As of recently, I’ve been involved at all possible degrees with the new process of the Day 2 Deck Check Procedure: Creator, Head Judge, Team Leader, Floor Judge. Having experienced the process from each of these positions taught me an incredible amount of things I’d like to share today.
Warning: This article is only quickly mentioning the global concepts about the Constructed Grand Prix and Pro Tour Deck Check Procedure. Give it a look or you may fail to understand some of the things described further.
Recognizing outstanding contributors
But first, I’d like to thank a few individuals:
- Joe Klopchic ran the team in Vancouver, where I was the Head Judge. It was my first opportunity to check out the practical applications and discuss some of the challenges associated with the position. Joe crafted a piece of software to get a better view on which players were more important to Deck Check.
- Florian Horn was my team Leader in Utrecht. He crafted his version of the file which can be used either online or with excel that’s inspired from Joe’s work. Throughout the day and the debrief, he gave me a great insight as to which problems he met. The file linked later in his article is mostly his design.
- Aruna Prem Bianzino was the Deck Check Team Leader in Barcelona. He sent me a comprehensive document featuring a data analysis of the tournament, which gave me a good insight to identify which goals should be met at which moment.
- QJ Wong was one of my team members in Shizuoka. He quickly took ownership of some parts of the process, which allowed me to focus on the process itself and how to rationalize it optimally so it becomes easier for everybody to use.
And the numerous GP Head Judges who proofread that article and gave me tips on how to make it easier to understand!
Analyzing the tournament’s behavior
Predict Final results based on Day 1 results
Here are some numbers from various recent Grand Prix:
|Day 1 Players||2700||1631||1265||1228|
Note#1: X represents any combination of wins and draws. For instance, 7-X means 7-2 or 7-1-1 or 7-0-2
Note#2: I consider draws as losses since it removes each player from a perfect record and it needs 3 draws to be equivalent to one win, which is incredibly rare.
Note#3: With 370 players less than New Jersey on Day 1, Barcelona’s Day 2 was only 35 players smaller. This is explained by the huge percentage of players with byes starting Day 1.
I computed these numbers into my Results prediction sheet and it gave me the following prediction for Final Standings (still excluding draws):
|Day 2 Players||858||500||465||398|
Note#1: 0,16 implies that there is a 16% chance that one player ends up 15-0 if you consider that no result is a draw and all games are 50/50
Note#2: This method of calculation is only an estimate and is by no way 100% accurate: First, it doesn’t take draws into account. Then, it doesn’t consider paired up/paired down pairings.
Focus does not need to be put on any player with a worse-than-X-2 record in Shizuoka. Same in New Jersey.
In Barcelona, the sum of the probable results is slightly above 8, which means that excluding any draws (intentional or natural), all Top8 players should be X-2 or better. However, considering a Draw acts as a double loss in the reasoning, it’s reasonable to believe that there will be players with less than 39 points making top8 (mostly due to Intentional Draws).
In Utrecht, the sum doesn’t even reach 7, which could instinctively means that there will be X-3 players making top8. However, it turns out that this mostly makes room for players to Intentional Draw into the Top8.
If the sum of the 13-2 and better players exceeds 8, then players need 39 points to make Top8.
If the sum of the 13-2 and better players is below 8, it’s likely players can draw into the Top8.
Note: The available data doesn’t allow to determine the number of Day 2 players needed to draw the line between the two aforementioned cases. Also, that’s dependent on the number of byes, draws and drops on Day 1.
Which players are making Top8?
Aruna wrote a statistical analysis of Barcelona. The most interesting takes were to me:
- Ignoring natural draws, 30% of the undefeated players after day 1 will make top8.
- Ignoring natural draws, 30% of the X-1 players after R11 will make top8. This possibly overlaps with 1-
- It’s only from the beginning of round 13 that players with 2 losses start having a non-negligible chance to make top8.
- It is possible that players with an X-1 record or better (depending on the size of the event) go for ID-ID in R14 and 15, hence they should be deck checked by R13.
The theoretical plan
Based on these factors, I decided to craft the following theoretical plan for Shizuoka:
- Gather information about which players were deck checked on Day1
- Request to perform DC in the 8-0 range in Round 9.
- Having all undefeated players DC by the end of R10
- Having all X-1 players DC by the end of R12
- Deck Check the remaining X-2 players by the end of R14
- Use R15 to spot the last or last two players in contention, if needed.
In a smaller tournament, you may be able to achieve having deck checked all X-1 players by the end of R11.
Note: In this system, Deck Checks performed during Day 1 aren’t taken into account (except those performed in R9). Indeed, while it is desired that Day 1 deck checks are tracked, this system is designed to work as if this didn’t happen. If that happened, that’s of course even better
Keeping an overview on the process
Note: This process can be run without the Deck Check tracker Tool. The tool only helps you, it doesn’t automate the entire process. It aggregates information in a way that making decisions is easier for you.
So as to have a better view on what’s going on, Joe Klopchic and Florian Horn designed their own versions of a tool tracking deck checks. I have taken a significant part of it and worked on providing instructions on how to use. You can find the Deck Checks Tracker here.
This tool allows to check in real time which players have been deck checked, how many times they were and what is their decklist number for easier lookup.
It requires the SK to provide Pairings for each round and the most recent Standings to fully operate. Note that if you operate the Tracker online, the SK can directly copy/paste at the appropriate place.
This tool also helps locating the player’s decklist based on the principles described here. Note that if decklists have not been indexed, the tool won’t work. But indexing has become the default for a few years now.
Pairings by table
Considering the SK will not always be able to fill in Standings and Pairings early enough for you to be able to determine the Beginning of Round Deck Check tables, I recommend you request the first page of Pairings by Table to be printed first (before Pairings by name). It’s definitely better if they can, but thanks to the printed copy, you can manually select tables, which works just fine.
Choosing the correct tables
Beginning of Round Deck Checks
Using the printed Pairings by table, select the first 2 or 3 (depending on your team’s size) tables in the category of the current target, choosing if possible tables where no player has been deck checked.
For instance, if you’re still targeting, X-1 players, do not deck check a table with two X-2 players, even if this is the first one where no player has been deck checked.
Mid-Round Deck Checks
Give your judges a list of several tables relevant for a deck check. Give them enough latitude that they should not need to wait for long before a table is ready for game 2.
For instance, if you could perform 2 Mid-Round Deck checks, give your team a list of 4-6 tables.
If time permits, perform more than one series of Mid-Round Deck Checks.
Especially when the format is fast, this shouldn’t impact the tournament’s timing.
Pre-select next round’s Deck Checks
Because it’s not always possible to have the standings for the round early enough and since Pairings by table do not display Tiebreakers, take a look at the latest Standings you have (for instance Standings after R11 when preparing for R13) to pre-select the top ranked players not yet deck checked for a Deck Check in the next round. This way, you will be able to use Pairings by table even more efficiently.
The importance of Coverage
Deck Checking Feature Matches
It is possible to deck check Feature Matches. The only exception is the video table.
Coordinate with Coverage ahead on time to know which table is the “live” table. It is also possible, if you need to deck check a very specific table, to request from Coverage they do not put it on video. They usually agree on the request (either in the current round or the next one).
Anticipate “to-be-featured” players
Check with Coverage which players or decks they feel like featuring soon.
If at any given point you have to choose between a table featuring one of these players and a table without, check the one that’s more likely to be on-camera later.
Again, we don’t want to deck check on-camera tables.
Goals by round
Round 10 Target: Deck Check all undefeated players
It’s likely that undefeated players after Day 1 will be called to Feature Matches in Round 10 and that one table will be on video, which can delay the completion of the goal. These players will need to be DC as soon as possible.
This should be anticipated by asking the Day 1 DC Team leader to perform as many deck checks as they can amongst the undefeated players in Round 9.
On top of deck checking undefeated players, this also checks the decks of many X-1 players.
Round 11: Focus on X-1 players
While at that point 50% of the 10-0 players will statistically make top8, only 20% of 9-1 players will.
If some X-0 players haven’t been deck checked yet, focus your beginning of rounds efforts on tables where no player has been deck checked yet and make the remaining X-0 players a priority for mid-round deck checks.
Keep in mind Lunch breaks will start.
Especially true if the Breaks Plan is very aggressive, for instance if two teams going on break in R12 and the other two in R13, this will limit your ability to achieve your goal of having Deck Checked all X-1 players.
Run a full round of deck checks in R11 to avoid the impact of Lunch Breaks.
Do your best to run two series of mid-round deck checks.
Round 12: Lunch break
Considering at the end of round 11, statistically 5/8th of the top8 players have been deck checked and you’re now in the dark as to who will be the other three, R12 is the greatest moment to go for a break.
If needed, perform a series of BoR DC to finish with deck checking the X-1s and leave.
Round 13: Support Floor Coverage
You may need to help cover the Floor during that round.
Perform the BoR DC before going back to the Floor. Keep in mind to coordinate with Coverage to see which players or decks are more likely to be featured and on-video in the later rounds.
Round 14: Finalize X-2 players
With every judge back on the Floor, it’s time to run a full round of deck checks to be one step ahead before the last round.
Do your best to run two series of mid-round deck checks.
Round 15: Finish the job, then Deck Check the bubble
To prevent Bribery issues, the pairings for the last swiss round of a tournament are ordered by Standings and not random inside the same amount of points. This means you can easily predict which tables are in contention for Top8 since Tiebreakers are rarely massively modified.
Perform some security deck checks in the tables that should place players as 9th and 10th. Indeed, if one of the top tables ends up unintentionally drawing, it’s as if two players had lost and therefore, the foreseen 9th may finish 8th.
Once this is done, focus on tables playing for Top64.
Some Practical Results
|Day 2 size||858||500||426||398|
|DC Team size||7||5||4||4|
|D1 R9 DC||6||2||0||0|
|Total D2 Top8 DC||29||22||14||18|
|Other D2 DC||6||unavailable||0||1|
|Top8/Top32 players DC||8/26||8/30||7/unavailable||8/24|
These results show that, to achieve this goal, less than 1 DC/judge per round is needed. Therefore, you do not need to deck check a lot more than usually: In a classic team of 4, the usual 2 BoR and 2 MR DC per round scheme sums up to 4 DC per round, which is about 1/judge.
It therefore comes down to coordination, since there is usually one round where the team is on a break. To account for this, DC need to be reported to other rounds, the most efficient being R11 and R14, as shown earlier in this document.