Scoring / Analysis Tools for Practice

This is where all of the games are discussed.

Moderators: alietr, trainman, econgator, dhkendall

Post Reply
User avatar
koam
Loyal Jeopardista
Posts: 205
Joined: Thu Mar 29, 2012 1:46 am

Scoring / Analysis Tools for Practice

Post by koam »

I've seen a couple of different home practice scoring tools. Which ones are your favorites?

I've used some that do a good job of calculating your $ score for the game, by round.

I would ideally like one that keeps my score for a complete episode, then accumulates statistics by round and by $ level within the round (e.g., what % right/wrong/skipped at the $200 level in Jeopardy round) across multiple episodes. Simpler yet would be "top/bottom half of board" stats.

Real pie in the sky would be if something could keep such stats by generalized categories (geography, history, humanities, sports, pop culture, etc.)..but that's just a dream. I know TOC Champ Roger Craig had built his own computer programs for practice tools that allowed him to analyze his performance by subject area. Since that's out of my league, I'm wondering what tools are out there.

User avatar
the_phil
60% or Bust
Posts: 197
Joined: Tue May 15, 2012 6:08 pm
Location: Raleigh, NC

Re: Scoring / Analysis Tools for Practice

Post by the_phil »

I [somewhat] recently built myself a scoresheet for use in Excel. I use it mostly when I'm "playing" a game from a j-archive on my lunch break, but you can use it while watching the show as well. When I get some time tonight I'll put a link and some screenshots in case anyone's interested in using it. Essentially it functions like this:

1 - Copy template scoresheet
2 - Fill in the date and categories
3 - Leave the box blank if you don't answer. Fill in a "1" for a correct answer, "0" for incorrect. If the question is a daily double, use "D1" and "D0", respectively.
4 - Fill in the combined show Coryat (if you're interested in comparing)
5 - The scorecard for the game updates Coryats automatically
6 - Click the "Calculatez0r!" button on the "Calculators" worksheet to update stats. This sheet captures:

For each game you've played, the following stats are tracked:
Date
Round 1 Coryat
Round 2 Coryat
Total Coryat
% of Coryat accumulated in Rounds 1 and 2 individually
Combined Show Coryat
Coryat from a hypothetical game of 3 people with identical skill level to and random crossover of correct/incorrect responses
% of Show Coryat you scored both by yourself or with 3 of yourself
how you did on DD's in each round
whether or not you got FJ right.

For the collection of games on the sheet, the following stats are tracked:
Average Coryat
Average 3-person Coryat
Average Show Coryat
% show Coryat average (1-person)
[median values for all those stats]
Round 1 DD % right, Round 2 DD % right
Batting average [i.e. % right] for each row in each round, both by yourself or with 3 of yourself



I got into a rhythm and probably went overboard. I've got over 300 games logged on my lunch break, just in the past 8 months...

User avatar
BobF
Watches Jeopardy! Way Too Much
Posts: 2123
Joined: Thu Jul 07, 2011 3:03 pm
Location: Westmonster, MD
Contact:

Re: Scoring / Analysis Tools for Practice

Post by BobF »

I have a "worst case scenario" scoring system because I figure I'd be beaten to the buzzer on some answers. I keep track of number right and number wrong (okay, I'll sound like a true geek here, but by counting in binary on my right and left hands - pinky=1, ring=2, middle=4, index=8, thumb=16). First 6 right I count as the lowest clue score $200 or $400, next 6 count double, next 6 count triple, etc. I never get more than 6 wrong (I clam) and each one wrong I count minus highest clue score $1000 or $2000. I keep track of daily doubles by "wagering" a multiple of $200 or $400 and giving myself extra points plus or minus if I get it right or wrong. If I didn't get the previous question, I don't count the DD.
Was once hugged by Maggie Speak!

User avatar
koam
Loyal Jeopardista
Posts: 205
Joined: Thu Mar 29, 2012 1:46 am

Re: Scoring / Analysis Tools for Practice

Post by koam »

Thank you both. That spreadsheet sounds interesting.

User avatar
DadofTwins
Watches Jeopardy! Way Too Much
Posts: 1206
Joined: Fri Jul 01, 2011 10:11 am
Location: Fortress of SHC-itude
Contact:

Re: Scoring / Analysis Tools for Practice

Post by DadofTwins »

They're too long to re-post here, but feel free to check out the "DoT Matrix" series of posts from last November here:

http://dadoftwins02.blogspot.com/2011/1 ... -luke.html

slam
Auditioning Since 1985
Posts: 256
Joined: Sat Jul 02, 2011 10:05 pm

Re: Scoring / Analysis Tools for Practice

Post by slam »

I'm a bit confused, DoT. How did you determine how many clues out of the clues revealed that a particular contestant knew? I agree that's a good way to measure "buzzer proficiency" (though ideally we'd want to see the percentage that was recognized by Alex out of the times that the contestant tried to buzz in - whether they knew the correct response or not). However, I don't see how that's knowable by us without an immediate post-game debriefing of a contestant.

User avatar
Woof
Swimming in the Jeopardy! Pool
Posts: 4212
Joined: Fri Jul 01, 2011 9:53 pm

Re: Scoring / Analysis Tools for Practice

Post by Woof »

slam wrote:I'm a bit confused, DoT. How did you determine how many clues out of the clues revealed that a particular contestant knew? I agree that's a good way to measure "buzzer proficiency" (though ideally we'd want to see the percentage that was recognized by Alex out of the times that the contestant tried to buzz in - whether they knew the correct response or not). However, I don't see how that's knowable by us without an immediate post-game debriefing of a contestant.
If I read correctly, he's using their FJ get rate to compute their %age, which is why he was only looking at people who had answered a statistically significant number of FJs.

User avatar
jpahk
Jeopardy! TOCer
Posts: 615
Joined: Sun Aug 28, 2011 1:16 am

Re: Scoring / Analysis Tools for Practice

Post by jpahk »

Woof wrote:If I read correctly, he's using their FJ get rate to compute their %age, which is why he was only looking at people who had answered a statistically significant number of FJs.
you're not reading it correctly. he's using the TS rate and some statistical guesswork, which is a lot better because only one player in the history of the show has a statistically significant number of FJs.

the basic idea: if players A and B both answer 20 clues per game, but player A's games have more triple-stumpers than player B's, we can reasonably conclude that player B knows more but player A is faster on the buzzer. there is one clearly false assumption involved* (namely, that the clam probabilities for the three players on the same clue are independent of each other) but nevertheless it is a useful calculation.

*edit: upon a closer reading, it looks like DoT did try to compensate for these by using average correlation rates for FJ. that is pretty clever actually, although of course FJ is a different animal. but it is in some sense the only thing there is to go on, since no other clue tells you what all three players are doing.

User avatar
Woof
Swimming in the Jeopardy! Pool
Posts: 4212
Joined: Fri Jul 01, 2011 9:53 pm

Re: Scoring / Analysis Tools for Practice

Post by Woof »

jpahk wrote:
Woof wrote:If I read correctly, he's using their FJ get rate to compute their %age, which is why he was only looking at people who had answered a statistically significant number of FJs.
you're not reading it correctly. he's using the TS rate and some statistical guesswork, which is a lot better because only one player in the history of the show has a statistically significant number of FJs.

the basic idea: if players A and B both answer 20 clues per game, but player A's games have more triple-stumpers than player B's, we can reasonably conclude that player B knows more but player A is faster on the buzzer. there is one clearly false assumption involved* (namely, that the clam probabilities for the three players on the same clue are independent of each other) but nevertheless it is a useful calculation.

*edit: upon a closer reading, it looks like DoT did try to compensate for these by using average correlation rates for FJ. that is pretty clever actually, although of course FJ is a different animal. but it is in some sense the only thing there is to go on, since no other clue tells you what all three players are doing.
Yes, he's using triple stumper cube hypothesis, which he then adjusts using statistics derived from FJs.

User avatar
RCraig
Jeopardy! TOCer
Posts: 111
Joined: Sat Jul 02, 2011 9:33 am
Contact:

Re: Scoring / Analysis Tools for Practice

Post by RCraig »

koam wrote: Real pie in the sky would be if something could keep such stats by generalized categories (geography, history, humanities, sports, pop culture, etc.)..but that's just a dream. I know TOC Champ Roger Craig had built his own computer programs for practice tools that allowed him to analyze his performance by subject area. Since that's out of my league, I'm wondering what tools are out there.
Others have tracked themselves as well, e.g. Volante's 0/1/2 strings.
I can't promise anything, but I'd be willing to give you (and anyone else) a breakdown if you send me your Right/Wrong info. (PM me for details.) Warning: The most time intensive part will be the data massaging into a nice machine readable form and I'm not going to have time to do that. Also, I may not get to it for weeks or even months, just don't want anybody getting the idea this will be instantaneous or anything...

User avatar
RCraig
Jeopardy! TOCer
Posts: 111
Joined: Sat Jul 02, 2011 9:33 am
Contact:

Re: Scoring / Analysis Tools for Practice

Post by RCraig »

My bad, I should be more clear about what I would need to do a breakdown for someone.

I would need your Right/Wrong (I count clams as Wrong) stats on a per question basis. Typically, this will be the 61 (Let's hope) questions for an entire episode. (It need not be for entire episodes though, but could instead be for some subset like "just FJs", "just DDs", "just bottom row questions" , etc.).

The important thing is to be able to map to the actual questions that appeared on the show. One way (but not the only or easiest way) would be to have a spreadsheet like the following:

Code: Select all

Round     Category     Depth     RightOrWrong
J            1           1              1
....<snip>
DJ           6           5              0
which would mean you got the top left clue of the J! round right, and the bottom right clue of the DJ round wrong. You don't have to use this system or something even remotely similar, just so it's regular and you know how to map it to the game board. My system identifies every clue by the Airdate/Episode number, Category (numbered 1 to 13), and Depth (1 to 5). (With tiebreaker questions after FJ being 13-2, etc.)

User avatar
Le Master
Loyal Jeopardista
Posts: 111
Joined: Sat Jul 16, 2011 2:00 am

Re: Scoring / Analysis Tools for Practice

Post by Le Master »

I demand commented source code. :P

User avatar
Volante
Harbinger of the Doomed Lemur
Posts: 7086
Joined: Thu Jul 14, 2011 11:42 pm
Contact:

Re: Scoring / Analysis Tools for Practice

Post by Volante »

RCraig wrote:My bad, I should be more clear about what I would need to do a breakdown for someone.

I would need your Right/Wrong (I count clams as Wrong) stats on a per question basis. Typically, this will be the 61 (Let's hope) questions for an entire episode. (It need not be for entire episodes though, but could instead be for some subset like "just FJs", "just DDs", "just bottom row questions" , etc.).

The important thing is to be able to map to the actual questions that appeared on the show. One way (but not the only or easiest way) would be to have a spreadsheet like the following:

Code: Select all

Round     Category     Depth     RightOrWrong
J            1           1              1
....<snip>
DJ           6           5              0
which would mean you got the top left clue of the J! round right, and the bottom right clue of the DJ round wrong. You don't have to use this system or something even remotely similar, just so it's regular and you know how to map it to the game board. My system identifies every clue by the Airdate/Episode number, Category (numbered 1 to 13), and Depth (1 to 5). (With tiebreaker questions after FJ being 13-2, etc.)
I'm wondering, would my series be effective for this?

(Sorry Robert...but I think people might find this useful... http://www.sinecreations.com/coryat_beta.php
Disclaimer: Yeah, I totally stole his HTML to make this )

User avatar
RCraig
Jeopardy! TOCer
Posts: 111
Joined: Sat Jul 02, 2011 9:33 am
Contact:

Re: Scoring / Analysis Tools for Practice

Post by RCraig »

Le Master wrote:I demand commented source code. :P
Comments? We don't need no stinkin' comments!

Code: Select all

//comments come after double slashes
Round     Category     Depth     RightOrWrong   //tab separated
J            1           1              1       //tab separated
....<snip>                                      //represents 58 (hopefully) snipped missing clues
DJ          6           5              0        //onomaTopoeia
FJ          1           1              1        // FJ was not a TS

User avatar
RCraig
Jeopardy! TOCer
Posts: 111
Joined: Sat Jul 02, 2011 9:33 am
Contact:

Re: Scoring / Analysis Tools for Practice

Post by RCraig »

Volante wrote:
RCraig wrote:My bad, I should be more clear about what I would need to do a breakdown for someone.

I would need your Right/Wrong (I count clams as Wrong) stats on a per question basis. Typically, this will be the 61 (Let's hope) questions for an entire episode. (It need not be for entire episodes though, but could instead be for some subset like "just FJs", "just DDs", "just bottom row questions" , etc.).

The important thing is to be able to map to the actual questions that appeared on the show. One way (but not the only or easiest way) would be to have a spreadsheet like the following:

Code: Select all

Round     Category     Depth     RightOrWrong
J            1           1              1
....<snip>
DJ           6           5              0
which would mean you got the top left clue of the J! round right, and the bottom right clue of the DJ round wrong. You don't have to use this system or something even remotely similar, just so it's regular and you know how to map it to the game board. My system identifies every clue by the Airdate/Episode number, Category (numbered 1 to 13), and Depth (1 to 5). (With tiebreaker questions after FJ being 13-2, etc.)
I'm wondering, would my series be effective for this?

(Sorry Robert...but I think people might find this useful... http://www.sinecreations.com/coryat_beta.php
Disclaimer: Yeah, I totally stole his HTML to make this )
Yes, your series is one of the ways I had in mind. I'm assuming it is a listing of a raster type scan across the 2 boards with FJ at the end.

User avatar
Volante
Harbinger of the Doomed Lemur
Posts: 7086
Joined: Thu Jul 14, 2011 11:42 pm
Contact:

Re: Scoring / Analysis Tools for Practice

Post by Volante »

RCraig wrote:
Volante wrote:
RCraig wrote:My bad, I should be more clear about what I would need to do a breakdown for someone.

I would need your Right/Wrong (I count clams as Wrong) stats on a per question basis. Typically, this will be the 61 (Let's hope) questions for an entire episode. (It need not be for entire episodes though, but could instead be for some subset like "just FJs", "just DDs", "just bottom row questions" , etc.).

The important thing is to be able to map to the actual questions that appeared on the show. One way (but not the only or easiest way) would be to have a spreadsheet like the following:

Code: Select all

Round     Category     Depth     RightOrWrong
J            1           1              1
....<snip>
DJ           6           5              0
which would mean you got the top left clue of the J! round right, and the bottom right clue of the DJ round wrong. You don't have to use this system or something even remotely similar, just so it's regular and you know how to map it to the game board. My system identifies every clue by the Airdate/Episode number, Category (numbered 1 to 13), and Depth (1 to 5). (With tiebreaker questions after FJ being 13-2, etc.)
I'm wondering, would my series be effective for this?

(Sorry Robert...but I think people might find this useful... http://www.sinecreations.com/coryat_beta.php
Disclaimer: Yeah, I totally stole his HTML to make this )
Yes, your series is one of the ways I had in mind. I'm assuming it is a listing of a raster type scan across the 2 boards with FJ at the end.
FJ's at the start then J 200s left to right, 400s, 600s... etc. (The array math was just easier having J! start at '1' and DJ at '31', so I put FJ at '0')

User avatar
the_phil
60% or Bust
Posts: 197
Joined: Tue May 15, 2012 6:08 pm
Location: Raleigh, NC

Re: Scoring / Analysis Tools for Practice

Post by the_phil »

Volante wrote:
RCraig wrote:
Volante wrote:(Sorry Robert...but I think people might find this useful... http://www.sinecreations.com/coryat_beta.php
Disclaimer: Yeah, I totally stole his HTML to make this )
Yes, your series is one of the ways I had in mind. I'm assuming it is a listing of a raster type scan across the 2 boards with FJ at the end.
FJ's at the start then J 200s left to right, 400s, 600s... etc. (The array math was just easier having J! start at '1' and DJ at '31', so I put FJ at '0')
As I mentioned earlier in the thread, here's a couple screenshots of my spreadsheet. From the look of it, it's essentially an excel version of Volante's .php way. I like mine for speed (I use it primarily for "playing" an archived game), the ability to save large numbers of played games, (and for a VBA learning experience).

template gamecard - http://imgur.com/cgvIX
populated gamecard - http://imgur.com/cu19E
empty calculators - http://imgur.com/dCQ1e
populated calculators - http://imgur.com/CeU9J

Note* - before you download, be aware that this file contains macros. You might get a security notification. I promise there's no malicious payloads.
download link - http://www.filedropper.com/scoresheet_1

Once I finish up a different J! related VBA project, I'll probably set to work on building a stats aggregation script, to compile stats across multiple scoresheets. For now you can just keep adding gamecards within this sheet, you'll just have to extend a couple columns on the 'Calculators' sheet.

User avatar
Volante
Harbinger of the Doomed Lemur
Posts: 7086
Joined: Thu Jul 14, 2011 11:42 pm
Contact:

Re: Scoring / Analysis Tools for Practice

Post by Volante »

Volante wrote: I'm wondering, would my series be effective for this?

(Sorry Robert...but I think people might find this useful... http://www.sinecreations.com/coryat_beta.php
Disclaimer: Yeah, I totally stole his HTML to make this )
To clarify how I use this, the page creates the string at the bottom and I copy/paste it into a text file (both a google docs and a txt file on my desktop) after appending the date and score.

There's no built in database for the page itself.

val_legal7
Just Starting Out on JBoard
Posts: 3
Joined: Wed Jan 15, 2014 12:19 am

Re: Scoring / Analysis Tools for Practice

Post by val_legal7 »

the_phil wrote:
As I mentioned earlier in the thread, here's a couple screenshots of my spreadsheet. From the look of it, it's essentially an excel version of Volante's .php way. I like mine for speed (I use it primarily for "playing" an archived game), the ability to save large numbers of played games, (and for a VBA learning experience).

template gamecard - http://imgur.com/cgvIX
populated gamecard - http://imgur.com/cu19E
empty calculators - http://imgur.com/dCQ1e
populated calculators - http://imgur.com/CeU9J

Note* - before you download, be aware that this file contains macros. You might get a security notification. I promise there's no malicious payloads.
download link - http://www.filedropper.com/scoresheet_1

Once I finish up a different J! related VBA project, I'll probably set to work on building a stats aggregation script, to compile stats across multiple scoresheets. For now you can just keep adding gamecards within this sheet, you'll just have to extend a couple columns on the 'Calculators' sheet.



Hey "the_Phil" - any chance you could re-share that Excel Coryat scoresheet? It is exactly what I've been looking for (and I am horrid with Excel on my own).

Post Reply