No Terran player advanced to the semi-finals of the WCS Global Finals 2018. The biggest shock is Maru’s 0-3 defeat to sOs. This post discusses sOs’ answer to Terran’s proxy strategy.

Background

Proxy is the mainstream strategy in TvP for the last few months. It is so ubiquitous that two variations are discussed in the September and October entries of the Metagame Build Orders. The key points for the proxy strategy are:

• Terran do not want to play the “standard” macro opening versus macro opening against Protoss.
• The central tenet of the strategy is to create information gap in Protoss’ decision making. It is difficult for Protoss to deduce what exactly Terran are doing, so it is difficult to prepare well.
• There is a broad range of variations for proxy. Terran can fake that they are doing proxy by placing their buildings near their base. On the other extreme end of the spectrum, Terran can also execute an all-in with the proxy.
• Protoss have been experimenting different answers to this strategy.

You can read about the proxy strategy in-depth in this article.

sOs had showed some of tricks that he thinks are effective against proxy in these two months, and he showed how much he refined them in the series against Maru.

Stage 1: Probe scout

There are two main choices for Probe scout:

• Go straight to Terran’s main base.
• Scout the plausible locations.

Players typically send a worker to opponent’s base directly. When Protoss see no Barracks, they deduce that proxy is the likely opening. They deduce further by scouting the whole main base and areas behind the natural to ensure this is not a simple fake proxy. When Protoss still do not see the production buildings, they usually conclude that this is a “standard” proxy. The problem remains: what is Terran up to?

Consequently, Protoss adjust their scouting habits and send the Probe(s) to nearby plausible locations (see below vods for examples).

Though this seems like a reasonable adjustment, it has one weakness. What if Protoss do not find any proxy building? Does this mean that Terran did not proxy? The only logical conclusion is that Terran did not proxy at those scouted locations. In other words, Terran could have proxy in the unscouted location or they did not proxy at all. This is a high risk high reward move, because Protoss have the potential to delay or even force a cancel on the Barracks (see below vods for examples).

Protoss have now further refined the Probe scout by sending it straight to the Terran’s base after the Probe puts down the first Pylon. This is a relatively early worker scout, and it has two obvious scouting advantages. The first is that it gives you time to scout for the proxy at the plausible locations after you have confirmed there is a proxy. The second is to know whether the Factory is placed in Terran’s main. sOs uses this Probe scouting method.

What separate sOs from other Protoss players in this area is the things that he does with the Probe that is unrelated to scouting directly. sOs has two additional moves with that Probe. Once he saw that there is no Barracks at the ramp, the Probe goes straight to the geyser to build a Assimilator (see vods below). If Terran do not prevent him from doing so, it is a big bump on the plan to tech up with the proxy strategy.

The other move is to build a Pylon at the ramp to prevent Terran to wall off properly. The positioning of the Pylon needs to be precise, as you want to create gaps on both sides of the Pylon (see the first image below). Terran would have to either destroy the Pylon to wall off or build two buildings to block the gaps. This move allows Protoss units to attack the workers. sOs used this technique well in game 1 and game 2 against Maru (see the second and third images below). sOs had actually used this move before in GSL more than two months ago, but he misplaced his Pylon and allowed Terran to wall off (see the fourth image below).

Terran then have two choices. One is to bring the units back from the proxy to deal with the Protoss units, and this drastically attenuates the strength of proxy in the early game. The other is to proceed with the attack on Protoss’ side as planned. Logically, if Protoss find it hard to defend the attack when they have the Gateway units at home, then shouldn’t it be impossible to hold if Protoss had sent the units across the map? This leads to the next stage of sOs’ plan.

Stage 2: Expansion

To expand, or not to expand, that is the question. You can take a look at games played in the last GSL two months ago, and you quickly notice that Protoss’ plan is to expand and hold off whatever Terran are going to throw at them. Protoss then build Shield Batteries at the natural to defend against Terran’s attack. Contrary to this conventional approach, Protoss recently have started to opt to expand later.

sOs did just that. It is a lot easier to defend on top of the main ramp than at the natural. In game 1, when Maru decided to send the Cyclones to attack sOs instead of going back to Terran’s base to kill the Adept and Zealot, sOs had Stalkers with a Shield Battery ready on top of the main ramp. Maru immediately turned away (see image below).

Stage 3: Tech choice

Robotic Facility is the most common choice when Protoss want to expand and defend, because Immortal is the best option for that purpose. However, when Protoss do not expand, Stargate and Twilight Council are the more popular options. Stats and Neeb prefer to proxy a Stargate near Terran’s base, and Terran usually do not have the adequate units to defend against an Oracle at that timing due to their production buildings being elsewhere (see first image below). This can even force Terran to invest in an Engineering Bay and a Missile Turret as a response (see second image below). The rationale is the same as sOs’ approach to have the Gateway units at Terran’s base to attack the workers, whereby Protoss want to deal damage to Scvs when Terran do not have their units at home early on for the proxy strategy.

sOs prefers Blink research with Twilight Council to Oracle with Stargate. This syncs well with the Pylon block harassment technique, as an additional Oracle does not add more value to the attack of Adept and Zealot. Blink is effective against the subsequent attacks, and this is very well demonstrated in game 1 against Maru (see the images below). Blink, with additional Gateways instead of a Nexus, also has the potential to punish the lack of units in the transition from the proxy phase to the standard macro phase.

General discussion

sOs showed a master piece in game 1 by showing a well tailored answer to the current popular proxy strategy. Brilliance, with flair.

Game 2 is very interesting. Lost and Found is a good map for Cannon rush, and sOs pulled it off nicely with the follow up pressure that seals the game. While Cannon rush appears to be a straightforward and effective counter to proxy, it relies on Terran to actually use the proxy strategy. Protoss have to put down the Forge before they have information whether Terran proxy, so the move is not a reactive-counter to the strategy per se. Cannon rush is not a good option against Terran in general especially at the highest level when Terran players can react well accordingly. This is an anti-Maru move.

Game 3 is the worst game Maru played this year. He forgot to put down the second Supply Depot, and a supply block that early in the game for a proxy strategy basically ended the game. I wonder whether sOs’ Pylon at the ramp actually caused this mistake indirectly.

After the quarter-finals, people were so quick to criticize the Terran players for insisting on the same proxy builds, saying that the stubbornness to keep using these bad builds are causing their defeats. Well, these are also the people who say the proxy strategy is too strong and unbeatable up until yesterday.

Some have asked me whether this means proxy is dead in TvP, and whether Terran should still use proxy. I think the removal of the current Cyclone in the upcoming patch has a bigger impact than the games played yesterday on the viability of the proxy strategies.

Lastly, thanks, Mike.

---

If you enjoyed this article, I’d love you to share it with one friend. You can follow me on Twitterand Facebook. If you really like my work, you can help to sustain the site by contributing via PayPal and Patreon. See you in the next article!