Back in 2004, I tracked Ken Jennings’ statistics daily and updated a toteboard. With James Holzhauer dominating his Jeopardy! games in ways unseen in years, I’ve decided to track James’ daily statistics in many of the same categories, alongside Ken’s statistics at the same point in his streak from 2004. Links are also provided to help define any statistics that may not be readily familiar or intuitive to the casual viewer.
This page will be publicly updated daily between 8:00 and 9:00 PM Eastern. (If you are a member of the media wishing to see the comparison earlier in the day, you’re welcome to contact Andy privately.)
Last Updated: June 3, 2019 (through 33 games)
|James Holzhauer||Ken Jennings|
|Unforced Errors (buzz in w/ incorrect response)||31||88|
|Correct Response % When Buzzing In||97.21%||92.45%|
|Times First In||1077||1112|
|First In %age||57.81%||59.31%|
|Rebound %age (of all rebound opportunities)||48.53%||40.19%|
|Categories With 5 Correct||36||38|
|Categories With 4 Correct||95||95|
|Categories With 3 Correct||122||115|
|% Of Categories With At Least 3 Correct||63.89%||62.63%|
|Categories With 0 Correct||11||15|
|% of Bottom-Row Clues Responded To Correctly||58.59%||53.16%|
|Daily Doubles Played||76||69|
|Daily Doubles Correct||72||59|
|Daily Double %age||94.74%||85.51%|
|Average Daily Double Wager||$8,984||$3,265|
|Total Net Gain On Daily Doubles||$654,416||$159,299|
|Average Net Gain On Daily Doubles||$8,611||$2,309|
|Average Score At End Of First Round||$12,564||$9,327|
|Average Lead At End Of First Round||$8,273||$5,555|
|Average Score At End Of Second Round||$47,655||$30,015|
|Average Lead At End Of Second Round||$36,750||$20,730|
|% Of Runaway Games||87.88%||84.85%|
|Final Jeopardy Clues Correct||32||22|
|Final Jeopardy %age||96.97%||66.67%|
|Average Final Jeopardy Wager||$27,891||$7,292|
|Average Final Score||$75,364||$33,347|
|Average Margin of Victory||$62,269||$25,144|
|Average Coryat Score||$30,576||$27,552|
|Average Opposition Player Coryat||$7,415||$5,948|
|Average Combined Coryat||$45,400||$39,448|
|Triple Stumpers Average||3.39||5.36|
|Unrevealed Clues Average||0.55||0.18|
Since Alex Trebek’s diagnosis of stage 4 pancreatic cancer, many community members have been raising money. The Jeopardy! Fan Online Store is as well! All proceeds from any “Keep The Faith And We’ll Win” shirt sold will be donated to the Pancreatic Cancer Action Network. To date, over $370 has been raised.)
If you are going to quote any statistics from this page or this website, attribution is required.
Have you had a chance to listen to our new podcast game show, Complete The List, yet? Check it out! It’s also available on Apple Podcasts.
Can I recommend linking or explaining “unforced errors”? I managed to find an explanation on the Ken Jennings page but I imagine I’m not going to be the only person with that question.
Questions a contestant is not forced to answer. James has 20 incorrect responses, but 3 are DD and 1 was FJ, meaning the other 16 were times he buzzed in but answered incorrectly, which is an ‘unforced error’.
I suppose a forced error is an incorrect Daily Double or Final Jeopardy answer. Is there any other incorrect answer that is not an unforced error?
I’m so thrilled to discover your webpage! I’ve been doing some basic calculations on James while re-watching his biggest wins, boring my husband with daily updates. So it’s a lot of fun to see your comprehensive breakdown and comparison to Ken. Thanks!
I’m still a little confused on the “unforced errors”? Wouldn’t that be the same as an incorrect response? Because in order to get an incorrect response, you need to have buzzed in.
On Daily Doubles and Final Jeopardy, you can be incorrect without having buzzed (essentially, your response is forced).
Ohhhh! Okay, that makes sense!
Also, thank you for doing this! Until we found this, my mom and I were doing our best to do these stats on my planner with very little room.
What is a rebound ? Thank you. Love this site.
A rebound is when one player responds incorrectly, and another player then gives the correct response.
I read a bit of your methodology for predicting the win streak and it’s basically fitting the coryat score to a curve to get win probability. I’m not sure if that approach will work for James since the coryat score totally ignores daily double betting and James’s betting strategy is pretty unusual for a player of his caliber.
Traditionally players go with the risk averse small bets but James relies on larger bets to build up an insurmountable lead early on and getting a margin of safety later. It’s not clear to me if his aggressive betting will put him above the curve or below it.
He’s betting correctly, not aggressively. He’s playing a more volatile game, when the odds of of him winning are lowest, but later in the game, if he’s acquired a large enough lead he bets based off of his and his opponents chip stacks so to not put his lead in jeopardy, should he get it wrong (that is also controlling for category confidence).
Despite the bets being larger later in the game, they’re more conservative relative to his stack. If anything, the way he opts for clues and sizes his bets should give him more of an advantage and bump his odds of winning because he plays a correct strategy. Not one dollar amount or category is chosen randomly, as others have done when “jumping around the board.”
Hello there, Andy. I just want to say that you’re doing a great job comparing the stats between James Holzhauer and Ken Jennings and it’s all very impressive, but there’s some questions I want to know if you can answer: 1st, What are the odds that James manages to break $1,000,000 next week if he’s still on the show and 2nd, what are the odds he breaks Julia Collins’ 20-day winning record? Until then, have a Happy Easter!
I am new to this website I have a few questions. James has been on different shows doing interviews so he must be under contract not to say how many more matches he’s going to win.Right? I know the shows have been pre-taped but what I don’t know is how long ago they were taped? So it’s a closely guarded secret how many more matches he’s going to win. When did they stop taping for the season? And how many more matches before this season actually ends?
Taping stopped for the season April 6; the final game of the season airs July 26.
This is great! Feels like 96% is too conservative given that it doesn’t take into consideration with how a player plays the board. Feels like you need to add in efficiency for finding the daily double to Coriat score, especially for multi-day winners. The way he picks and chooses his clues is 100% not random and based off of stack sizes, confidence, and a consequent probability of winning. His bet sizes work the same way on the daily doubles. Obviously with average Coriat score, comes higher probably of winning, but an ability to understand and adapt strategy based on your intra-game probability of winning is a massive advantage to place yourself in a significant position to lock in a win.
Also, can you guys look at studying and weight ring in percentage (first to ring in) by dollar value? Feels like it should be weighted by that — players are clearly more apprehensive to ring in for higher clues, I’d suggest $ rung in / total $… but obviously can altered based on any other study that makes sense.
I mean, there is something to be said for the fact that his style of play inherently keeps his opponents from building their own score off Daily Doubles; the current prediction model doesn’t account for that at all. The fact that it’s at 96%, though, is because there are going to be times where he’s either going to miss a crucial Daily Double (or his opponent will get one), and his score will be at a level where he’s only projected to win 80% of the time (or less!)
His daily double and final jeopardy percentages are unusually high. I wonder how statistically unusual it is that his percentages for those things that require betting are so high… Would the hypothesis that his chance of getting daily double questions correct are the same as his chances of getting similarly-valued questions correct be rejected in a t-test?
If you have data at that detail I’d love to run the analysis.
Brian: It would not be possible to run a comparison because a “normal” clue requires a competition on the signalling device in order to respond. We have no way of knowing which clues James is not attempting to signal on, or even his accuracy rate on clues where he has attempted to signal but was not first in.
I asked this previously, and perhaps it got lost in the nether. I was curious if there was chance at see a comparison between Ken and James that shows the difference in each day? Once you post the new day, seeing what happened on any previous day is not available. Thanks, love your stats.
I’m working on this and should have this up in the next few days.
One factor in James’ favor is that none of the contestants that he faces for the rest of the season will have had an opportunity to watch him compete, and been prepared to counter his tactics. I assume that they arrive on the set “blind”, and only then learn that they are facing a multi-game, $$$$$-winning champion. It will be interesting to see if he manages to win-out through the end of the taped season. If he continues on to the next season, everyone he faces would have had a chance to watch his streak.
You’re the man Andy Saunders, Thanks!
Not enjoying watching past weeks episodes as no doubt about who will win, hopefully someone will come along to beat him soon
Even Ken Jennings lost eventually. Someone will come along and beat James. As for me, I love watching the game being played to its utmost potential.
Michael Schwartz – his opponents watch him play before they play him. Contestants for all of the day’s shows watch the matches before theirs. I think this helps build an intimidation factor.
Thank you for this great web site!
And your speed is greatly appreciated.
This is the first place I visit after each game.
Keep up the great work!
What about adding in the only other person “Brad Rutter” to come near this much money won to this equation.
Brad’s winnings have been mostly accumulated as a result of tournaments. It would not be an apples-to-apples comparison.
Thank you Andy. I did not start watching until my mother got me hooked on it.
First, thanks for a fascinating website!
When Ken Jennings was competing, were the board dollar amounts the same as they are now? Someone told me she thinks they were half of what they are today.
The board values were the same in Ken Jennings’ day compared to today.
This is great stuff, thanks. I wonder if it would be useful to have data on how much a contestant bets on Double and Final Jeopardy clues in relation to their pre-bet total (the true Daily Double being 100%, for example). In addition to the bet in terms of raw numbers, the percentage can give an idea of how aggressive a player plays (because $4,000 may be aggressive when one has a score of $6,000 but conservative when one is at $20,000).
I’ll think about this. I do track “Daily Double Efficiency” already on the daily threads.
One small quibble with this otherwise fantastic collection of data.
“With a projected 83 regular-play games to go prior to the Tournament of Champions cutoff: James Holzhauer qualified 100.000% of the time.”
Should James still be the champion at the cutoff date, he will technically not qualify for this 2019 TOC. You might consider putting a second line for James, which includes the model projections that he’s actually still the champion at the cutoff date.
All the model cares about is whether James will qualify for any ToC, not necessarily the next one. If you’re curious as to the chance the he is still champion, you can look at Jonathan Dinerstein’s chances, as he is only in if James is still champion.
I dont get why this comparison keeps going because he’s clearly better than Ken Jennings in every category worth comparing.
James Holzhauer is a phenomenal player. Prodigious knowledge & rapid recall are matched by thorough mastery of wagering theory & quick calculation (he’s a professional gambler with a math degree). Ken Jennings played much more conservatively, knowing that his best chance of winning a lot of $ was to win a lot of games (which he did), not bet a lot in any one game. The styles are different, each appropriate to the player. One is not “clearly better” than the other.
The reason this comparison keeps going is that we enjoy thinking about such things. If you don’t…
Hi, can you show me where you got your info about him having a math degree? My mom and I have been trying to find out what his education level is, just for the heck of it.
BS Mathematics University of Illinois.
As far as I know that’s the extent of his formal academic education. He claims (in jest) to have majored in online poker at school, certainly played IRL poker, and moved to Las Vegas soon after graduating, so his informal education overlapped (one might even say “lapped”) his formal education. However, he’s clearly skilled at mathematical calculation, at least.
Thanks for the information and statistics. Really fascinating. I’ve been hooked on Jeopardy! for years but now it’s heart -stopping to watch. I’m rooting for that genius James. And he probably has a photographic memory.
How about his remark about studying children’s books for quick information!
I’ve been a Jeopardy fan since before Alex’s Day…..at 88, I can’t remember the previous host’s name. I loved the time when he was betting, threw a kiss to Heaven and said “this is for you Grammy.” I record it and watch with my son and two grandsons on Sunday….they call me “Grammy”. I told them to remember that when they’re on Jeopardy.
Love the stats. Thanks for creating and updating for all of us to enjoy!
One possible additional metric that would be interesting to see is “Total net gain on Final Jeopardy”
I believe at 21 games it is roughly 577k to 74k.
Pure speculation, but is there any value to changing the game rules such that it is not so controlled by buzzer skill and becomes more of a pure knowledge test? I have the sense that, were Brad, Ken and James to meet only once, the winner would be the one who wins the buzzer.
No. That wouldn’t be Jeopardy!; it would be something Jeopardy!-like, but not actual Jeopardy!.
Very interesting site.
Thanks for crunching all the numbers!
Here’s an interesting scenario that could happen with James. The T.O.C. stats indicate that James is 100% qualified for the tourney. But, if he crosses seasons, like Ken Jennings did, then he doesn’t qualify for this year’s T.O.C. Instead, he will qualify for next year’s T.O.C. And, furthermore, some agreement could take place, as it did with Ken Jennings, where he never takes part in a T.O.C. Ken Jennings waived his right to take part in the T.O.C. when he agreed to be an automatic finalist in the Ultimate Champions Tournament. Just some food for thought.
Thanks for all the stats. Do you happen to have the same stats you compiled for Ken for some of the other Jeopardy! greats? With Holzhauer exceeding Jennings on most stats, I’d be curious to see how he stacks up to other champions as well across all these metrics. To that end, do you also have a sense of what the prediction model was implying for some of these champions. i.e. with Holzhauer at a 97-98% chance to win any given game, how does that stack up to say Julia Collins or Austin Rogers?
You can search for Austin Rogers; the prediction models were operational for his games. For the first 48 games of Ken’s run, challengers received very little practice time, so there isn’t as much of a direct comparison possible.
In the photo of Ken, are those James’ fingers creeping up on Ken’s left shoulder?
I think you will find that to be Alex Trebek’s hand from a picture of Ken posing with Alex. So, I’m guessing your question is in jest.
Yes – just trying to be funny as I knew it was due to cropping and his picture was with Alex. But looking at it post-cropped there is no arm or other person readily visible so at it appears that it is someone slowly creeping up from behind ready to pull him down (like in some bad horror movie) and then I went to – who could that threat be – clearly it is James.
From a Chicago Sun-Times Q&A published after game 25.
Q. What’s your single greatest memory from this historic run?
A. My greatest memory hasn’t aired yet!
That sounds revelatory, but it’s still cagey. Only Season 35 has been recorded. If the “historic run” extends into Season 36 or oh, say, 40, his greatest memory is likely to be of something that hasn’t happened yet, so by definition it hasn’t aired yet. All he’s revealed is that he’s on more games than we’ve seen so far, and we knew that.
hah! correct Karl. I just liked it cause it sounds like something Forrest Gump might say.
As James keeps winning, doesn’t the odds that are on your list of players with a chance to play go up the longer his streak goes on? Haven’t been playing a lot of attention to that but was curious if you looked at.
That was a horrible question, since I failed to identify what list of players I was looking at. TOC list. Sorry, though I think I do know the answer and know how to determine the answer. So, sorry again for such a stupid question. My next comment will be better.
If James makes it through into June, current averages says he’ll break Ken’s regular season win total on June 3. A little less than current averages would make it on June 4.