses, he needs to pick up 3,200 (exactly 1/3 of the remaining values) to achieve a lock. If he gets the DD, he sits at 34,323 and needs 1,600 to ensure a lock (or, alternatively, if he sits pat, Megan needs 8,000 of the remaining 9,600 to break the lock).

Which of these is best depends on various assumed probabilities, but if you have the calculation speed of a computer, the optimal wager is almost certainly either 6,589 or 2,911. If you are human, I'd still probably go with "either make your usual big wager or the minimum wager" mainly because everything considered in this post is too much to do in your head in those few seconds you have to decide on the wager.
User avatar
gnash
Jeopardy! Champion
Posts: 1678
Joined: Fri Jul 01, 2011 9:24 am

Re: Thursday, May 30, 2019 Game Recap and Discussion (SPOILERS)

Post by gnash »

jeff6286 wrote: Fri May 31, 2019 1:36 pm I think it's very hard to quantify both his DD and FJ get rates. Many of his DDs have seemed absurdly easy, but it's hard to attach any kind of figure to it, and it's highly possible that even if the DDs were harder he'd have no problem handling most of them. The FJs we do have a way to quantify, thanks to the jboard polls, but I don't know if anyone has done any kind of stat-compiling. It has also seemed like a historically easy stretch in FJ, and it feels impossilbe that he'd be able to go 30/31 over most stretches in the show's history. I'd love for someone to ask him if he's ever tracked FJ get rates over time and if he's shown 85% or 90%, I"d be stunned if he said 97% or anything in that range, but who knows, maybe his brain is just wired perfectly for FJ solving, which along with knowing all the things he knows is a deadly combo.
For his FJs, I agree they have probably been easier than average. I got 26/31 and I'd expect more like 23 based on my long-run average of 75% (back when I kept track of it, but if anything I should be worse now that I'm rusty). Then again, I got Atticus Finch, Elgin and crowdfunding, all of which had 50% or less in board polls.

Back when they were posting their stats on the board, Austin Powers and Schliemann posted FJ runs comparable to James's, so I don't think it's implausible that his long-run average could be 90% or even higher. Also, LL questions are a good practice for FJ, so today's top players may be in better form than those of a decade ago and before.

As for DDs, I'm sure "many" did seem "absurdly" easy, but that's by no means unusual. And a sample of 72 is not so small. I would be very surprised if his DD average changes much. Besides, as I said before, Madden had a higher DD get rate. I think a bigger puzzle is Ken Jennings's 83%, which to me seems absurdly low given his other statistics.
User avatar
Robert K S
Jeopardy! Champion
Posts: 5249
Joined: Thu Jun 30, 2011 1:26 pm
Location: Cleveland, Ohio
Contact:

Re: Thursday, May 30, 2019 Game Recap and Discussion (SPOILERS)

Post by Robert K S »

BRD-98 wrote: Thu May 30, 2019 9:02 pm Haydn = Austrian. Just pavlovian.
And even if you don't know that, the clue is basically "pick a German-speaking country that isn't Germany". The string of incredibly easy Finals continues for James. I thought they got considerably harder during the Teachers Tournament. The easy Finals aren't really helping his streak, given all his locks, but they are helping his winnings total.
User avatar
gnash
Jeopardy! Champion
Posts: 1678
Joined: Fri Jul 01, 2011 9:24 am

Re: Thursday, May 30, 2019 Game Recap and Discussion (SPOILERS)

Post by gnash »

Robert K S wrote: Fri May 31, 2019 3:27 pm The string of incredibly easy Finals continues for James. I thought they got considerably harder during the Teachers Tournament.
For a balancing data point, I'm 84% on James FJs and was 90% on Teachers'.
Caboom
Watches Jeopardy! Way Too Much
Posts: 718
Joined: Mon Mar 24, 2014 9:49 am

Re: Thursday, May 30, 2019 Game Recap and Discussion (SPOILERS)

Post by Caboom »

gnash wrote: Fri May 31, 2019 3:42 pm
Robert K S wrote: Fri May 31, 2019 3:27 pm The string of incredibly easy Finals continues for James. I thought they got considerably harder during the Teachers Tournament.
For a balancing data point, I'm 84% on James FJs and was 90% on Teachers'.
Here's another data set for comparison, from a much weaker player: I'm 46,9 % during James's run (15/32), while I was 39.8 % (49/123) before him since the start of the season, disregarding the all star tournament. Last season I was 38 % (76/200) disregarding the tourneys. However, my stats for his run were inflated by a 10/11 streak early on during his run, I'm 5/20 for the last four weeks of his run. I was 3/10 for the Teachers tournament.

It would be interesting if someone went through the FJ get numbers for the board to compare them to earlier seasons.
Bamaman
Also Receiving Votes
Posts: 12897
Joined: Sat Jul 02, 2011 1:39 pm

Re: Thursday, May 30, 2019 Game Recap and Discussion (SPOILERS)

Post by Bamaman »

Ironhorse wrote: Thu May 30, 2019 10:08 pm I'm curious how many here came up with Elizabeth I's death year. I certainly did not.
I ran the dates category. I knew Elizabeth thanks to TPH. Sometime ago, there was a clue related to her death and he said he didn’t know how we should know that. It motivated me to look it up and I retained the information.
thenextofken
Watches Jeopardy! Way Too Much
Posts: 442
Joined: Fri Jul 08, 2011 2:54 pm

Re: Thursday, May 30, 2019 Game Recap and Discussion (SPOILERS)

Post by thenextofken »

Ironhorse wrote: Thu May 30, 2019 11:30 am I confused Haydn with Chopin and went with Poland. In retrospect, "1947" should have eliminated that because one would think Poland would not have waited two years to revert back from a Nazi-associated anthem.
I went with Poland as well, simply because I knew the Nazis had invaded that country.
User avatar
DBear
Denier of Pop Culture
Posts: 2548
Joined: Sat Sep 03, 2011 1:57 pm

Re: Thursday, May 30, 2019 Game Recap and Discussion (SPOILERS)

Post by DBear »

I had 1601 for Liz I's death date. I think I was partially conflating her with Vicki who died in 1901. : :|
Post Reply