Looping in Tetris

Looping is a type of playing forever which involves reaching the same position as in the beginning of the game.
That means performing a Perfect Clear at the point in bag structure where it simultaneously leads into Bag 1.

Recommended reading

If you're not familiar with 7-bag Randomiser, Hold key and modern Tetris rules, I recommend reading PC bibble ↗️.

Bag Structure

bag-structure.png
Fig. 1: Bag structure.

Notice how bag structure repeats after 35 pieces.

Loop types

Full Loop

Full Loop clears 28 lines using 70 pieces, clearing 280 squares.

full-loop.png
Fig. 2: Full Loop (7*4L=28L = 70p)

Full Loop is only used in Pure PC.
Notice how an uninterrupted series of 4-line Perfect Clears misses the boundary of half-length loop and forces tackling the same bag structure in different scenarios relative to individual Perfect Clears.
Pure PC Loop has good theoretical efficiency but it's very difficult to perform consistently and (without extra T-spins and Tetrises) falls short of best DPC loops in terms of average score.

💡 Taking a 2L switches you to the other row of the above diagram than you would end up if you did a 4L.

Half-length loops

Half-length loops clear 14 lines using 35 pieces, clearing 140 squares. Outside the context of PC Mode, "loop" typically refers to half-length loop.

DPC Loop

dpc-loop.png
Fig. 3: DPC Loop (8L+6L = 35p)

DPC loops involve performing an 8-line Perfect Clear setup (or two 4-line Perfect Clears) followed by a 6-line setup.
DPC loops are popular due to their relative simplicity and consistently high average score. Guaranteed T-spins in DPC compensate for the lower PC rate compared to Pure PC.

6L Loop

6l-loop.png
Fig. 4: 6L Loop (6L+8L = 35p)

6L loops involves performing a 6-line Perfect Clear followed by two 4-line Perfect Clears (or an 8-line Perfect Clear setup).
6L loops are impractical due to the difficulty of 6L opener setups. However, there is a good potential for very high average score if a good 6L opener is found. As of writing this article (October 2022), this area is underresearched.

14L Loop

14l-loop.png
Fig. 5: 14L Loop (14L = 35p)

14L loops are 14-line Perfect Clear setups that loop unto themselves.
Basically BT Cannon loop. 14L loops are generally bad because T-spins and Tetrises cannot compensate for the lost Perfect Clears.

How to loop

In practice, loop setups will often feature segments where continuation is not guaranteed. This can be somewhat alleviated by using recovery setups when the main goal of the segment (typically a Perfect Clear or a T-spin) is impossible to perform with current queue. I strongly discourage common practice of learning an incomplete setup and hitting the reset button whenever you don't get RNG that allows you to execute the single route you know.

My recommended method to loop is to start by first learning a simple minimal loop such as Ultimate simplified Jigsaw loop ↗️ which you can always build. Then gradually increase your efficiency by learning extra setups. You can always fall back to your minimal loop when you can't build any better setup with current queue.

Changelog