Tuesday, December 30, 2014 Game Recap & Discussion [SPOILERS]

[jjwaymee]

I'm a little surprised that the Vikings haven't asked this already, but...

Vaughn, why didn't you wager to cover a double-up by Marcia? Did you have some intuition that she would wager only to lock out third place and not make the "suicide wager"? You kinda defied conventional wisdom last night and fortunately you were right.

I'm not Vaughn, but his wager was fine. The Archive had Vaughn's DD wager as $2000 instead of $200. I've fixed it now.

Aha! That makes much more sense. Thanks Mark. Obviously, I did not see last night's show and only caught up with Archive this morning. I was out playing Thrift Store Trivia for fabulous (thrift store) prizes at my favorite microbrewery.

[John Boy]

In ISRAEL $800, the response was given "miracle of the loaves and fishes". Would "feeding of the 5,000" work? (I know feeding of the 5,000 and feeding of the 4,000 were different things, and I'm afraid that I picked the wrong one.)

Pretty sure the reference was Matthew 14, so 5000 is the correct one and I would say yes, you were correct. The feeding of the 4000 is in Matthew 15.

ACCEPTABLE: What is the miracle of the feeding of the 5,000?

ACCEPTABLE: What is the miracle of the loaves and fishes?

NOT ACCEPTABLE: What is the miracle where the guy loafs and fishes?

[John Boy]

Congratulations to Vaughn on another great win. This was a fantastic game, very enjoyable to watch! I got 36 clues correct including garter snake, Lucid (DD), and FJ.

Let me second the motion re: congrats to Vaughn, with extra kudos for qualifying for the Jeopardy! Players Hall of Fame with $50K+

Indeed, a well played game except for Dan's horrible FJ wager (which, fortunately for him, was irrelevant).

[TenPoundHammer]

So far, no real complaints that this was a tough board. So I'm guessing that just all 12 categories were in my hammerhouse?

And I will never, for the life of me, be able to keep any of those kings straight. Nor will I ever understand how anyone can.

[bpmod]

And I will never, for the life of me, be able to keep any of those kings straight. Nor will I ever understand how anyone can.

I can understand not wanting to learn things. I have no desire to learn about most of what passes for entertainment these days. But to not understand how anyone else would, especially things that are relatively important (relative to current entertainment, for instance) in the grand scheme of things, is truly sad. Would that you would have at least as much interest in world history as in, say, the retail world.

Brian

[clprez]

I got lucky with FJ because earlier this year they found the remains of Richard III, and mentioned he was killed in battle. I would have had no idea else wise.

[econgator]

So far, no real complaints that this was a tough board. So I'm guessing that just all 12 categories were in my hammerhouse?

Not that tough at all (20 J, 24 DJ), so apparently they all were.

[deling]

Congratulations to Vaughn on another great win. This was a fantastic game, very enjoyable to watch! I got 36 clues correct including garter snake, Lucid (DD), and FJ.

Let me second the motion re: congrats to Vaughn, with extra kudos for qualifying for the Jeopardy! Players Hall of Fame with $50K+

Indeed, a well played game except for Dan's horrible FJ wager (which, fortunately for him, was irrelevant).

Harsh, man. I think "horrible" is a little strong, but you deserve an explanation. Given what I had already seen of Vaughn and Marcia, I figured 1) both could be reasonably expected to be strong in such a category (commonly studied, used on show, etc), 2) Marcia would bet to cover me OR bet all, or leave $1, 3) Vaughn would bet to cover Marcia's bet to cover me but never come close to a Clavin, and 4) it was a category I felt very strong in.

I'm a proponent of a more Bayesian FJ wagering strategy as opposed to the purely arithmetical methods. Given my high confidence that both would give the correct response, and at least approximately how they would bet, the only routes I could see to advance from 3rd were if both missed (unlikely), if one missed and bet traditionally so I would finish second if I got it (slightly more likely), or hope one missed and bet big so even if I missed and was left with $2 I'd be in 2nd.

The trouble with attempting this type of wagering approach is that you don't have much time to work out your various degrees of confidence to calculate your bet. I would not at all be surprised if I took the time to actually do a true Bayesian calculation, it would indicate a different bet. And in truth, given the standings at the end of DJ and my very high certainty one or both would give the right response and bet accordingly, I did not see any real chance of even moving out of 3rd. If Vaughn got it, he would win. If Marcia got it, she would at least cover me. Realistically for me to win, both would have to miss while I got it, and in a category like "Royalty" that simply was not going to happen.

[bpmod]

Realistically for me to win, both would have to miss while I got it (so I wouldn't have to get it either), so wagering for the triple stumper would have given me the best chance to win.

There. Fixed that for you.

Brian

[deling]

Realistically for me to win, both would have to miss while I got it (so I wouldn't have to get it either), so wagering for the triple stumper would have given me the best chance to win.

There. Fixed that for you.

Brian

Per the arithmetical models for wagering, you're absolutely right. But I believe a more complex wagering approach can be more successful, i.e. integrating more than just the dollar amounts into the decision making process. This is actually one of the few areas where new contestants can have an advantage over returning champions. Having watched Vaughn for 2 matches and Marcia for our match together, and knowing the FJ category, I had various degrees of confidence in my ability to provide the correct response, their abilities to do the same, and at least some estimation of how they would wager (given prevailing strategies, relative scores, and how they had wagered in DD and FJ opportunities to that point). I believe that disregarding all that I had learned before making my final decision would have been foolish.

Having said that, I've never seen a detailed, informed analysis of using this kind of wagering vice the suggested wagers you'll see at the Archive. I'd be interested to hear what others think.

[bpmod]

Realistically for me to win, both would have to miss while I got it (so I wouldn't have to get it either), so wagering for the triple stumper would have given me the best chance to win.

There. Fixed that for you.

Brian

Per the arithmetical models for wagering, you're absolutely right. But I believe a more complex wagering approach can be more successful, i.e. integrating more than just the dollar amounts into the decision making process. This is actually one of the few areas where new contestants can have an advantage over returning champions. Having watched Vaughn for 2 matches and Marcia for our match together, and knowing the FJ category, I had various degrees of confidence in my ability to provide the correct response, their abilities to do the same, and at least some estimation of how they would wager (given prevailing strategies, relative scores, and how they had wagered in DD and FJ opportunities to that point). I believe that disregarding all that I had learned before making my final decision would have been foolish.

Having said that, I've never seen a detailed, informed analysis of using this kind of wagering vice the suggested wagers you'll see at the Archive. I'd be interested to hear what others think.

Not to belabour the point, but you stated that for you to win, your opponents needed to get it wrong and you to get it right. That is simply mathematically not true. True, you needed both to miss, but once they do, you are already sitting at a higher score than they would fall to, especially considering your confidence in how they would wager (which turned out to be well-founded). If one or the other gets the correct response, you might be able to play for second vs. third, but there is no way you can win. But what you did with your wager is guaranteed you would lose if you did not get the correct response, regardless as to whether the others did or what their wagers might be. Before that wager, you had been in the fortunate position of not needing a correct response to win.

Brian

[MarkBarrett]

off topic: Brian, I put a clue for you in today's Sports J! thread.

[deling]

Realistically for me to win, both would have to miss while I got it (so I wouldn't have to get it either), so wagering for the triple stumper would have given me the best chance to win.

There. Fixed that for you.

Brian

Per the arithmetical models for wagering, you're absolutely right. But I believe a more complex wagering approach can be more successful, i.e. integrating more than just the dollar amounts into the decision making process. This is actually one of the few areas where new contestants can have an advantage over returning champions. Having watched Vaughn for 2 matches and Marcia for our match together, and knowing the FJ category, I had various degrees of confidence in my ability to provide the correct response, their abilities to do the same, and at least some estimation of how they would wager (given prevailing strategies, relative scores, and how they had wagered in DD and FJ opportunities to that point). I believe that disregarding all that I had learned before making my final decision would have been foolish.

Having said that, I've never seen a detailed, informed analysis of using this kind of wagering vice the suggested wagers you'll see at the Archive. I'd be interested to hear what others think.

Not to belabour the point, but you stated that for you to win, your opponents needed to get it wrong and you to get it right. That is simply mathematically not true. True, you needed both to miss, but once they do, you are already sitting at a higher score than they would fall to, especially considering your confidence in how they would wager (which turned out to be well-founded). If one or the other gets the correct response, you might be able to play for second vs. third, but there is no way you can win. But what you did with your wager is guaranteed you would lose if you did not get the correct response, regardless as to whether the others did or what their wagers might be. Before that wager, you had been in the fortunate position of not needing a correct response to win.

Brian

Excellent point, Brian! In the scenario I described for both other contestants missing on FJ, and me making it, I would have been well-served if I had bet $0. My explanation above was poorly written. I didn't put much stock in that scenario at the time - I couldn't see them BOTH missing in such a category. But given what I had seen of Vaughn and Marcia, I was dead certain she would bet to cover me, so if she got it I would lose no matter what. I was also relatively sure that Vaughn would bet to cover Marcia. So If either of them get it right, which I thought was a near certainty, I'm done. The only remaining scenarios are the ones where they both miss. And since I felt very confident I would get it (given that I had studied what felt like a lot in that topic), I bet for the admittedly unlikely scenario of myself to win as the only correct respondent (and make almost as much as possible in the process), with a small margin held in case Marcia went for the jugular and missed.

I would love if someone more mathematically rigorous would put together a probabilistic wagering calculator, based on your confidence in various factors (opponent R/W, wagers, etc.).

[Winchell Factor]

It's really limiting to play FJ from behind. In the last 3 games, I have played from behind a lot (and been very nervous about it), but I've been lucky to be ahead going into FJ twice out of 3 times. (Because, as dhkendall tells us, Jeopardy is two parts luck and one part luck!)

[Bamaman]

I remembered the FJ from a few years ago where we had to give Henry VII for killing Richard III. I got it then, but I recall it was a bit of a guess. This time it was an instaget.

I heard him say rhythm gymnastics and knew they were going to reverse it.

[bpmod]

I would love if someone more mathematically rigorous would put together a probabilistic wagering calculator, based on your confidence in various factors (opponent R/W, wagers, etc.).

You're not likely to be able to put odds on each of the 8 possible outcomes of FJ! in the amount of time you have to make your wager (and I don't mean you; I mean anybody playing the game). But if you came up with the odds of WWR being much higher than WWW (or any of the other 6 (in your game)), then you made the right wager. Since neither of those was the outcome, we'll never know.

Brian

[Caboom]

So far, no real complaints that this was a tough board. So I'm guessing that just all 12 categories were in my hammerhouse?

And I will never, for the life of me, be able to keep any of those kings straight. Nor will I ever understand how anyone can.

You're not alone. This was my worst result since I started tracking my scores in mid-October (If we disregard the one TOC episode where I got 3 right ).
I did get FJ though.

[Caboom]

I would love if someone more mathematically rigorous would put together a probabilistic wagering calculator, based on your confidence in various factors (opponent R/W, wagers, etc.).

You're not likely to be able to put odds on each of the 8 possible outcomes of FJ! in the amount of time you have to make your wager (and I don't mean you; I mean anybody playing the game). But if you came up with the odds of WWR being much higher than WWW (or any of the other 6 (in your game)), then you made the right wager. Since neither of those was the outcome, we'll never know.

Brian

Here's a quick calculation (with some rough estimations and assumptions that might be way off) for how often you should get the sole right to make the all in a +ev play in that situation:
If you get it right, you gain 11800. If you get it wrong, you lose 10800 (or 9800) and whatever you would have won in the next games. (The assumption here is that the other two miss and bet as you assumed, therefore giving you the win with either a sole get or a small enough bet. Like was stated above, the other players missing is the only situation where you can win, if they bet rationally).
Let's assume your win probability in the next episodes, as the defending champion, is around 40%, and your average win would be around 20k. In the next episode your average take would be 0.4x20k=8k. In the one after that, 0.4x0.4x20k=3.2k. Then 0.4x0.4x0.4x20k=1280. And finally 512 in the next one. That results in around 13k combined, to which we add the 2nd or 3rd place money that you are guaranteed to get at some point.
So you're risking 10800+13000+1500 to win 11800. Therefore you need to get it right 25300/(25300+11800)=0.68=68% of the time.

68% doesn't sound too high, if it's a category in your wheelhouse. It does, however, seem high when you factor in that two high level opponents have already missed the question. In addition, there's also the fact that you're risking the title of Jeopardy champion, which some people would consider to be worth quite a lot. (Note: you never gain the title of jeopardy champion with your all in bet, if, as you assumed, the other two bet rationally)

[Bamaman]

I said feeding the multitudes on the miracle clue. I assume they'd have taken that, though I would have been ready with more info if given a BMS.

[Volante]

I think that the Beatles deserved more Grammys than U2, oh well . I remembered Shannon Lucid from the J! App game.

'more grammys' really depends on longevity. The Beatles were active for, what, a decade? U2 has been around for three, going on four. It's basically the Woody Allen principle: you're talented at what you do, you keep doing what you do, you'll rack up accolades by being persistent even if not all your body of work has been critically acclaimed.

Plus there's probably been more awards U2 has been able to get simply due to award bloat.