you forgot to factor in the cost of years of therapy after you turn a sure win into a loss and alex trebek makes fun of you in his slimy way

interesting analysis though

Dude blogged about this decision:

Pretty funny.

There are other factors you haven't considered.

Mastery of the buzzer is of paramount importance, and the more games you play, the better you master this ability. Each and every game you are stronger simply from this.

The Tournament of Champions, it invites players based on wins and has a grand prize of $250k with lesser prizes down the line. Anything you do to hurt your chances of winning any game (irrelevant after 5 wins) also affects your chances of reaching the tournament and another payday.

Now, if some hack has a lock and the perfect category, then risking everything MAY be a good move. But it's pretty rare that a poor player would actually have a lock going into FJ.

Winning games is just too important.

i) Mastery of the buzzer can't be that big of a deal? Some guy who is 20% to win his first show isn't going to be 30% to win his next one.

ii) Didn't consider the Tournament of Champions. How many wins do you typically need to get in? It's worth an average of ~$32,000 per player but with very convex payouts. The guy who is {1/5,2/5,2/5} and playing his first game is i) really unlikely to make the Tournament of Champions, ii) even more unlikely to do better than the semi finals. In addition his best chance to reach the tournament might be to play super aggro and hope to qualify through three big wins. i.e, it's possible that going all in on FJ in a lock game increases his chances of making it.

I will concede that your TOC point makes it much less likely that a strong player who has a shot at final tabling it should gamble a lock game before he's locked up a TOC spot. But for the {1/3,1/3,1/3} guy? Even if he locks this game up, if he has to win another 3 games that's a 1/27 chance of getting an EV of say $13,500. So $500. And if he gambles on a topic he's 70% on he's giving up $150 of that. So it's essentially an irrelevant factor for him.

Now, the average player gets 49% of FJ questions right, and this guy needs to get it right 61.2% of the time to gamble it all. How often will this happen? Let's assume Mr. Average's expected probability based on the Final Jeopardy Topic is ~N(49,23). This says that 1/40 shows there is a topic that he's 90% to get correct, and 1/40 shows there is a topic he's 10% to get. If this is true then he should wager all in on ~27% of topics. Adjusting for non linear utility and so forth will lower this percent but it's still going to be well above 0. And given that AFAIK pretty much noone ever wagers a lock game, some of the decisions not to do so have been egregious mistakes.

BTW this discussion was merited by me tweeting to friends before today's FJ that I would wager all in even if I had a lock game. The topic was 'International Sports'. There's just extremely few questions about non-American sports that the Jeopardy producers would think reasonable that I wouldn't know. The q was

'This country with a population of ~ 4.5 million has hosted the Winter Olympics twice, and has more medals in it's history that any other nation'.



BTW is there a method to choosing the FJ questions? Cos some of them are ridiculously easy.

Recently:

Q. This 5-letter name appears 7 times in Shakespeare titles, more than any other name

Someone got that wrong, but c'mon. That's ridiculous.

Another one:

Q. It's once again in demand repairing old stone infrastructure, & is 1 of the 5 most popular U.S. boys' names today

Interesting, I snap answered Austria, thinking for whatever reason that Oslo had hosted the Summer Olympics, not Winter.

I couldn't believe the champ answered Russia, second time she clearly didn't read the entire FJ clue.

Um, yeah. A contestant winning their first game will have at least a 10% better chance to win their second. It is the single most important element of the game. Don't forget that each and every contestant has passed the 50 question test twice in order to be on the show, so all the contestants know many of the questions.

Let's put it another way, in a normal game I'll know 40-50 of the answers, if I buzz in first each and every time that's $30-40k without factoring in DD's. I'd have a lock every game.

Just how do you think Ken Jennings won 74 games? He was very good but it's not like he knew twice as much information as the other contestants every game. No, he simply mastered the buzzer and dominated.

Each and every Jeopardy contestant and aspiring contestant knows the buzzer is of paramount importance. Without question.

The cutoff is usually in the middle of 4 wins, tie-breaking is done by money won. Like the buzzer, it really helps to totally understand each and every aspect of the game especially when making statements like risking locks.

Going back to buzzer theory, a guy that is 1/3 to win the first game is at least 2/5 to win the second, etc. So you've got to adjust for that in your figures.

I watch Jeopardy each and every day having done multiple auditions and passing the online test each and every year I was eligible. FWIW, I got all three above questions within two seconds.

But for you, it should be a good real world experience of why you don't risk locks. Sometimes wheelhouse categories can ask nasty questions, sometimes horrible categories can ask questions in such a way that everybody will get it.

FJ clues are just like any other clue, sometimes they're easy, sometimes not. But almost every clue can be rationed out even if you don't immediately know the answer. And by the way, if you asked the Shakespeare question to the typical poker table, I'd be surprised if 20% could answer correctly. It's a yawner to many Jeopardy players, impossible to the general public.

Would've liked to have seen more aggressive guessing from Zachary in the last few DJ! answers to prevent the runaway by Karen (who, is Jeopardy! 8 fyi).

What happens if 2 players have 0 (or negative) balance going into FJ? Does the remaining player get a freeroll or does he end with what he has at the end of DJ? What if all 3 are in the red?

Yesterday:

Topic: 10 letter words.

A. This was Geppetto's job.

Answer was 'woodcarver' - would they have credited my 'woodworker'?

If 2 players can't play, then yeah, remaining player gets to freeroll. So long as they don't foolishly bet it all, they get to go back the next day.

If all 3 are unable to play (has only happened once), there's no FJ! round that day and 3 new players show up the next episode.

Tuesday 1/28

Champ (woman obviously) 18,600. Leader 20,000. Third 9,000.

Leader bet 17,200. Champ bet 18,400 and third went all in. First's bet is obvious, and second is equally obvious (but atrocious) and third's bet I thought was irrelevant, but I guess he should count on second's moronity and bet no more than 6200 or so.

There was something off about woman champ. Not very attractive to me as a whole, but a lot of attractive parts for a Jeopardy champ. Glad to see Asian back up his DD hunting and go all-in about 7500, smacking his big balls on the forehead of the pretender who got the first two DDs and wagered a total of 1600.

The bets *should* have been:

Leader - 17,201
2nd - 0-599
3rd - 0-6,200

Right?

Precisely.

The dude tonight decided he wanted to gift the girl he was playing against another game?

The current champ is on twitter and is kinda funny. @arthur_affect

Says he did not have sex w/ challenger in exchange for tie wager.

yeah, wtf. and betting 5 dollars, ok you dont know it but atleast guess a sport.

Pretty sure that's exactly what happened. I'm cool with it overall, but as a 3-day champion, you'd think he'd want two opponents that are fresh with the buzzer.

ETA: I really like this champ's wagering, and his win today was especially impressive given that he blew all his SJ! earnings on a missed DD towards the start of the DJ! round. He takes smart risks and plays to win... I like that.

Wednesday 1/29

A few curiosities in this episode.

1) The way the champ jumps around the board is annoying, but bearable assuming he has a strategy of finding the daily double...and then actually doing something constructive with it. I can't remember if it was in the 1st round or if it was in DJ, but it doesn't matter. He finds the DD and wagers $5!! What the **** is the point in looking for the DD if you plan on wasting it? If you aren't confident at all in the category, why are you choosing that category, especially when you are hunting the DD??? Doesn't it make sense to look for it, and hope to find it, in a category you are more familiar with? Also, his answer to the DD question was "I don't know", and it came within a microsecond of Trebek finishing the clue. Can't he at least guess any ****ing sport? Jeez.

2) On to FJ. Champ 18,200. 2nd 13,400. 3rd 8,200.

Sound strategy suggests that 2nd shouldn't bother to go all in, because she can't win if 1st gets it right, and also she has to watch out for 3rd place, who can catch her. Therefore I was expecting 2nd to wager at least 3,000. However, she did go all in, and was rewarded when Champ bet exactly 8,600 and they both got FJ and the game ended in a tie! (Third idiotically went all in; clearly he shouldn't bet more than 2,200 because this is enough to win in the (unlikely) event 1st and 2nd whiff FJ)

So seeing that first sometimes (even if rarely) allows the tie, is the all in strategy of 2nd now more acceptable? Many times we castigate 2nd for a dumb all in which loses but would have won if she had bet smaller, but what about the times she could have won by tying, but didn't bet enough? BO?

1 of 2 capital cities that end in Z, one in europe? and one in the americas.

I would think so.

Yeah, that was weird. I thinking jumping around the board is actually a decent strategy if you can handle it, because you're prepared for the category a couple seconds before your opponents and can throw off their rhythm.

And there's something to say for just killing the DD to reduce variance if you think you're the best player. But you should still hunt among categories you know before those you don't. Why go Sports for $1000 if you don't know sports? You'd have to have a really strong suspicion the DD was in that slot, and given the board, there was no reason to believe that.

It depends. I'd need to see someone show a predilection for betting a tie amount before I'd adjust my strategy based on it.

When you find a DD in a good category you should wager big to make it count. When you find a DD in a poor category you wager next to nothing and it's a non-issue. Pretty basic.

Regardless of category you want to find all the DD's, that way your opponents can't utilize them. Ask yourself this, the category is Mongolian Opera and you know there is a DD in the middle column. Now given first pick of the board, would you rather take it out with no harm done or give your opponents a chance to find it and possibly double their score? Easy decision.

Now onto the FJ wagering, as a rule I'm not fond of offering the tie unless you are significantly better than the person you are offering the tie to.

But 2nd place blindly wagering it all is definitely not acceptable in most circumstances. But there is one thing nobody has mentioned in today's discussion and that is the champ also offered a tie in yesterday's game as well. This is important. A challenger watching the previous game knowing he offered the tie prior might guess he will do the same and go for the double up and the tie. Makes sense.

This now brings up an interesting game dynamic, let's assume a multi-day champ has offered the tie each and every time when the game was still in doubt. If 2nd place is aware of this, he/she will (correctly) bet everything and win/tie if correct. But the possible hidden genius in the could be in what happens if everybody misses FJ. If 2nd misses they have zero and can't win whereas with the typically proper small wager gives them a win on a triple stumper. So the only way the champ can lose (assuming 3rd is out of the picture) is to miss FJ while 2nd answers correctly.

It's certainly interesting to think about the long term dynamic of this.

You'd think the guy probably didn't think she'd wager it all when offering the tie. Perhaps he strategically offered the tie thinking she was a weaker opponent that he would like to go up against again. Not sure that would be a wise move as she was consistently beating him on the buzzer.

Arthur with a poker reference on jboard ... "There's much less 'theory' on DDs out there than there is on the FJ wagering, but the strategy I decided on with DDs was what poker players call 'tight and aggressive' -- bet big on DDs that I probably do know, but as small as possible on DDs that I probably don't know."