EPT[G/D] ZvP - Defending the Immortal/Sentry all-in
Great post by Orek, explains the all-in in its true form:
Let me help set standard for what Zerg should assume for this push. I present no solution, but just raw facts and analysis so that we can discuss based on the same timings/assumptions.
As LonelyClock posted in last page, I think this game needs to be added to the OP to show everyone what we are dealing with.
Introduction
Hello TeamLiquid, I'm Mavvie, a (now mid!) Masters Zerg on NA. ZvP is my best and favourite matchup, yet there is one all-in that is simply really, really good and often beats even the highest level GSL players. This all-in is known as the immortal/sentry all in, or 3 immortal timing, really whatever you want to call it. It's 3 immortals, 6+ sentries, and 7 gates of stalker/zealot reinforcements, often with warp prism/observers, that can be attacking your third as early as 10:20. This thread is designed to discuss and compile the various ways of defending the immortal/sentry all-in as Zerg. I will start off by describing the ways I am familiar with, if you have your own way (against a true immortal/sentry all-in that is attacking you no later than 10:45) that I have missed, please post and include a replay for us to see it in action!
Note: This is intended as a compilation of methods that can stop the all-in. I'm not pretending to invent these, in fact I have little to no experience with most of these methods. I've just been reading TL a lot, and know of these methods. As a result of this, I happen to have no recent replays except for one
Please post replays of you defending this all-in! Not to be elitist, but really masters and higher replays only. Odds are, in lower leagues it won't be executed properly, and often Protoss don't know how to forcefield and abuse terrain. This is a topic that has been discussed heavily before, but never in a really comprehensive fashion. If you read every post in the Zerg Help Me Thread, and Belial's ZvP guide, and the [D] about Nestea's fast mutalisk style, you would know how to defend the all-in. Unfortunately, this is neither fast nor practical, so I have decided to throw it all together into one happy thread to discuss. Also, it's my first guide, criticism will be accepted and I will be actively editing this with new replays, vods, and methods. Also expect some sexy images of Protoss armies getting crushed.Orek is also a boss and described the possible ways of scouting this:
"Squirtle's build order execution looks very close to optimal. As Khaldor said in the video, it is "textbook immortal push" that Zerg players should keep in mind. I doubt that an average NA/EU master/GM Joe can pull off this level of play at the moment, so probably add 10-30sec to every timing for your ladder play.
Timings (only important timings)
6:00&6:25 3rd 4th gas taken
9:00 move out
9:45 arrive near or inbetween Zerg bases if undisturbed. (depend on map distance)
10:10 engagement unit count =3immortals, 12 sentries, 1 stalker, 5 zealots, 1 warpprism, 1 observer(not arrived yet) with +1 attack upgrade.
10:30 +1 armor upgrade finish if no chronoboost is used. -10sec for each chronoboost. In the video, it is not clear if squirtle used chrono or not, but 10:30 is upgrade complete timing at the very latest.
Depending on map and where you set up the defense, Zerg has at least 35-60sec to prepare after lings confirm the push in front of protoss base. Map distance reference:[G]Map Distance & Travel Time. Check 2.25 movement speed section for immortal/sentry.(Ohana has different image, but distance is the same)
Zerg Scouting
I think we need to assume the worst when skill is not involved.
Standard scouting assumption:
1. Zerg can scout 3rd&4th gas timings (Cloud Kingdom 4th gas is the only one hard to scout on ladder)
2. Zerg cannot scout any tech buildings including robo with overlord sacrifice. (not always reliable, so assume the worst)
3. Zerg cannot see the sentries with overlord sacrifice. (1 stalker is already out at 6:15 to kill the overlord)
4. Zerg can know that protoss doesn't have 7:30ish fast 3rd. (1 ling can easily scout it)
5. Zerg can see the move out around 9:00-9:30 in front of protoss base. (lings with careful micro is skill-based, not luck-based)
6. Zerg cannot find/kill all proxy pylons, espcially non-close ones. (It is OK to assume you can kill close proxy before 9:30, but hidden mid map ones are not always found/killable)"
In a typical game, you will not be certain that you've seen every last gateway. You can't always scout every last inch of his base, so you can't definitively rely on this. The difference between a 5gate robo expand and a 7gate robo all-in is minimal -- only 2 gateways difference! Fortunately, you typically react to both of these the same way:
-Take a fourth
-Go more ling heavy than usual
Without further ado, how do we stop this terrible all-in?
Method 1: Out-power it with roach/ling
Good news: This is the most reliable, consistent, skill-based way to stop it. sCCrooked has done a lot of work and research, so props to him
Here is what he has to say about roach/ling:On December 25 2012 14:35 sCCrooked wrote:
I realized I never really compiled all my findings spread out in this thread into 1 post so here goes:
sCCroooked's findings regarding stopping the immortal/sentry all-in from the last 5 months of extensive testing
The best response in my opinion is roach/ling but with a variation on how you execute it.
First off, most Zs go lair first vs P. This is no longer possible with immo/sentry being so popular as it strikes before lair tech units can be out. The immo/sentry expand also has given P a very safe option to expand with this composition and its incredibly hard to tell the difference as a Z, while even a slight delay in the proper reaction could result in a game over.
To stop the immortal all-in, we must recognize its formation and its execution first.
-3 immortals + many sentries move out of natural base with warp prism on the way and a probe or 2 to build pylons
During this first stage, all the P will have is immortal/sentry with no zealot/stalker except possibly 1 single stalker which isn't a significant factor we have to consider in this since the counter-unit is the same.
So what kills 1 stalker/sentries and immortals? Bunch of speedlings! If they forcefield, even better because it still helps your cause out.
-The remaining immortals/sentries and the pylons/warp-prism will gather together in a single spot and defend it with ffs and firing behind the wall. Seeing such mass lings during the first stage of our engagements will almost always force zealot warp-ins. If they warp in stalkers, its just begging for more lings to be made.
This stage is incredibly important because its right around now that the attack looks a lot more threatening with all the warp-ins adding to its power. Since you forced a lot of zealots, make a roach round to deal with it and pre-rally them into a flank at home. They won't have speed done just yet so they're useless off creep. Position these in a nice big circle to lead them into. After 1-2 warps of zealots, you probably won't be able to force any ffs unless the P gets careless since most Ps will realize they're safe behind 14 zealots from your lings and can stop the sentry/immo from being attacked. Try to pick off sentries if you can, but whatever you do, don't waste your lings.
-Finally, the P is completely set up with immo/sentry and will be attacking into one of your bases with the intention of continually reinforcing with zealot/stalker and then winning right there.
Hopefully by this point, you've exhausted almost all the sentry energy. They might have enough for 1 more big donut of forcefields, but that should be it and you can get in with the next wave of roach/ling.
----------------------------------------------------------------------------------------------------------------
Now that we've broken that down and how to respond in each stage, how do we achieve all these timings?
I'd recommend following fenner's build order from this video.
Its very greedy and gets a warren too late for things like 8 gate +2 blink allins or something, but its perfect for the latest meta which is 3 base collossi pushes off immo/sentry or SG expand or the immo/sentry all-in off 2 bases. You really don't see much 4 gate +1 pressure into all-in or those as much these days.
However we're going to make a few changes to Fenner's build order for this particular all-in.
The beauty of this version is that I've found it not only works against immo/sentry all-in, but it also is fast enough and with enough economy and tech to completely kill P who use the immo/sentry as an expand off 5 gates or so.
At this point, I'm assuming everyone has seen the video or can get to 70+ supply before 7:30 rather easily (you really want before 7:15, but 7:30 is a good round number for general use here) so I shouldn't have to explain that build part. We're just going to point out all the variation changes we need to stop this push.
-Speed first, then lair (reason is we need speed to be done no later than 8:30 or so for that 8:50 first stage)
-Stop at no more than 56 drones. (reason for this is larva + timing. Those 4-8 larva can be 16 speedlings instead of drones and stopping at this number does not dent your economy badly)
-Delay macro hatch until your first ling round is made
-Spread Creep with at least 2 tumors. I usually do one with my natural queen and then once again around 50 supply mark and 6:00 from the third base
Using these variations, you get speedlings out quicker in time to deal with the oncoming push. You also have roach tech (which wouldn't have speed even close to done anyways even if lair was done) massing slowly at your base.
The purpose of these lings is more harassment. Treat them like you would a muta flock of 10 or so when facing a Terran base that already has turrets but is trying to build more on the edges or something. Its a volatile force you're facing and a few good ffs can catch some lings, so pay attention to this force very carefully.
----------------------------------------------------------------------------------------------------------------
Well now that we see the variations, how do we execute this?
-While active with your lings on the map engaging his first push-out, target sentries and probes/pylons above all else. If you're REALLY lucky he'll screw up and you can surround an immortal in which case definitely take it, but this would be a rather large blunder so don't count on it. You can quite possibly pick off all probes thus crippling his reinforcement abilities and might even get a sentry or 2 killed.
-Make a round of pure roach at home after your 20-30 speedlings
-Continue to engage and surround with lings while on the map. Do not move your roaches! You basically want to set them up and then lure them into this attacking spot with the flank already set up.
-Its ok to make a few lings if you see your group of lings got ff'd or something and it caused you to lose a lot of them, but concentrate mainly on roaches. If you see they still have a considerable amount of ffs when they reach your base, you can suicide the remaining ling bunch trying to bait some more or take out some sentries.
As soon as you see them actually at your base, its time for the big attack. You need to snap that big 3 or even 4 way arc of roaches around them and prepare to back them off once he makes a donut of ffs. If you did a good job with the lings, this should be his final ffs you're seeing.
After this point, its ok to make either roaches or lings depending on what you see happening. Its very dependent at this point so I can't really guide farther than this.
------------------------------------------------------------------------------------------------------------------------
Are there other factors I should consider?
Yes! There are things like engagement places and looking for likely places for pylons to be hidden.
On a map like Entombed, make sure that small arc behind the middle-left and middle-right rocked-off expansions is cleared. P love hiding stuff back there.
Any areas that have vision-blockers like shrubs or trees need to be checked thoroughly. A good P will send 2 probes to make pylons just in case you see one of them with the army, there's another one running around sneakily setting up places they can reinforce on the map. Sniping those probes is HUGE if you can pull it off.
In a proper engagement of the roaches, you want to set up the flanks long before you'll actually have him in there. The longer you're able to keep him busy, the more roaches you can have waiting for him there. On a map like Ohana, its very likely they'll snake up around onto where you'd normally build your 4th base and try to set up a position there.
Since you already know probably where he'll engage (Daybreak on your third's choke, Cloud Kingdom on that snaking back path to your third, etc), it shouldn't be hard to rally the eggs to different places so you have a huge flank already set up. I can't stress how important this is.
Try to save the lings as much as possible while harassing the crap out of sentries. They buy you time and weaken his potential with baiting ffs and killing sentries/probes/pylons.
The roaches are your hammer though. Think of the time you buy and the damage you do with your lings as a gauge. By killing more stuff or preventing more pylons or every time you make him use a forcefield, that hammer is raised higher and the head is given more weight as time goes on.
Ahead there be some spoilers from a game of the Blizzard Cup Finals, do not read if you don't want Parting vs Life spoiled!
+ Show Spoiler +
------------------------------------------------------------------------------------------------------
Extra concerns I will add later into another well-written part include:
-what if they manage to get down pylons?
-actual in-depth descriptions of how to engage on every map in the current pool
-more replays of myself pulling this off at GM league level
-what if they go for 5 gate immo/sentry expand and I went for this variation? Won't I be behind?
I hope you all find my research helpful and I will continue to post my findings here.
I realized I never really compiled all my findings spread out in this thread into 1 post so here goes:
sCCroooked's findings regarding stopping the immortal/sentry all-in from the last 5 months of extensive testing
The best response in my opinion is roach/ling but with a variation on how you execute it.
First off, most Zs go lair first vs P. This is no longer possible with immo/sentry being so popular as it strikes before lair tech units can be out. The immo/sentry expand also has given P a very safe option to expand with this composition and its incredibly hard to tell the difference as a Z, while even a slight delay in the proper reaction could result in a game over.
To stop the immortal all-in, we must recognize its formation and its execution first.
-3 immortals + many sentries move out of natural base with warp prism on the way and a probe or 2 to build pylons
During this first stage, all the P will have is immortal/sentry with no zealot/stalker except possibly 1 single stalker which isn't a significant factor we have to consider in this since the counter-unit is the same.
So what kills 1 stalker/sentries and immortals? Bunch of speedlings! If they forcefield, even better because it still helps your cause out.
-The remaining immortals/sentries and the pylons/warp-prism will gather together in a single spot and defend it with ffs and firing behind the wall. Seeing such mass lings during the first stage of our engagements will almost always force zealot warp-ins. If they warp in stalkers, its just begging for more lings to be made.
This stage is incredibly important because its right around now that the attack looks a lot more threatening with all the warp-ins adding to its power. Since you forced a lot of zealots, make a roach round to deal with it and pre-rally them into a flank at home. They won't have speed done just yet so they're useless off creep. Position these in a nice big circle to lead them into. After 1-2 warps of zealots, you probably won't be able to force any ffs unless the P gets careless since most Ps will realize they're safe behind 14 zealots from your lings and can stop the sentry/immo from being attacked. Try to pick off sentries if you can, but whatever you do, don't waste your lings.
-Finally, the P is completely set up with immo/sentry and will be attacking into one of your bases with the intention of continually reinforcing with zealot/stalker and then winning right there.
Hopefully by this point, you've exhausted almost all the sentry energy. They might have enough for 1 more big donut of forcefields, but that should be it and you can get in with the next wave of roach/ling.
----------------------------------------------------------------------------------------------------------------
Now that we've broken that down and how to respond in each stage, how do we achieve all these timings?
I'd recommend following fenner's build order from this video.
Its very greedy and gets a warren too late for things like 8 gate +2 blink allins or something, but its perfect for the latest meta which is 3 base collossi pushes off immo/sentry or SG expand or the immo/sentry all-in off 2 bases. You really don't see much 4 gate +1 pressure into all-in or those as much these days.
However we're going to make a few changes to Fenner's build order for this particular all-in.
The beauty of this version is that I've found it not only works against immo/sentry all-in, but it also is fast enough and with enough economy and tech to completely kill P who use the immo/sentry as an expand off 5 gates or so.
At this point, I'm assuming everyone has seen the video or can get to 70+ supply before 7:30 rather easily (you really want before 7:15, but 7:30 is a good round number for general use here) so I shouldn't have to explain that build part. We're just going to point out all the variation changes we need to stop this push.
-Speed first, then lair (reason is we need speed to be done no later than 8:30 or so for that 8:50 first stage)
-Stop at no more than 56 drones. (reason for this is larva + timing. Those 4-8 larva can be 16 speedlings instead of drones and stopping at this number does not dent your economy badly)
-Delay macro hatch until your first ling round is made
-Spread Creep with at least 2 tumors. I usually do one with my natural queen and then once again around 50 supply mark and 6:00 from the third base
Using these variations, you get speedlings out quicker in time to deal with the oncoming push. You also have roach tech (which wouldn't have speed even close to done anyways even if lair was done) massing slowly at your base.
The purpose of these lings is more harassment. Treat them like you would a muta flock of 10 or so when facing a Terran base that already has turrets but is trying to build more on the edges or something. Its a volatile force you're facing and a few good ffs can catch some lings, so pay attention to this force very carefully.
----------------------------------------------------------------------------------------------------------------
Well now that we see the variations, how do we execute this?
-While active with your lings on the map engaging his first push-out, target sentries and probes/pylons above all else. If you're REALLY lucky he'll screw up and you can surround an immortal in which case definitely take it, but this would be a rather large blunder so don't count on it. You can quite possibly pick off all probes thus crippling his reinforcement abilities and might even get a sentry or 2 killed.
-Make a round of pure roach at home after your 20-30 speedlings
-Continue to engage and surround with lings while on the map. Do not move your roaches! You basically want to set them up and then lure them into this attacking spot with the flank already set up.
-Its ok to make a few lings if you see your group of lings got ff'd or something and it caused you to lose a lot of them, but concentrate mainly on roaches. If you see they still have a considerable amount of ffs when they reach your base, you can suicide the remaining ling bunch trying to bait some more or take out some sentries.
As soon as you see them actually at your base, its time for the big attack. You need to snap that big 3 or even 4 way arc of roaches around them and prepare to back them off once he makes a donut of ffs. If you did a good job with the lings, this should be his final ffs you're seeing.
After this point, its ok to make either roaches or lings depending on what you see happening. Its very dependent at this point so I can't really guide farther than this.
------------------------------------------------------------------------------------------------------------------------
Are there other factors I should consider?
Yes! There are things like engagement places and looking for likely places for pylons to be hidden.
On a map like Entombed, make sure that small arc behind the middle-left and middle-right rocked-off expansions is cleared. P love hiding stuff back there.
Any areas that have vision-blockers like shrubs or trees need to be checked thoroughly. A good P will send 2 probes to make pylons just in case you see one of them with the army, there's another one running around sneakily setting up places they can reinforce on the map. Sniping those probes is HUGE if you can pull it off.
In a proper engagement of the roaches, you want to set up the flanks long before you'll actually have him in there. The longer you're able to keep him busy, the more roaches you can have waiting for him there. On a map like Ohana, its very likely they'll snake up around onto where you'd normally build your 4th base and try to set up a position there.
Since you already know probably where he'll engage (Daybreak on your third's choke, Cloud Kingdom on that snaking back path to your third, etc), it shouldn't be hard to rally the eggs to different places so you have a huge flank already set up. I can't stress how important this is.
Try to save the lings as much as possible while harassing the crap out of sentries. They buy you time and weaken his potential with baiting ffs and killing sentries/probes/pylons.
The roaches are your hammer though. Think of the time you buy and the damage you do with your lings as a gauge. By killing more stuff or preventing more pylons or every time you make him use a forcefield, that hammer is raised higher and the head is given more weight as time goes on.
Ahead there be some spoilers from a game of the Blizzard Cup Finals, do not read if you don't want Parting vs Life spoiled!
+ Show Spoiler +
Your object here is to make "The Hammer" part as heavy and as painful as possible. This was showcased in some replays but I don't have many of my own to give. The best execution in a pro match I've seen was in the last Blizzard Cup Finals on Entombed. Life did exactly as I've been testing and it stopped parting cold. He was trying to warp in nothing but sentries at the end because Life did so well with the lings and had so many roaches waiting when Parting finally made it over.
Now granted, I think he got pretty damned lucky with the probe/pylon snipe, but the concept is very sound and still works at top 50 GM level as I've shown before.
Now granted, I think he got pretty damned lucky with the probe/pylon snipe, but the concept is very sound and still works at top 50 GM level as I've shown before.
------------------------------------------------------------------------------------------------------
Extra concerns I will add later into another well-written part include:
-what if they manage to get down pylons?
-actual in-depth descriptions of how to engage on every map in the current pool
-more replays of myself pulling this off at GM league level
-what if they go for 5 gate immo/sentry expand and I went for this variation? Won't I be behind?
I hope you all find my research helpful and I will continue to post my findings here.
Everything I say below still applies, but like the rest of the guide is a bit outdated. For example, you should hit 70 supply at 7:20, not 8:00. "Macro better" is hugely important!
Okay, so you're playing a standard ZvP, when all of a sudden you scout immortal/sentry all-in. You have balls of steel, so you decide to use roach/ling to kill his all-in (or pressure his expand)
Be extremely ling heavy. Like really ling heavy. As in, 20 roaches MAX, preferably fewer. You want as many lings as possible, you really need to macro perfectly. If you don't hit 70 food by 8:00, and you don't make a macro hatch at 8:00, you'll simply lose with roach/ling. Also, if you just 1a into his army, you /should/ lose. Unless Protoss sucks.
How to engage:
You need 2 to 3 flanks minimum. If you come at him from one direction, he'll just laugh and forcefield you to death, while his army (and lead) snowballs. You must engage from multiple directions, hopefully in the open. If you manage to fully surround his units, you've won the game. Try to micro so that zerglings are attacking the immortals and stalkers, but at the very least try to draw the fire away from the roaches. The AI makes units target the closest units, and if the immortals are shooting zerglings then you're in a pretty good shape.
Oboeman recommends flanking with roach/ling from both directions, and that creep spread is extremely important, even worth an inject or two:
When trying to use roach ling, make sure you have both roaches and lings on all sides of your flank. If you hit with roaches from one side and lings from the other, it's pretty easy to forcefield you out while still killing many units. If you have roach/ling on both sides, it's much harder and he needs to use more than twice as many forcefields.
Also I know we are all very greedy with our inject,. but a bit of creep really does help out so much.
Since I'm a noob, I'll typically have my split my army into three parts. I have half my lings on 1, half on 2, and my roaches on 3. 1 and 2 should be opposite directions, while your roaches hit the side flank.
For example:
Roaches
(down arrow)
Lings--->>Protoss<<----Lings
----------Wall--------------
Protoss has to forcefield 270 degrees of Zerg forces, and if he misses even one, you swarm in. Even if he forcefields correctly, he's trapped himself, hopefully out of range of your hatchery. Just wait for the FF to dissipate, then re-engage. It's possible to defeat immortal/sentry with roach/ling, but you need to macro perfectly then engage perfectly. Very hard.
Replays:
+ Show Spoiler +
I have a few replays of plat league me vs diamond league Protoss during a lesson with TangSC, I will try to find these ASAP.
In the mean time, please post replays of this method!
Method 2: Basetrade with roach/ling
As far as I know, this is the only method of defeating the all-in consistently at the pro level. It's really easy. Once you see him move out, you mass spines at your natural, while taking your current army to his base, while avoiding his army. You sac your third, but kill Protoss's natural, so it's a 2base vs 1base situation. Honestly I don't know how to transition out of this, I guess your best bet is to go mutalisk/zergling and immediately double expand. This style has been done before by blade55555, and he's had extreme success with it in the grandmaster league.
Focus fire the probes + nexus, then try to snipe the main if you can.
Upon playing more with the various ways of defending the all-in, this is totally the best one imo. I'd like to emphasize just how easy and simple it is; even with bad macro it's a free win. Just remember to mass spines around when he moves out, preferably make ~5 about 30 seconds before you think he'll move out.
Replays:
+ Show Spoiler +
Replays of me:
http://drop.sc/241141
Vs Master league Protoss: http://drop.sc/243983
Blade's Replays:
Game 1
Game 2
Edit: Big thanks to blade55555, for not only providing new replays, but also a video analysis! Props!
I saw the replays that op uploaded were a bit old (month in a half) so here are a few in here:
http://www.mediafire.com/?2kp5vvvt5yk422v
One of the replays in there I analyzed if people are curious about that:
Method 3: Fast 3 base mutalisks
Originally done by Nestea in the GSL, a truly amazing build that wins against robo expands and wins against robo all-ins.
While any fast mutalisk build works, this is the rough outline of Nestea's:
1) Fast gas after third (~4:30)
2) First 100 gas into ling speed, next 100 into lair
3) Drone up as usual, take 3x gas at 6:30
4) If you scout no gateway all-in, skip roach warren + evo chamber entirely
5) If you scout robo, Spire as soon as lair finishes
6) 2x gas at 7:00-7:30
If he all-ins, mass spines in your natural, sac third (leave guys mining gas) and basetrade with muta/ling. Reinforcing mutas go to his base, reinforcing lings stay in your spine forest.
If he expands, take a fourth and pressure with mutas. Honestly, you should win the game then and there. 10-15 mutas pop at 10:30; no protoss is ever prepared for that.
Original Thread
Replays:
+ Show Spoiler +
Method 4: Mass zergling into 3 base muta (not fast)
So this is a style that Belial has used, and it's my personal favourite. Fortunately for me, people in diamond suck, so it works just great.
Open standard gasless 3 hatch, 3x gas at 6:30, blah blah blah open standard, hit 70+ by 8:00, and macro hatch immediately. Mass lings. +1 Carapace helped me a lot, shouldn't usually get this though. So if you mass lings from 8:30+, you can sort of kind of engage if you flank a lot. I really just stall for mutas, then I'm able to win the engagements and proceed to win the game through not playing like a moron. I imagine that at higher levels you could pull this style off by making a bunch of spines to add critical DPS to your zerglings when you engage, but you'd have to make them ~9:00 because of how long they take to morph.
Replays:
+ Show Spoiler +
Method 5: Baneling drops
Here's one that I hope you guys can explain and give replays of working! The problem is that you can't get drops in time to stop the all-in...maybe a modification of Nestea's build could work, but there's no high level replays of this working. Despite this all, every second post about dealing with sentry/immortal all-ins suggests baneling drops. Even Kevin te Raa has said that he will be coming out with a guide(?) or at least replays on the topic. I'm looking forward to that!
Awesome, first replay I've seen of baneling drops being used against sentry/immortal. Here's the build that I think would be "optimal", from seeing the replay:
-gas @5:00
-gas @6:00
-gas @7:00
-ling speed first, then lair
-no roach warren/evo unless needed
-@100% lair, get drops + overlord speed
-@50 gas, baneling nest
-macro hatch when money allows, should be ~8:45
-cut drones ~50 (I really don't like this part of the build, but it seems to be necessary to defend this particular all-in. I imagine that if he had skipped the roach warren, he would've had up to 10 more drones here)
-morph 24+ banes upon completion of baneling nest
-don't get supply blocked!
Engagements:
In the replay, he engaged with the overlords from one direction, zerglings flanking. It didn't work, along with having 1500/500 banked. If there was no supply block, it would've been an easy cleanup. Banelings killed much of the sentries, and i think having ling/overlord on one side and ling/bane on the other would be terrific. This looks pretty promising. I still don't like 15 hatch, because the slight economic gain is negated by almost always having to pull drones.
Replays:
+ Show Spoiler +
Analysis of new ways to stop it:
An excerpt of Orek's post, explaining the FF count and general rules for possible solutions:
+ Show Spoiler +
10:10 Forcefield count analysis
energy regeneration rate is 0.5625energy/second. Time it takes to regenerate 50 energy for forcefield = 50/0.5625=88.8888=almost 90sec=1min30sec. Therefore, whichever sentry made after 8:40 has only 1 force field, and those before 8:40 have 2 forcefields. Among before 8:40 group, sentry made before 7:10 can even have 3rd forcefield as well. From video analysis, it looks 1sentry =3FF, 3 sentry =2FF, 8 sentry =1FF. 1*3+3*2+8*1=17 forcefields are available for protoss at 10:10 mark. Since there are 12 sentries and 90sec is 50energy regeneration interval, 90/12 =7.5sec is average FF refill period. That is to say, protoss gains 1 additional forcefield every 7.5 seconds on average. Therefore, 10:10 17FFcount -> 10:40 21 FFcount ->11:10 25 FFcount ->11:40 29 FFcount if protoss keeps all 12 sentries alivem
With all these in mind, Zerg should go for a build that
A. can defend immortal/sentry all-in (obviously)
B. can defend other 2 base all-ins that involve 6:00-6:25ish 3rd&4th gas.
If you do a build that can satisfy A, but not B, then it relies on your luck or opponent's lack of skill. Your build has to satisfy both A&B to play consistently. Important thing to remember is that you never know 100% if it is immortal/sentry push until 9:00 mark. It could be delayed 2 stargate(usually 5:30&6:00 gas) as far as you know from the scouting information.
I hope these information set common ground for discussion.
Closing notes:
Thanks for reading, hopefully you enjoyed and can try out some various ways to stop this powerful all-in! Since I don't have access to many recent replays against this style, please post any replays and mention which style you used! Honestly, I hope that we can kick this all-in out of the metagame. Take that, Protoss! Just kidding love you guys :D
Change log:
+ Show Spoiler +
Edit 1: Formatting changes, changed the header images to some that were "less offensive"
Edit 2: Added content to baneling drops, including a replay thanks to Chaos. Also added change log.
August 17, 2012: Added in blade55555's new replays + video analysis the the roach/ling basetrade section.
August 19,2012: Added the video at the top, also included excerpts from Orek's post that help define both the all-in itself, and guidelines for builds that may attempt to stop it (and other similar-looking all-ins, from a worst-case-scenario zerg scouting point of view)
