So...what have we learned so far from James Holzhauer?
Moderators: alietr, trainman, econgator, dhkendall
- Robert K S
- Jeopardy! Champion
- Posts: 5313
- Joined: Thu Jun 30, 2011 1:26 pm
- Location: Cleveland, Ohio
- Contact:
Re: So...what have we learned so far from James Holzhauer?
In practice I'm not sure it's even 250 milliseconds. That may be a case of rounding up. I recall John Lauderdale telling me it was some much lower number, like single-digit or double-digit milliseconds, and therefore, each contestant should pulse the button as rapidly as possible (just as Watson did). The lockout really only needs to be long enough to let the other contestant ringing in at the same time get in, for which about 10-50 ms should be plenty.
- MarkBarrett
- Watches Jeopardy! Way Too Much
- Posts: 16556
- Joined: Fri Jul 01, 2011 10:37 am
- Location: San Francisco
Re: So...what have we learned so far from James Holzhauer?
Fair enough, thanks.Robert K S wrote: ↑Wed Apr 17, 2019 4:34 pm In practice I'm not sure it's even 250 milliseconds. That may be a case of rounding up. I recall John Lauderdale telling me it was some much lower number, like single-digit or double-digit milliseconds, and therefore, each contestant should pulse the button as rapidly as possible (just as Watson did). The lockout really only needs to be long enough to let the other contestant ringing in at the same time get in, for which about 10-50 ms should be plenty.
Sporcle has tests. Here's one: https://www.sporcle.com/games/pr1nterbo ... -the-k-key
-
- Watches Jeopardy! Way Too Much
- Posts: 632
- Joined: Wed Jan 04, 2012 6:50 pm
Re: So...what have we learned so far from James Holzhauer?
Heh, I THOUGHT 3 seconds was too much; I did know there was SOME lockout.
- This Is Kirk!
- Jeopardy! Champion
- Posts: 6578
- Joined: Mon Dec 17, 2012 1:35 am
- Location: Seattle
-
- Watches Jeopardy! Way Too Much
- Posts: 2981
- Joined: Fri Jul 15, 2011 7:11 am
Re: So...what have we learned so far from James Holzhauer?
As we've already seen many times before this, a deep knowledge base is worth rather little if you never get called on to answer. I know handling the signaling thingy is part of the game, but I hate to see it become basically the whole game.
- AndyTheQuizzer
- Lots and Lots of Interviews
- Posts: 2606
- Joined: Fri Jul 01, 2011 11:01 am
- Location: St. John's, Newfoundland and Labrador, Canada
- Contact:
Re: So...what have we learned so far from James Holzhauer?
It's always been the whole game. It just becomes more prominent when one player has significantly better timing than his opposition.
- gnash
- Jeopardy! Champion
- Posts: 1678
- Joined: Fri Jul 01, 2011 9:24 am
Re: So...what have we learned so far from James Holzhauer?
James is one of the most dominant players on the buzzer, and it's key to his success, but there (in fact, here, for the most part) are people who have been (slightly) more dominant on the buzzer: Ken Jennings, Joon Pahk, Jerome Vered, Brian Weikle, Michael Dupee. (The last three were limited to five games, so I am basing their comparison on each player's first five games.)OntarioQuizzer wrote: ↑Thu Apr 18, 2019 9:46 amIt's always been the whole game. It just becomes more prominent when one player has significantly better timing than his opposition.
(The measure I'm using is how many times they buzzed in first. Of course, that's driven by a combination of knowledge and timing, but it's the best observable measure of dominance in the control of the board.)
- Volante
- Harbinger of the Doomed Lemur
- Posts: 9265
- Joined: Thu Jul 14, 2011 11:42 pm
Re: So...what have we learned so far from James Holzhauer?
One thing about buzzer dominance, James tends to ease up near the end of the game which would throw off the average if you're expecting the same level of intensity through the game.gnash wrote: ↑Thu Apr 18, 2019 10:52 amJames is one of the most dominant players on the buzzer, and it's key to his success, but there (in fact, here, for the most part) are people who have been (slightly) more dominant on the buzzer: Ken Jennings, Joon Pahk, Jerome Vered, Brian Weikle, Michael Dupee. (The last three were limited to five games, so I am basing their comparison on each player's first five games.)OntarioQuizzer wrote: ↑Thu Apr 18, 2019 9:46 amIt's always been the whole game. It just becomes more prominent when one player has significantly better timing than his opposition.
(The measure I'm using is how many times they buzzed in first. Of course, that's driven by a combination of knowledge and timing, but it's the best observable measure of dominance in the control of the board.)
The best thing that Neil Armstrong ever did, was to let us all imagine we were him.
Latest movies (1-10): An Autumn Afternoon (7), Europa Europa (7), Tampopo (9), Baby Doll (6)
Latest movies (1-10): An Autumn Afternoon (7), Europa Europa (7), Tampopo (9), Baby Doll (6)
-
- Watches Jeopardy! Way Too Much
- Posts: 2981
- Joined: Fri Jul 15, 2011 7:11 am
Re: So...what have we learned so far from James Holzhauer?
Yeah, sure. When the football team leads 38-3 with a minute to go, they put the starting QB on the bench and have the sub take a knee and let the clock run out. Nice when you can afford to do that.
- gnash
- Jeopardy! Champion
- Posts: 1678
- Joined: Fri Jul 01, 2011 9:24 am
Re: So...what have we learned so far from James Holzhauer?
If you assume that it's his conscious decision. Alternatively, he may only be able to sustain the adrenaline rush for so long and his reflexes get slower near the end. That would be consistent with his FJ wagers that are almost prize-maximizing, but leave a little bit of money on the table, as if he's worried that he'll make an arithmetic error at that point.Volante wrote: ↑Thu Apr 18, 2019 11:18 amOne thing about buzzer dominance, James tends to ease up near the end of the game which would throw off the average if you're expecting the same level of intensity through the game.gnash wrote: ↑Thu Apr 18, 2019 10:52 am James is one of the most dominant players on the buzzer, and it's key to his success, but there (in fact, here, for the most part) are people who have been (slightly) more dominant on the buzzer: Ken Jennings, Joon Pahk, Jerome Vered, Brian Weikle, Michael Dupee. (The last three were limited to five games, so I am basing their comparison on each player's first five games.)
(The measure I'm using is how many times they buzzed in first. Of course, that's driven by a combination of knowledge and timing, but it's the best observable measure of dominance in the control of the board.)
If he can't sustain the dominance for 60 clues, then that has to be captured by the measurement, rather that think of something like "peak half" being the "true" measure of board control. (It also has implications for the optimal way to play against him.) If he just relaxes because it's over, then you may have a point that his dominance is under-measured. But then, why would a rational professional gambler give away the easy clues that, for him at that point, represent real money?
-
- Voyeur
- Posts: 1846
- Joined: Mon Sep 26, 2011 12:00 pm
- Location: Princeton, NJ
Re: So...what have we learned so far from James Holzhauer?
We've learned that it is most unlikely that he will ever be a contestant on American Idol or The Voice.
- Volante
- Harbinger of the Doomed Lemur
- Posts: 9265
- Joined: Thu Jul 14, 2011 11:42 pm
Re: So...what have we learned so far from James Holzhauer?
He doesn't maximize FJ wagers either so squeezing every last dollar out doesn't seem to be the highest priority.gnash wrote: ↑Thu Apr 18, 2019 11:52 amIf you assume that it's his conscious decision. Alternatively, he may only be able to sustain the adrenaline rush for so long and his reflexes get slower near the end. That would be consistent with his FJ wagers that are almost prize-maximizing, but leave a little bit of money on the table, as if he's worried that he'll make an arithmetic error at that point.Volante wrote: ↑Thu Apr 18, 2019 11:18 amOne thing about buzzer dominance, James tends to ease up near the end of the game which would throw off the average if you're expecting the same level of intensity through the game.gnash wrote: ↑Thu Apr 18, 2019 10:52 am James is one of the most dominant players on the buzzer, and it's key to his success, but there (in fact, here, for the most part) are people who have been (slightly) more dominant on the buzzer: Ken Jennings, Joon Pahk, Jerome Vered, Brian Weikle, Michael Dupee. (The last three were limited to five games, so I am basing their comparison on each player's first five games.)
(The measure I'm using is how many times they buzzed in first. Of course, that's driven by a combination of knowledge and timing, but it's the best observable measure of dominance in the control of the board.)
If he can't sustain the dominance for 60 clues, then that has to be captured by the measurement, rather that think of something like "peak half" being the "true" measure of board control. (It also has implications for the optimal way to play against him.) If he just relaxes because it's over, then you may have a point that his dominance is under-measured. But then, why would a rational professional gambler give away the easy clues that, for him at that point, represent real money?
The best thing that Neil Armstrong ever did, was to let us all imagine we were him.
Latest movies (1-10): An Autumn Afternoon (7), Europa Europa (7), Tampopo (9), Baby Doll (6)
Latest movies (1-10): An Autumn Afternoon (7), Europa Europa (7), Tampopo (9), Baby Doll (6)
- squarekara
- J! Reactionary
- Posts: 1564
- Joined: Wed Feb 08, 2017 12:49 am
- Location: USDA Zone 5
-
- Newbie
- Posts: 5
- Joined: Wed Apr 17, 2019 8:30 am
Re: So...what have we learned so far from James Holzhauer?
I think there are some things of interest from watching his games.
Yes, be better on the buzzer and know more of the trivia are obvious, and the bulk of the reason for his incredible performance. But there are still meaningful things to be learned from his play.
1. The Daily Double in the Jeopardy round should be an all or nothing affair (generally, all).
Obviously, most people aren't aggressive enough. But at this point in the game, with how much money is still available (even if it's the last answer on the board), I don't see the value in holding things back.
If you get it right, and you didn't bet the max, you've left money on the board.
If you get it wrong, there's one of 2 scenarios:
Either you are good enough to recoup your loses with the large amount of money still available, like James has been.
Or you aren't good enough to do so, so your only chance of beating the better players that you are facing was to maximize your daily double opportunities.
The "nothing" part of the "all or nothing" is if you are hopeless in that category, and the only benefit is denying the double opportunity to your competitors.
2. The Holzhauer horizontal -
He's not just hunting for the DD, he's maximizing his score by hitting the bottom row first. If you are a stronger player (buzzer, questions) so that you are more likely to have control of the board, this seems like a great strategy. I'm not sure how much better you have to be to do this rather than daily double hunt (which are more likely to be in the 4th and then 3rd rows than the bottom).
Interestingly, in Double Jeopardy, James has switched to starting in the 3rd row rather than directly at the bottom. Probably trying to max his advantage and get the DD off the board while he has a big lead, rather than trying to max the amount he can achieve. I suspect if things were even or if he was behind going into Double Jeopardy (an uncommon occurrence), he would start at the bottom instead of the 3rd row.
We've seen people DD Hunt and the Forrest bounce plenty of times, but I'm not sure we've ever seen a player try to leverage value like he has.
Yes, be better on the buzzer and know more of the trivia are obvious, and the bulk of the reason for his incredible performance. But there are still meaningful things to be learned from his play.
1. The Daily Double in the Jeopardy round should be an all or nothing affair (generally, all).
Obviously, most people aren't aggressive enough. But at this point in the game, with how much money is still available (even if it's the last answer on the board), I don't see the value in holding things back.
If you get it right, and you didn't bet the max, you've left money on the board.
If you get it wrong, there's one of 2 scenarios:
Either you are good enough to recoup your loses with the large amount of money still available, like James has been.
Or you aren't good enough to do so, so your only chance of beating the better players that you are facing was to maximize your daily double opportunities.
The "nothing" part of the "all or nothing" is if you are hopeless in that category, and the only benefit is denying the double opportunity to your competitors.
2. The Holzhauer horizontal -
He's not just hunting for the DD, he's maximizing his score by hitting the bottom row first. If you are a stronger player (buzzer, questions) so that you are more likely to have control of the board, this seems like a great strategy. I'm not sure how much better you have to be to do this rather than daily double hunt (which are more likely to be in the 4th and then 3rd rows than the bottom).
Interestingly, in Double Jeopardy, James has switched to starting in the 3rd row rather than directly at the bottom. Probably trying to max his advantage and get the DD off the board while he has a big lead, rather than trying to max the amount he can achieve. I suspect if things were even or if he was behind going into Double Jeopardy (an uncommon occurrence), he would start at the bottom instead of the 3rd row.
We've seen people DD Hunt and the Forrest bounce plenty of times, but I'm not sure we've ever seen a player try to leverage value like he has.
- gnash
- Jeopardy! Champion
- Posts: 1678
- Joined: Fri Jul 01, 2011 9:24 am
Re: So...what have we learned so far from James Holzhauer?
I don't think it is possible for a human to objectively maximize the value of DD wagers in real time unless there are only a couple of clues left on the board. (Watson may do that.) I can't find any flaw in his DD wagering in the sense that I could suggest an improvement without relying on something like a computer simulation.Volante wrote: ↑Thu Apr 18, 2019 12:03 pmHe doesn't maximize FJ wagers either so squeezing every last dollar out doesn't seem to be the highest priority.gnash wrote: ↑Thu Apr 18, 2019 11:52 amIf you assume that it's his conscious decision. Alternatively, he may only be able to sustain the adrenaline rush for so long and his reflexes get slower near the end. That would be consistent with his FJ wagers that are almost prize-maximizing, but leave a little bit of money on the table, as if he's worried that he'll make an arithmetic error at that point.Volante wrote: ↑Thu Apr 18, 2019 11:18 amOne thing about buzzer dominance, James tends to ease up near the end of the game which would throw off the average if you're expecting the same level of intensity through the game.gnash wrote: ↑Thu Apr 18, 2019 10:52 am James is one of the most dominant players on the buzzer, and it's key to his success, but there (in fact, here, for the most part) are people who have been (slightly) more dominant on the buzzer: Ken Jennings, Joon Pahk, Jerome Vered, Brian Weikle, Michael Dupee. (The last three were limited to five games, so I am basing their comparison on each player's first five games.)
(The measure I'm using is how many times they buzzed in first. Of course, that's driven by a combination of knowledge and timing, but it's the best observable measure of dominance in the control of the board.)
If he can't sustain the dominance for 60 clues, then that has to be captured by the measurement, rather that think of something like "peak half" being the "true" measure of board control. (It also has implications for the optimal way to play against him.) If he just relaxes because it's over, then you may have a point that his dominance is under-measured. But then, why would a rational professional gambler give away the easy clues that, for him at that point, represent real money?
But not trying to grab all the 400 and 800 clues after you've secured a lock would obviously be leaving real money on the table. And I don't know how to interpret slightly underwagering FJ other than as an insurance policy against a stupid arithmetic mistake. (It's a real risk, BTW. My policy was doing the math several times, which had the benefit of being free.)
- gnash
- Jeopardy! Champion
- Posts: 1678
- Joined: Fri Jul 01, 2011 9:24 am
Re: So...what have we learned so far from James Holzhauer?
These are good points, but nothing new and nothing that anyone who has seriously thought about J! strategy didn't know before James appeared.ObrienP wrote: ↑Thu Apr 18, 2019 12:07 pm I think there are some things of interest from watching his games.
Yes, be better on the buzzer and know more of the trivia are obvious, and the bulk of the reason for his incredible performance. But there are still meaningful things to be learned from his play.
1. The Daily Double in the Jeopardy round should be an all or nothing affair (generally, all).
Obviously, most people aren't aggressive enough. But at this point in the game, with how much money is still available (even if it's the last answer on the board), I don't see the value in holding things back.
If you get it right, and you didn't bet the max, you've left money on the board.
If you get it wrong, there's one of 2 scenarios:
Either you are good enough to recoup your loses with the large amount of money still available, like James has been.
Or you aren't good enough to do so, so your only chance of beating the better players that you are facing was to maximize your daily double opportunities.
The "nothing" part of the "all or nothing" is if you are hopeless in that category, and the only benefit is denying the double opportunity to your competitors.
2. The Holzhauer horizontal -
He's not just hunting for the DD, he's maximizing his score by hitting the bottom row first. If you are a stronger player (buzzer, questions) so that you are more likely to have control of the board, this seems like a great strategy. I'm not sure how much better you have to be to do this rather than daily double hunt (which are more likely to be in the 4th and then 3rd rows than the bottom).
Interestingly, in Double Jeopardy, James has switched to starting in the 3rd row rather than directly at the bottom. Probably trying to max his advantage and get the DD off the board while he has a big lead, rather than trying to max the amount he can achieve. I suspect if things were even or if he was behind going into Double Jeopardy (an uncommon occurrence), he would start at the bottom instead of the 3rd row.
We've seen people DD Hunt and the Forrest bounce plenty of times, but I'm not sure we've ever seen a player try to leverage value like he has.
I argued point 1 on the old Sony board way back during Madden's run. DD wagers made by the vast majority of players are bad. Unless it's the end of the game and the optimal wager is calculable, you need to go either very low or very high. Madden figured out the "low" part of it. Roger Craig was the first to demonstrate on a consistent basis over a long enough run the advantages of the "high" part. (My own run was too short as I suck on the buzzer. Plus, my perception of what was "high" 11 years ago was influenced by the hive mind and biased down.)
Point 2 depends on the situation. If you have exceptional control of the board and are confident in all categories, then it's clearly optimal. But in that case, you already are a Ken Jennings-caliber player and Ken won 74 games with no strategy beyond Kids' Week level. In other words, the HH is not a useful strategic advice for a merely good player trying to maximize the win probability. It's an advice for an insanely good player to maximize the already almost assured winnings.
But some variants of this strategy are well-known and used by well-prepared mortals. For example, if you have to go first in DJ (which means you're behind), it's usually best to start your favorite category (the one where you think you have a comparative advantage) from the bottom, hoping you'll keep control and find the DD in the middle row. But if you go to the $2000 clue in your bad category, you'll probably just lose control; that's not optimal.
- gnash
- Jeopardy! Champion
- Posts: 1678
- Joined: Fri Jul 01, 2011 9:24 am
Re: So...what have we learned so far from James Holzhauer?
I should add: You cannot know whether you are an insanely good player (a monster on the buzzer) until you actually start playing - and a few such players took a game or two to reach that level.
-
- Watches Jeopardy! Way Too Much
- Posts: 632
- Joined: Wed Jan 04, 2012 6:50 pm
Re: So...what have we learned so far from James Holzhauer?
Yeah, I got that, but WHY don't they want people to buzz in early. Personally I feel if you know it, you should be able to buzz in whenever (with the exception of there being contestants like Eddie Timanus).This Is Kirk! wrote: ↑Wed Apr 17, 2019 5:58 pmBecause they want a disincentive for buzzing in early.
-
- Newbie
- Posts: 5
- Joined: Wed Apr 17, 2019 8:30 am
Re: So...what have we learned so far from James Holzhauer?
Well, they want Alex to finish the question. Jeopardy is an entertainment show for the audience first, and trivia contest second. If it is not entertaining for the viewing audience, the show ceases to exist.ACW wrote: ↑Thu Apr 18, 2019 1:00 pmYeah, I got that, but WHY don't they want people to buzz in early. Personally I feel if you know it, you should be able to buzz in whenever (with the exception of there being contestants like Eddie Timanus).This Is Kirk! wrote: ↑Wed Apr 17, 2019 5:58 pmBecause they want a disincentive for buzzing in early.
- jeff6286
- Watches Jeopardy! Way Too Much
- Posts: 5233
- Joined: Fri Jul 01, 2011 7:34 pm
- Location: Indianapolis, IN
Re: So...what have we learned so far from James Holzhauer?
I really don’t think worrying about a math error has anything to do with it. His wagering plan in final seemed clear, find the maximum safe bet he can make, and then find a number slightly below that with personal significance (usually within 1-2000 or so). It’s true that with as good as he is at final he is basically giving away sums in the thousands in exchange for making birthday shoutouts, I find that an odd choice but it is certainly his right. By the time you get to a half million in winnings those thousand dollars left here and there are pretty unimportant I suppose.gnash wrote: ↑Thu Apr 18, 2019 12:20 pmI don't think it is possible for a human to objectively maximize the value of DD wagers in real time unless there are only a couple of clues left on the board. (Watson may do that.) I can't find any flaw in his DD wagering in the sense that I could suggest an improvement without relying on something like a computer simulation.Volante wrote: ↑Thu Apr 18, 2019 12:03 pmHe doesn't maximize FJ wagers either so squeezing every last dollar out doesn't seem to be the highest priority.gnash wrote: ↑Thu Apr 18, 2019 11:52 amIf you assume that it's his conscious decision. Alternatively, he may only be able to sustain the adrenaline rush for so long and his reflexes get slower near the end. That would be consistent with his FJ wagers that are almost prize-maximizing, but leave a little bit of money on the table, as if he's worried that he'll make an arithmetic error at that point.Volante wrote: ↑Thu Apr 18, 2019 11:18 amOne thing about buzzer dominance, James tends to ease up near the end of the game which would throw off the average if you're expecting the same level of intensity through the game.gnash wrote: ↑Thu Apr 18, 2019 10:52 am James is one of the most dominant players on the buzzer, and it's key to his success, but there (in fact, here, for the most part) are people who have been (slightly) more dominant on the buzzer: Ken Jennings, Joon Pahk, Jerome Vered, Brian Weikle, Michael Dupee. (The last three were limited to five games, so I am basing their comparison on each player's first five games.)
(The measure I'm using is how many times they buzzed in first. Of course, that's driven by a combination of knowledge and timing, but it's the best observable measure of dominance in the control of the board.)
If he can't sustain the dominance for 60 clues, then that has to be captured by the measurement, rather that think of something like "peak half" being the "true" measure of board control. (It also has implications for the optimal way to play against him.) If he just relaxes because it's over, then you may have a point that his dominance is under-measured. But then, why would a rational professional gambler give away the easy clues that, for him at that point, represent real money?
But not trying to grab all the 400 and 800 clues after you've secured a lock would obviously be leaving real money on the table. And I don't know how to interpret slightly underwagering FJ other than as an insurance policy against a stupid arithmetic mistake. (It's a real risk, BTW. My policy was doing the math several times, which had the benefit of being free.)
As to the not trying to mop up every 400 and 800 clue at the end when the game is long decided, I haven’t seen any reason to believe he is doing this on purpose. There is certainly the possibility of a slight letdown in adrenaline, aggression, or focus once the game is well in hand, a perfectly reasonable thing for a human to have happen.
And I don’t think it has been that pronounced in most games. It is logical that he would lose a higher percentage of buzzes on the top rows because everyone is usually going to be buzzing on those, while the 1600/2000 boxes will frequently only have one or two players ringing in.