Denny McLain, the AJ Score, and Predicting the 2017 Jeopardy! Tournament of Champions
When someone finds out that I’m a Jeopardy! champion, one of the first things they usually ask is how many games I won. And, while I’m extremely proud and grateful to have won six games over the course of my original run on the show, I’ve always felt like that number alone doesn’t quite reflect the true strength of my performance over those seven episodes.
To me, wins on Jeopardy! are like wins for baseball pitchers. As a stat, it’s intuitive and memorable, which makes it ideal for records and milestones. Any Jeopardy! superfan could tell you that Ken Jennings is the all-time wins leader, and Julia Collins is the last player to win 20, just as any baseball historian could tell you that Cy Young is the all-time wins leader, and Denny McLain is the last player to win 30.
Going into the Tournament of Champions, all eyes are generally on the players who managed to rack up the most wins. But in the history of the show, few champions have been able to convert the de facto number-one seed into the $250,000 grand prize. In 2004, Tom Walsh (7 wins) finished 2nd to Russ Schumacher (4). In 2006, David Madden (19) was taken out in the semis by Bill MacDonald (4), who himself fell in the finals to Michael Falk (3). One exception is the 2009 event, when Dan Pawson (9) came from behind to win, the result of Larissa Kelly (6) missing the deciding Final clue. In 2010, Vijay Balse (4) took down Justin Bernbach (7) and Jason Zollinger (6). In 2011, Roger Craig (6) beat Joon Pahk (7) and Tom Nissley (8) en route to the title. In 2014, Arthur Chu (11) and Julia Collins (20) finished a respective 2nd and 3rd to Ben Ingram (8).
At this time two years ago, it looked like the winningest player in my tournament would be Greg Seroka (7), whose episodes began airing the same week my own stint at the champion’s lectern came to an end. But by the time tape day rolled around, the presumptive top seed was Matt Jackson (13), the reigning champ in TV-land, whose presence in the tournament was my only evidence that he’d ever been beaten. As it turned out, Greg didn’t make it out of the first round, and Matt lost in the finals to yours truly (6).
The 2017 Tournament of Champions is in the can, so to speak, and we now know that this year’s Matt Jackson was Austin Rogers (12), a complete unknown to the rest of the field at the time of taping. Austin, Seth Wilson (12), and Buzzy Cohen (9), will be considered by many to be the favorites to make the finals and compete for $250,000 on November 16–17. But are they truly the three best players in the tournament? In search of a better way to compare Jeopardy! champions, I’ve invented a new stat, which I’m simply calling the Average Jeopardy Score, or AJ Score.
Though there are three rounds in a game of Jeopardy, the second round is essentially a higher stakes version of the first round. So it makes sense to think of a game as made up of two distinct subgames, “Jeopardy” and “Final Jeopardy”. There’s much more randomness built into Final Jeopardy — imagine a basketball game in which field goals are worth 100 points in the final minute of play. In addition, the strategy of Final Jeopardy has an equalizing effect — a player with a small lead has to bet big, while one with a big lead often bets small. For those reasons, I’m much more interested in what happened over the first sixty questions than the last one.
For a given game, I define a player’s “Jeopardy Score” as the player’s score going into Final Jeopardy. Going back to the well of the sports analogy, there’s not really a defense element on Jeopardy!, so there’s not much you can do to cause your opponents to make errors. All you can do is run your offense as effectively as you can and hope it’s good enough to win. When comparing the performance of two Jeopardy! champions, I want to know who ran the better offense on average, irrespective of win total. Win total is a reflection of the outcome of a small sample of games, while AJ Score is based on the results of a much larger sample of clues. Therefore, I believe AJ Score is a better predictor of future success on Jeopardy!, like, say, in the 2017 Tournament of Champions.
If including Final Jeopardy would skew things too much, why not also limit the effects of the Daily Double, as the Coryat Score (http://www.j-archive.com/help.php#coryatscore) does? Now famous in the Jeopardy! superfan community, the Coryat Score was devised to gauge a hopeful’s ability to pass the qualifying test. Viewed through that lens, it makes sense why you’d want to make the Daily Doubles just like any other question, since all questions are weighted equally on the test. However, when you’re actually playing, Daily Doubles are a huge part of the game. There’s skill in finding a Daily Double, there’s skill in wagering, and there’s skill in responding correctly. All of that is fully reflected in the AJ Score, which is why I prefer it to the Average Coryat Score. Consider a hypothetical clairvoyant player who finds every Daily Double, always bets it all, and always responds correctly. This contestant would be trouncing people, and the AJ Score would accurately reflect that dominance, while the Average Coryat Score would not.
In 1968, Denny McLain of the Detroit Tigers became the first pitcher to win 30 or more games during the regular season since Dizzy Dean in 1934. His flashy 31 wins helped him win both the American League Cy Young and MVP Awards unanimously, meaning that he was voted not just the league’s best pitcher, but its best overall player. He even found himself on the cover of Time magazine. But in the World Series that year, he struggled, losing Games 1 and 4 to future Hall of Famer Bob Gibson. Teammate Mickey Lolich, on the other hand, carried the Tigers with 3 complete-game victories, including a gem in Game 7 over Gibson, and was rightfully awarded World Series MVP honors. His win total that season? 17.
So should McLain have won that regular season MVP award in the first place? Nowadays, many baseball analysts rely on advanced stats, one of the most popular of which is WAR (wins above replacement), which uses a complex formula to quantify a player’s value to his team. According to baseball-reference.com, McLain actually finished 4th in the league in WAR that year, behind future Hall of Famers Carl Yastrzemski and Brooks Robinson, and Indians starting pitcher Luis Tiant, who finished the year with a better ERA, better WHIP, and better K/9 than McLain. And it was actually Bob Gibson who led the majors in WAR that season, with only 22 wins, but an eye-popping 1.12 ERA, a mark that hasn’t been beaten since. Maybe he should have been on the cover of Time in September of ’68, instead of McLain.
Putting my sabermetrician’s hat on, I’ve put together a suite of Jeopardy! stats. You already know the AJ Score, and the rest are defined below. I’ve tabulated the results for the 2017 Tournament of Champions field, as well as the decade’s past TOC finalists, for historical context (data pulled from www.j-archive.com).
Lock%: percentage of games in which the player had a lock win (more than double the score of the closest opponent going into Final)
Lead%: percentage of games in which player had the lead going into Final
Find%: percentage of Daily Doubles which the player found
DD%: percentage of found Daily Doubles to which the player responded correctly
FJ%: percentage of Final Jeopardy clues to which the player responded correctly
UEPG: unforced errors per game (incorrect responses per game, not including Daily Doubles or Final Jeopardy)
Now that we have the stats, let’s take a look at this week’s matchups, and make some predictions for the tournament (for the record, I am completely spoiler-free.) For quick reference, I’ve included each champion’s AJ Score rank.
Monday, November 6
Buzzy Cohen (11th) vs. Hunter Appler (8th) vs. Pranjal Vachaspati (3rd)
Tuesday, November 7
Tim Aten (14th) vs. Lilly Chin (6th) vs. Jason Sterlacci (4th)
Wednesday, November 8
Austin Rogers (1st) vs. Alan Lin (7th) vs. David Clemmons (13th)
Thursday, November 9
Seth Wilson (5th) vs. Lisa Schlitt (10th) vs. Sam Deutsch (15th)
Friday, November 10
Andrew Pau (2nd) vs. Justin Vossler (9th) vs. Jon Eisenman (12th)
As expected, the top five seeds according to win total and money won (Austin, Seth, Buzzy, Tim, Andrew) have been separated. But I was delighted to see that my top five seeds according to AJ Score have also been separated! I’m going to predict that those five players (Austin, Andrew, Pranjal, Jason, and Seth) advance to the semis. The remaining four spots are tough to call — it may come down to which episodes have the easier Final clues. I think since the competition will be tougher, resulting in fewer scoring opportunities, it will be important for the players to limit their mistakes. Therefore, I’ll also pick Tim (0.86 UEPG, lowest in the field), and Alan (1.29 UEPG, 3rd lowest). In a one-game sample, accuracy on the high-value questions (DD and FJ) is also important, so for the last two spots in the semis, I’ll choose Lisa and Hunter, who were both over 70% on DDs and FJ.
At this point, even if we assume my nine picks advance, it’s extremely hard to predict what the matchups would be. It’s clear that Austin and Seth would be separated due to their field-topping win totals, and I think Tim would be placed in the third game, with the next highest win total. I’ll predict that Austin advances to the finals. I think that the three finalists will come from my top five seeds according to AJ Score, but I’ll make a wild guess and predict that Andrew will fall to Austin while Pranjal and Jason will knock off Seth and Tim, and make the finals.
Austin’s AJ Score of $24,331 is the highest in the field, bettered only by myself, Matt Jackson, and Roger Craig among the decade’s TOC finalists. His FJ% is also the highest in the field by far, having only missed one of thirteen. If he can translate that accuracy to the notoriously difficult Final clues in the TOC finals, he will be extremely tough to beat. If he has a weakness, it’s one that doesn’t show up in the stats — Final Jeopardy wagering. If he’s improved that part of his game since his original run, I really love his chances.
Pranjal’s strength is daily double play. His Find% (67%) and DD% (86%) are strong, and he also showed the aggressiveness necessary to beat great players like Austin and Jason. I want to flag up one moment during Pranjal’s run that particularly impressed me. With a big lead in Double Jeopardy, Pranjal found a daily double and wagered $7,500, almost half of his total. He missed, only to find the last daily double on the very next clue! Without hesitation, he wagered $7,000, almost all of his remaining money, when most players would have been gunshy. However, Pranjal’s aggressiveness can also be seen as a weakness when it comes to buzzing in. His UEPG of 4.57 is by far the worst in the field. But, as I said with Austin, if he’s aware of this and makes an effort to limit his errors, I think he has a great shot to win.
Jason dominated his Teachers Tournament, leading all four games before Final Jeopardy and locking up three of them. Even more impressive, he won his tournament in a super-lock (a guaranteed win regardless of his wager on the last Final Jeopardy clue), the first person to do that since… well, me. And he was able to do it while only finding four of the twelve daily doubles. However, I don’t think he’ll be able to run through this field without the benefit of daily doubles, and I see his low Find% as his biggest weakness. From what I can tell, looking for daily doubles isn’t part of Jason’s game. In fact, his Teachers Tournament final was remarkably friendly — all twenty-four categories were taken in order! I certainly don’t think that will be true for the 2017 TOC finals, and it will be interesting to see if Jason shows up with a different mindset regarding taking the clues in order.
As for my final prediction, I’ll stick with the AJ Score and predict that Austin will win it all. I can’t wait to watch and see how it plays out.
P.S. If you’ve made it this far, thanks for sticking with me. I hope you enjoyed this Jeopardy! nerd manifesto. You can follow me on Twitter and Instagram at @whoisalexjacob and you can like my Facebook page at www.facebook.com/whoisalexjacob.