Tetris Openers: TKI-3, DT Cannon and Albatross

The single biggest mistake in competitive Tetris openings is picking your opener before you have looked at your queue. You decided last week that you are a DT Cannon player, the game hands you an S, Z, T start, and now you are twelve pieces deep in a build that was never going to work.

Openers are not a personality. They are a lookup table keyed on your first bag, the table is small — roughly eight setups cover almost every legal opening — and the entries sort themselves by which of I, L, J, or O shows up early. Everything below applies to modern Guideline Tetris: TETR.IO, Jstris, Tetris Effect: Connected, Puyo Puyo Tetris, and Tetris 99. It does not apply to NES Tetris or the Tetris The Grand Master series, and the next section explains why not.

Why openers only exist in Guideline Tetris

Hard Drop's opener article opens by listing the rules an opener depends on, and that list is the whole argument. A named, memorised opening sequence needs all of the following:

• A bag randomizer. Every tetromino appears exactly once per seven pieces, so "I will get an O in this bag" is a fact, not a hope.
• A hold slot. Almost every T-spin opener parks a T or an I in hold at some point.
• SRS kicks. The T-spin triple in a DT Cannon or a C-Spin only exists because the Super Rotation System tests a kick that drops the T two rows down and one column across.
• Multiple previews. Guideline games show up to six upcoming pieces. Without that, you cannot verify the build is routable before you commit the first placement.
• A bonus for T-spins, combos, and perfect clears. Otherwise a flat Tetris well beats every build here.

NES Tetris has none of those five. TGM has a bag-free history randomizer, ARS instead of SRS, and no T-spin reward. That is not a technicality — it means any "best Tetris opener" advice you read is implicitly about Guideline versus play. For the mechanical background, the rotation systems and wall kicks guide covers what SRS actually tests and the randomizers guide covers why 7-bag makes any of this predictable.

Read your first bag before you commit

The useful way to organise openers is by which piece arrives early, because that is the constraint you cannot negotiate with. Here is the decision table.

Starting with S, Z or T is, in Hard Drop's own words, basically the worst thing that can happen if you want a T-spin opener. A T start forces you to either waste the T in the stack or hold it for a long time; an S or Z start forces a soft drop, a platform spin, or a downgrade to a T-spin single.

One coverage trick is worth memorising above all others. If your first piece is an O and you place it three columns from the left wall, either Hachispin or Albatross will always be executable — Hachispin wants J before L, Albatross wants L before J, and one of those is always true. One placement, guaranteed T-spin opener from any O start.

TKI-3: the I-piece opener with four faces

TKI-3 (Japanese 開幕TSD) was invented by Tetris DS player TKI, and confusingly the name means something narrower in Japanese than in English — in the Japanese community "TKI" refers to triple-double attacks generally, while the English scene uses TKI-3 for this specific opening family.

The base never changes: I flat in the middle, an S or Z horizontally on top of it, the O in the opposite corner. What changes is where the J goes, and that single choice names the variation.

The reason to learn Castle Top first is that it routes into an Imperial Cross with an early Z in the second bag, and that branch carries an 89.68% perfect clear rate — the highest of the documented TKI continuations. Flat Top's Yack's PC branch sits at 68.77%, climbing to 88.33% if you place the Z piece first, and its Mechanical TSD line is 71.67%. Dingle's 8-line PC example is 75.08%. If you get an early J alongside the I there is a dedicated Early J setup whose 8-line PC example runs 93.10%.

Those numbers are the actual reason strong players pick a variation mid-build rather than pre-committing. You are not choosing an aesthetic, you are choosing a probability.

DT Cannon: one wasted T, two big spins

Double Triple Cannon — DT砲 in Japanese, DT Cannon or just DT in English — is also a TKI design. It produces a T-spin double that feeds directly into a T-spin triple, using the SRS T-spin triple wall kick to drop the T through the gap.

There are two common bases and they build the identical stack:

• LS base. Fastest in keypresses. L, J, S and Z need no rotation at all — they either hard-drop immediately or sit against a wall. This is why DT is so popular despite being piece-inefficient.
• JZ base. Always available when your first piece is a J, because the Z can still be spun into place after the T and O have already been placed.

Two things to know before you commit. First, the first T piece is wasted in the stack — that is the cost of the setup, and Albatross and Hachispin exist specifically to avoid paying it. Second, the first T-spin is normally delayed as long as possible so both spins land back to back. In any variant without instant lock down, hold the soft drop button until the T-spin double completes, so the T falls through the gap instead of rotating early into the wrong slot.

Hard Drop makes a counter-intuitive point about DT's inefficiency: more pieces sitting in the stack is sometimes an advantage, because that stack transitions into a 4-wide or a Tetris well faster than a piece-efficient build does. Wasteful is not automatically wrong.

The standard three-bag continuation leaves a flat stack with a perfect clear opportunity after 10 cleared lines. If the PC does not solve, the documented fallback is a donation T-spin double.

There is also a DT STSD variant, identical in appearance except one extra cell is left empty at the bottom, converting the follow-up into an STSD for three T-spin doubles total and no leftover overhang. It is more fragile: a garbage row arriving with its hole in the wrong column breaks it outright.

Albatross and Hachispin: the O-piece pair

Starting with an O looks like a dead hand and is actually one of the best openings in the game. Both of these setups drop an S or Z on top of the O and place its counterpart beside it — Hachispin places the top piece vertically, Albatross places it horizontally.

Albatross Special is a single-bag opener that produces a platform (air) T-spin double, and its headline feature is that it wastes no T pieces, unlike DT Cannon or C-Spin which each burn one to build. It is always executable from an O start when L arrives before J. The orientation rule is trivially simple: if L comes first, prop the S on the left and spin left; if J comes first, prop the Z on the right and spin right.

Its follow-up is a T-spin triple, usually completed at the end of the second bag, then another platform T-spin double — with a documented 26.67% perfect clear rate on both the O-then-L and L-then-O routes, two solutions each. Albatross into Infinite TST is specifically called out as the strongest way to open a game of Ultra, because you attack with your first two T pieces instead of burying one.

Two real weaknesses. If you cannot see an O in your five previews, play something else — a late O means the S and Z occupy the space and the O no longer fits. And Albatross is garbage-sensitive: roughly a 10% chance an incoming garbage hole lands along the edge before you place your J or L, which nullifies the T-spin triple and the T-spin double after it. An S, Z, T or Z, S, T start kills it too; that is Pelican territory.

Hachispin is the mirror-dependency twin: always executable from an O start when J arrives before L, producing a T-spin single into a T-spin triple across the first two bags. Its third bag always allows a T-spin of some kind, and a T-spin double needs nothing more than an L piece.

MKO, Pelican and the bags nobody wants

MKO Stacking (MKO積み), named after its creator MKOMIZ, is the L-and-J opener that pivots toward a perfect clear rather than a spin. Both L and J must arrive early. You place them flat-side down one column apart, drop the S and Z vertically on top of them, and keep the T on hold. The payoff is a 33.3% chance of a perfect clear after ten pieces, which needs an early O in the second bag. That is a gamble, not a plan — which is why the STMB Cave and S-on-S continuations matter. The S-on-S branch, where you drop an S vertically onto an S (or Z onto Z), reaches an 88% perfect clear chance after 8 lines if the LJ or OI pair goes on the S. Those solves are hard to spot, because a piece usually gets split by an in-between line clear.

MKO's honest flaw: the first T-spin cannot be fully prepared inside the first bag. If the T arrives first in bag two you are holding two T pieces and must spend one on the overhang.

For the bags nobody wants, three salvage options:

• Pelican. Albatross-shaped, single bag, platform TSD into TST. From an S start it is always executable if the J arrives before both the L and the O, and its three-bag continuation adds a TSD plus a perfect clear chance after 10 cleared lines.
• C-Spin. A triple-double attack: T-spin triple first, then a T-spin double, across three bags. Built on an S, Z, T core with the L and J forming the overhang — the natural home for the exact start that kills Albatross.
• BT Cannon. DT-shaped — TSD then TST within three bags — but usable from a T start with early S and Z. It chains into a reversed C-Spin and another T-spin triple, with a perfect clear window after five bags.

Also worth knowing: Pokemino's STD has almost no first-bag dependency at all, just J before Z, which makes it nearly always stackable — and also one of the weakest options here, because its T-spin triple cannot be finished inside the second bag.

What each opener actually sends

Attack values are not universal, and quoting them without naming the game is how bad advice spreads. The table below is Hard Drop's garbage table for four specific clients. Note that TETR.IO and Tetris 99 use their own systems — Tetris 99 layers badge multipliers on top, which the Tetris 99 strategy guide covers separately.

Run the openers through the Puyo Puyo Tetris column and it clarifies fast. Two TKI T-spin doubles are 4 plus 5 with back-to-back, so 9 rows across two bags. DT Cannon is a T-spin double into a back-to-back T-spin triple: 4 plus 7, 11 rows. Albatross adds a third spin — TSD, b2b TST, b2b TSD — for 16 rows across roughly three bags, without burying a T. A Perfect Clear Opener sends 10 rows in ten pieces, which is why it competes despite failing outright on a meaningful share of bags; the perfect clear guide covers its 61.2% base rate and the Grace System fallback.

One more variable most players ignore: how messy the garbage you send actually is. In Tetris Friends the hole column never shifts during an attack, so garbage arrives clean and cheap to clear. In Puyo Puyo Tetris there is a 30% chance the hole column shifts between rows inside a single attack, so even T-spin and perfect clear garbage lands messy by luck. Games with garbage blocking — Tetris Friends and Tetris Online among them — hold garbage back until a combo breaks, which is exactly what makes side-stacked combo openers stronger there. The scoring and attack guide has the full points side of this.

Combo openers: 4-wide changes the format

Combo openers reject the T-spin premise entirely. You build a tall rectangular tower, leave a fixed number of columns empty, and clear one line per piece for as long as you can hold it together.

Four wide beats two wide for a mechanical reason worth understanding: a 2-wide stack shrinks by about 2 lines per dropped piece, while a 4-wide shrinks by only 1. More residual height per piece means a longer combo. A 4-wide tower also ends up slightly taller for the same pieces, so it builds faster. Three wide is the compromise — slightly worse than 4-wide, but you can run it without soft drops.

The cost is control. On a 6-wide tower you have far fewer legal placements than on an 8-wide, and in a 4-wide your pieces must clear exactly one line to keep the count alive, with very few exceptions. Three filled cells in the empty columns — the residuals — is the standard, and that number holds as long as every piece clears exactly one line.

Center 4-wide is worth knowing about even if you never build one. Instead of a 6-wide tower on one side, you build two 3-wide towers with the four empty columns down the middle. There are only two ways to top out in Guideline games: place a piece entirely above the visible field, or have a piece spawn into a filled cell. Center 4-wide sidesteps both, because you only die once garbage has pushed the bottom of the center channel to the very top. That makes it the defensive answer to T-spin and perfect clear openers, at the price of keeping two towers at matching heights.

Combos are also where an opener can be interrupted mid-run. A tower that breaks at combo six sends a fraction of what it promised, and a well-timed spike tops you out before you reach your own garbage. Highest ceiling, worst floor.

Quick Action Checklist

• Look at all five or six previews before your first placement, not after it.
• Learn one opener per first-piece category — I, L/J, O, and the S/Z/T salvage — rather than one opener total.
• Place an O start three columns from the left wall so either Hachispin or Albatross is always available.
• If you cannot see an O in previews, do not start an Albatross.
• Learn the DT Cannon LS base first, and hold soft drop through the T-spin double so the T falls through the gap.
• Know which client you are on before you trust an attack number — perfect clears send 10 in Puyo Puyo Tetris and 6 in Tetris Online.
• Practise center 4-wide as a defensive answer, not as your default opening.