Pinning the third wave: who crosses the map first, and how

There is a moment in the game when the map gets stretched: your wave is sitting at their T3 in one lane, and their wave is sitting at your T2 or T3 in the opposite lane. When two of your waves are pushing and only the third one comes against you, Astini calls this pinning the third wave. Someone has to cross the map. Does whoever crosses first come out winning or losing? And do they cross by TP, smoked or walking?

The tower study answered what to do after taking the T2. This one answers what to do with the stretched map, on the same data: the 240 maps from The International 2026 and the Esports World Cup 2026, with the trail of the ten heroes at one point per second.

How the situation was found

Every five seconds of every game, the front of each lane comes from the last lane creep killed by any hero (it is the same gauge as the smoke study). The situation opens when one lane is past their T2, on the way to the T3, and another lane is past our T2, and it counts once every three minutes. That gave 953 situations from minute 10 on, 279 of them with two lanes pushed.

Two names the tables use all the time: X is the team whose wave is at the other's T3; Y is the team with the threatened T3 (and whose wave comes against X's tower). And crossing is having three or more living heroes in one of the two zones (Y's T3 or X's threatened tower) where the team had at most one when the situation opened, within the next two minutes. The method is what the trail shows: TP when two or more jumped there, smoke when the team used Smoke of Deceit between 30 seconds before and the arrival, walking for the rest.

Crossing first decides nothing

X crossed first in 321 situations and won 58% of them. Y crossed first in 332, and X won 59%. In 300 nobody crossed within two minutes, and X won 62%. The median time until the first arrival is 45 seconds, and the team that arrives within 30 seconds does not win more than the one that arrives after a minute. The advantage of whoever crosses first does not exist. What exists is the advantage of whoever crosses to the right place.

The wave owner: follow it, do not go back

When X crossed first, it went to one of two places. It followed the wave to their T3 in 112 cases: 73% wins, 2,133 more gold in two minutes, 187 kills for 86 deaths, 93 towers taken for 29 lost. It went back to defend its own tower in 209 cases: 49%, 299 less gold, 206 kills for 221 deaths, 66 towers taken for 83 lost.

The objection is the usual one: whoever goes forward is whoever is winning. That is why the comparison below is within the same range of score.

With the gold even, X following to the T3 won 57% (19 kills for 6 deaths, 12 towers for 4) and X going back won 40% (49 for 63, 9 towers for 21). Two to six thousand ahead: 81% against 66%. Six thousand or more: 88% against 71%. Behind: 35% against 27%. In every range, going back to defend the tower was X's worst decision, and in every range the tower it went back to save fell more than when it kept going. Going back does not save the tower: it trades a T3 of theirs that was 45 seconds from falling for a fight on our half of the map, where the enemy arrives first.

The clip is the typical case: X's wave is at their T3, X abandons the wave to go back to its own tower, arrives walking, and loses the trade without taking anything.

The defender: do not run to the T3

The mirror applies to Y. When Y crossed first and went to defend the threatened T3, it won 30% of the games, with 190 kills for 265 deaths and 54 towers taken for 133 lost. When Y crossed first and went to push X's tower in the opposite lane, it won 61%, with 184 kills for 107 deaths and 68 towers for 20.

Again by score: with the gold even, Y defending the T3 won 30% and Y pushing X's tower won 36%; with Y six thousand ahead, 64% against 81%; with Y two to six thousand behind, 26% against 60%. Their T3 is 45 seconds from falling and the answer that pays is the trade, not the run. The team that runs to the T3 arrives spread out, on top of a wave that has already landed, against a group that was waiting for exactly that.

TP or smoke?

Without discounting the score, the TP defense looks like the disaster of the study: Y crossing first by TP won 38% and lost 709 gold in two minutes; smoked, 51% and plus 588. But whoever TPs to defend is whoever is losing the wave, so the table below separates by score.

Within each range, the TP to defend is not worse than arriving on foot: with the gold even, 45% against 28%; with X two to six thousand ahead, 38% against 22%. The problem was never the scroll. The problem is defending the T3. Whoever TPs and lands on top of a group of five loses the same way as whoever arrives walking, only sooner.

For the wave owner the question is different, because nobody reaches the enemy T3 by TP (zero cases with the score unsettled). X arrives walking or smoked:

With the gold even, X smoked to the T3 won 73% (16 kills for 4) and X walking won 60%. Two to six thousand ahead, 89% against 75%. With six thousand or more, the smoke stops mattering (81% against 95%): whoever has already won does not need to hide the crossing. The smoke is the tool of the stretched map with the score even, and it is exactly what the smoke study already said: a smoke with an objective in the window converts, a smoke without one blows up.

The duel at the T3

Who arrives first at the threatened T3? When it is X, and Y does not even show up (91 cases), X wins 74% with 160 kills for 77 deaths. When X arrives first and Y responds by TP, X wins 67%; when Y responds walking or smoked, 78%. And when Y arrived at the T3 first and X went anyway, X won 79% with 55 kills for 31 deaths. Their T3, with our wave sitting on it, is the best place on the map to be, whoever gets there first.

The two clips are the TP defense in both outcomes: in the first, Y arrives by TP at the T3 before X moves and punishes X for not going with the wave; in the second, Y TPs too early, lands on top of X's group and pays.

Pinning the third wave

With one lane pushed (674 situations), X won 51%. With two lanes pushed and the third coming against (279), X won 79% and was 1,239 gold ahead after two minutes. It is the state the doctrine calls pinning the third wave, and the number says why it exists: with two waves at their base, the enemy has two towers to defend and one to push, and the trade always favors whoever has the two. Within that state, X crossing first won 75% with 1,685 gold; Y crossing first, X won 75% as well, with 559. When the two waves are pinned, the enemy's crossing does not change who wins: only by how much.

The honest reading is due: two lanes pushed is also a symptom of who is already winning, and the table does not separate the cause from the symptom. What it shows is that, in that state, the team ahead does not lose the trade because Y crossed first.

Nobody crossing is also an answer

In 300 of the 953 situations (31%) nobody had three heroes in one of the two zones within two minutes, and X won 62% of those. With six thousand or more ahead, nobody crossing yielded 94% wins and 2,170 gold, more than any crossing. Both waves hit towers, each team farms what is on its side, and the score moves for whoever is ahead. Crossing is a decision, and not crossing is one too.

The TP of a single hero to defend his own tower, with the rest of the team on the wave, is the most common middle ground, and the study neither condemns nor recommends it: 60% wins with or without it. What it costs is gold (minus 141 against plus 494 when nobody goes back), because the hero who went back stopped farming the pushed wave to defend a tower that almost always falls anyway.

What to take to ranked

1. Stretched map? Look at where YOUR wave is, not where theirs is. If your wave is at their T3, your team goes there. Going back to defend wins 49%; following wins 73%, and within each score range the gap holds.
2. If the threatened T3 is yours, do not run to it. Defending the T3 wins 30%. Pushing their tower in the opposite lane wins 61%. The tower is worth less than the trade.
3. The TP is not the villain. Discounting the score, TPing to defend is not worse than arriving on foot. What loses is defending. If you are going to TP, TP to trade objectives, not to land on top of five.
4. With the score even, the wave owner's crossing is smoked. 73% against 60% walking. Six thousand ahead, the smoke stops mattering.
5. Arriving first is not an advantage. 58% when X arrives before, 59% when Y arrives before, 62% when nobody arrives. The destination decides.
6. Two pinned waves is the state you want. 79% wins against 51% with one lane, and the enemy's crossing does not take that from you.

Method and caveats

The wave front is only known when someone kills lane creeps; a wave hitting a tower alone is invisible to the gauge, which is why the window is 75 seconds and the situation requires both lanes to be known. The crossing is read from the trail (three heroes in the zone), so a team that arrives with two does not count. The TP is inferred from the position jump and lumps portal and scroll together. The gold of the window is the team's gold minus the enemy's, without what the dead drop, to match the OpenDota chart.

The usual caveat: nothing here is an experiment. Teams do not decide at random, and the tables by score range reduce the bias without eliminating it. What they show is what each decision yielded when the best teams in the world made it.

The sister guides are After the tower: force the T3 or go hunting?, Dealing with waves and When to use the TP scroll.