TvP mech build evolution

Briefly, for anyone reading this not already familiar with lyyna's or the mech resource thread, here are some tvp mech laws that I keep:

Tank/hellion as a tvp mech style is inviable-- it uses money too inefficiently to have a lategame, and thus is a 2-3base all-in. The problem with that is protoss can hold and/or outmaneuver such aggression and win a pseudo base-trade. Thus management styles of mech must be used instead (think: bw Flash).

1rax cc gasgas is inviable-- there are some all-ins that you can't scout before choosing gasgas (as opposed to rax rax). And 3rax before gas is inviable for mech because it is exceptionally vulnerable to immortal all-ins at your 3rd either because you've delayed gas too badly or because you wasted too many minerals making marine/bunker trying to hold. Thus you must open with a gas build (not necessarily a tech build).

The endgame composition of bc/ghost/raven + whatever else can beat any protoss composition. Winning is thus a matter of staying alive until you obtain this composition.

When I say "the lyyna build", I'm referring to some variation on 1tank cc reactor port armory raven thor +1 banshee fact ghost acad cc, make 3 thors then 8 tanks then 3 more thors then start bc production. I don't know if lyyna still uses this build or not; last I saw he was experimenting with some early fact-based aggression. When I say anything about the lyyna build/style I really am referring to my own understanding/interpretation of it, and don't guarantee that it agrees with what lyyna actually thinks. I don't use his build because it doesn't work for me; he may very well have some skill and/or understanding that I lack which makes his build better than what I'm doing.

http://drop.sc/280418

Ok, it's a little messy, but this is what I think will become my standard build tvp. I've been quiet in lyyna's and the mech resource threads for a while because 1. I've been fairly inactive and 2. when I have been playing I've been doing a lot of theory-crafting and specific-scenario practicing. So I did a lot of theory-crafting for tvp and the build I used in this rep is the result. It needs refined and practiced in a variety of scenarios to confirm that it's safe and adaptable, some timings/numbers may need tweaked, and obviously my macro was very poor, but the core of it is present.

The following is by no means my complete thoughts and notes on tvp mech, but reveals some of my thought processes in creating a tvp mech build. I've used a great number of builds so far, and have found lethal inflexibility problems with all of them so far. That is to say I find possible scenarios where I'm either not aware of a solution or am unwilling to employ the only solution, thinking it a temporary fix within the game and not something that can actually prevent defeat-- for example lyyna will frequently stay on 3base for very long periods of time, turtling very hard, and just slowly build up bcs and hope he can outlast his opponent's patience. I'm perfectly fine with long games, and often employ great patience, but I want to play in a manner that makes me feel like I'm accomplishing something within each game that brings me closer to victory-- that my actions lead along a clear path to a possible win, and I abhor completely giving up the game to my opponent and hoping he doesn't win the game rather than doing something to prevent him from winning it. I also like expoing.

I've been quick to explain to others why their mech builds are bad (lyyna's being the exception-- I have nothing against his build or style, I only seek to see if I can make it better), but have not in a long time advocated any solution of my own. This is because I fell out of confidence with my own build, until now. This build still need not be the final word to tvp mech; it doesn't need to be perfect, it just needs to be adaptable. That is to say, I need to be able to scout the possibilities that I'm not prepared for and need to be able to adapt my play to be prepared for them having either scouted them or failed to rule them out.

I analyzed my previous builds, focusing on my assumptions-- the things I took for granted, and tried to understand them and see which were accurate and which were not, and design a build around those. I hope that my explanation of the evolution of my thought process in such detail as I present it here helps others come up with their own builds. I'm not here to teach you how to mech tvp (yet); I don't think you can learn the matchup from this, but maybe you can learn how I think about the matchup, or maybe see the fallacies in my thought processes thereby helping me understanding the game better.

My previous build/style focused on securing expos as fast as possible under the assumption that 2fact tank/thor and 1reactorfact hellion production along with constant ghost production (with early 2nd rax) could hold anything provided battles were fought near pfs. Compared to the classical lyyna build I was putting a greater focus on ghost upgrades and production, and completely removing port units from play until a certain thor/tank count was achieved. It was a very efficient build and gave me a lot of utility in the form of the ghosts and quick nuke-harass potential. Not making a port or ravens or banshees or cloak all gave me gas to streamline 2fact 2rax production and fast +3 attack. The build was an answer to a strong 3-4base immortal/archon + speedlot push. The only apparent hiccup in efficiency was making 3 thors before cc, which are expensive and slow to build; I made them because lyyna had determined they're good at holding any early aggression. What it obviously lacked was harass potential prior to nuke harass, good scouting, and a means to push/secure pfs by.

Ultimately the purpose of the build was to hold 1base aggression, 2base immortal all-in, and the 3-4 base immortal/archon attack.

This build/style gave me my highest win% of all matchups for a couple months. I was getting the money I needed to build a strong fleet of bcs with a good core of ghosts to support them. I stayed alive by responding to his army with sensor towers. Once I got my final composition there was nothing protoss could do; I just moved around, harassing with nukes as was possible, looking for a moment to EMP his mothership before engaging and destroying his army.

There were some early warning signs that the build might not be the catch-all I'd hoped it would be. The most consistent worry was my hellion count being constantly low. I didn't mind the low harass potential, because I don't need early probe kills to win. What I did mind was the inability to scout his movement outside of tower range, his current unit count (before warp-ins), and more importantly his unit composition. I was aware of some annoying scenarios where I'd thought of solutions for them, but I couldn't employ those solutions if I didn't scout the right information in time. Even worse were scenarios where after using my hellions to scout I would lose enough of them such that it made it hard to defend his next aggression.

The build flat out simply cannot work on antiga. This gave me a hard time in several playhem tournaments. I can't count on future maps not having similar difficulties.

One problematic game was my match vs Sage in playhem. I defended his two-pronged blink/colos aggression pretty well for a while, and was starting to build up bcs, but eventually took too much damage and couldn't defend. In analyzing the game I concluded that a pf at the front of my nat (cloud kingdom) was the solution to that scenario, and that with that I would've had a good chance to win game. Unfortunately it's not easy to scout protoss going double pronged colos/stalker count. Thus apparently I needed a preemptive pf. But a pf in my nat isn't going to be very useful later in game. A pf on the tower would cover my nat and help me take the nearby expo as a 5th base, but my army isn't big enough to actually secure such a pf if protoss notices it being built. Pushing pfs to key locations is very important on all maps, and if protoss contests its placement you simply can't defend some forms of aggression once you get spread out. I don't like the lyyna approach here of simply staying on a low number of bases and slowly building up a force that can secure further expansions, as I feel it gives protoss too much freedom. This almost never actually matters because protoss don't make the most out of their money/units vs mech most of the time, being unaccustomed to it still.

Another annoying game I faced was a scenario where protoss opened phoenix, which I laughed at since I was making thors. I scouted to see if I had an attack timing, and determined that I did not, so I just did my thing. He looked to be making immortals, which I didn't want to attack into. This reminds me of the immortal/archon attack I'd designed my build around holding, so I figure I'm fine. I built up my army, secured expansions, and noticed he still had no gateway units. I figure I might have an attack window now, and move out. I'm building bcs by now, so even if I have to fall back I have pfs and bc production to fall back on. Well I try to attack him, and discover he has a formidable carrier force.

My hellions couldn't get past his immortal/phoenixes so I couldn't know he didn't have more immortals behind the ones I could see, or archons or something. Even if I'd been aware that he was making carriers I'm not convinced I could have adapted in time to stop them. Carriers build faster than bcs when chrono'd. If I had tried to attack him before the carriers were up he could've defended by warping-in. I'd scanned his main and seen no signs of the tech (his stargates were in the back of his 3rd on ohana). So in the end I tried to stall by making turrets and repairing my thors, but his carrier count alone was tough to manage, and the situation was simply hopeless after adding in his other units.

So there are a few problems found in this game. 1. I can't know what lies in the fog. I can't know his unit count, and that unit count is easily increased if needed (warp-in). 2. A small number of protoss units can defend small amounts of harass so easily that I gain no scouting information by it. I've faced the same problem using builds involving banshees (such as/similar to the classical lyyna build). 1port banshees and 1fact hellions are easily shut down by a protoss who's on their toes about defending. They don't even need to be hiding secrets behind this for it to be formidable. 3. hard/fast tech can be difficult to even scout, much less punish. It's hard to catch protoss off guard, because they might have more good units than you're aware of (more colos/immortals/archons than you've seen), and they can easily augment their force with chrono'd warp-ins, as is needed.

Moving out vs protoss before a good bc/ghost count is obtained is risky, hence lyyna's maxim of never attacking. This is absolutely the best plan in most scenarios with the classical lyyna build. This is because not losing units allows you use your money very efficiently, so even if you're stuck on a lower base count than you'd like you can still build up a formidable endgame mech composition (bc/ghost). This works out usually because protoss typically is anticipating an attack from you (because that's how bio works, and that's how most meching t play-- with a 2-3base tank/hellion all-in).

Protoss rarely realize that in fact you have no intention of attacking in time to capitalize on that fact. If they do, they usually try to mass expos. Lyyna's play again accounts for this attempt, and his modest harass potential becomes potent once protoss spreads themselves out very thinly. Additionally his play is designed to absorb whatever protoss does, so it doesn't matter how much money they have. We've all seen the replays where protoss banks 10k/10k vs him and just throws a max at him, remaxes and repeats until they have no money at all, and he moves out and wins-- 40-60 minutes in. This method of maximizing all the advantages of mech is what I'd focused on too. And it really can work in most scenarios. However, I feel that this scenario where protoss exploits the fog and your passivity properly and mass carriers like this is a scenario where that style fails.

Turrets, +3 thors, and initial bcs (lacking upgrades and yamato energy) just can't defend a certain carrier count long enough. Emp/seeker can't guarantee kills. Even if he is stacked, the proper way to use a large carrier count is to hit and run-- running via mothership recall if necessary. Counting on a protoss who's smart enough to make this high carrier count early to throw his carriers away by stacking them heavily and having no means of escape isn't a viable strategy. It's not that terran units can't handle a good carrier count; again lyyna has shown us proof that literally a full max of only carriers can be beaten by bc/ghost/seeker. The problem is that we don't have the necessary unit count/energy accumulation at this timing to kill the carriers. And merely defending with mass turret/repair is no longer a viable strategy since his carriers can do so much damage quickly. Proper use of recall defensively or offensively by recalling to a different location to attack makes his carriers more cost efficient than your defenses. And it's not like his carriers are alone-- wherever he makes a hole with his carriers is a spot stalkers, zlots, or dts can now exploit.

Ideally such a strategy can be beaten by identifying it early and attacking at the right time, and committing to exactly the right extent before retreating as is necessary, but all that knowledge is difficult if not impossible to obtain through passive play with mild harass. An additional problem comes in misidentifying the protoss as using this strategy, and reacting grossly inappropriately vs what they're actually doing. For example, seeing only high tech units (e.g. phoenix/immortal, like the scenario) and no mass expansion attempts could make you think the right move is attacking to stop them from getting carriers. If it turns out they just have some extra robos in their 3rd making colos (or even more immortals) instead of carriers, you just lost the game if you move out.

You can't afford to scan everywhere, and if their obs spread is good they can block harass-scouting attempts without revealing more information. Or what if you were right, and they were starting to mass carriers, but cancel their current production upon seeing you move out, and just chrono warpgates and warp-in in the fog? You may not be able to know that the attack is no longer appropriate until you're in a vulnerable position. And expecting him to warp-in and backing off preemptively after showing him your army could be equally disastrous if you think you've delayed his carrier production and in fact he was unphased by you moving out and didn't stop carriers?

Vikings are not a solution-- they don't bring you closer to your bc/ghost/raven composition, but do delay it. Additionally, being low-hp, he can kill your vikings with his carriers alone with proper control/support. If he wants to use blink stalkers or storm if he has it, he can waste your money and port queues even further. There's no such thing as a "defensive viking count" vs carriers, since he can easily snipe the no-armour low-hp units out of range of your turrets/thors (the carriers themselves). Carrier + interceptor range exceeds viking attack range. Massing thors is also not a solution, because immortals can be produced faster thanks to chrono, and carrier/immortal has a much better attack surface area than do thors because of thor's size compared to immortals, and the fact that carriers fly. BCs are the best unit against carriers, and can simply be outproduced thanks to chrono.

When I described this scenario in lyyna's thread, he was confident that his build is designed to prevent it. This may be, but this long story is not just about the super specific scenario I encountered, where I probably could've done a great deal of things better to win. Rather I'm revealing the protoss' strengths relative to mech terran. Fog/warp-in is a powerful combo. The difference between a small number of defensive immortals and an immortal all-in (or at least attack, doesn't have to be all-in) is just warp-in cycle or two. There exist scenarios where sitting and trying to just weather whatever protoss does to build a bc count is the wrong choice, and it can be difficult if not impossible to identify such a scenario. Observers give protoss quick information which warp-in lets them react to rapidly.

The next scenario was troublesome for my exact build rather than the style. This is probably a scenario which helped guide lyyna to choose the build he usually employs-- that his build is a reaction to this scenario as mine was to immortal/archon. I played against Welmu in eu playhem. On daybreak he used a standard build: 3base colos. What he did differently than most protoss I play beyond having better macro and control than most (both of which are manageable), was that he was aggressive with his colos. He had a confidence in his attack that most protoss lack against my mech. He just moved near my nat, then walked to my 3rd, and back, on the outside (my 4th base area), back and forth. This forced me to unsiege my tanks and move around, reacting to him with sensor tower detection. I couldn't leave just a few tanks at either spot, because his force was great enough that he could pick off a small number of units easily. His units accumulated greater than mine as is the norm with protoss vs terran, which made it impossible to secure my 4th ever. I could prevent him from killing me, but not move out. Obviously that's not something that can last forever if I plan on winning the game.

In the next game, on entombed, he used a similar strategy. I altered my build this game adapting to what he'd done on daybreak, and made banshees. I was hoping that if he tried the same thing I would eventually accumulate enough banshees on 1port production that I could fend off his colos. I forget the details of the game, and quite possibly to my mechanical mistakes alone I lost the game. Just making banshees didn't solve the problem at anyrate, however I lost. The solution is simple of course: just put your port on a reactor and make some defensive vikings. Vikings are only a major problem when you need a great number (which is what makes bio so annoying). A small number of vikings complement your immobile tanks well enough to dissuade protoss from being aggressive, letting you progress through the game unscathed. However I don't want to make my build so much less efficient making a port just to have the potential to defend a very specific form of aggression that protoss need not even commit to. That would set me back vs every other possible scenario, and when the scenario I'm preparing for did arise, I would just go even. Going even isn't a problem in and of itself, because of course we as proper meching terrans become more likely to win the game the longer it goes, whatever happens in the middle. All we need to do is stay alive-- but ruining my build could make staying alive in other scenarios harder.

The last problematic scenario is pretty amusing. My friend used some strategies/builds against me in practice games that were designed just to blind counter my mech. The killer builds involved immortals, possibly with a very fast warp prism. I didn't study the games too well, but he told me that the builds could only possibly be used as a blind "counter", that they were completely unsafe taking into account other/standard tvp builds. That's fine though, since that's what players in tournaments and team-matches try when they know their opponent. These killer builds actually weren't very scary, and some easy adaptations enabled me to consistently defend them. Long story short: what I found to be best at holding them were making tanks, not thors (1 thor is fine), and not walling with depots. You actually can't prevent protoss from sniping your depots with immortals early on. Losing the depots forces you to replace them, and means you have no wall, making your thors less effective at defending (because a greater number number of units can attack).

There may have been other reasons/concerns, but I forget them. Tanks instead of thors early on was something I'd considered at other times for reasons I can't recall. Thors really are wonderful at defending, but they keep you honest too. Once you put up a turret in your nat protoss can't necessarily be sure what tank count you have, and thus isn't likely to try to aggress into your nat. You need a good marine count (12-16) to be completely safe, but usually you can forgo them if you're on top of scouting.

What was actually problematic was what he was doing in the longer games. He would just mass immortals with some stalker support. This force of course can stop harass, allowing him to expo as he pleases, and stop any attack/counter attempts. More importantly though was what he would do with them. Even if he didn't see a particularly good attack opportunity, he would just attack into me. It was suicide-- he'd lose all his immortals, but I'd be a bit broken. You can't stop suicidal immortals from killing key pfs for example.

The real deathblow though were his massive pure-stalker warp-ins. His immortal attacks would force me to reorganize my defenses if nothing else-- to secure the most vulnerable expansions or key strategic locations, etc. This means that something becomes vulnerable in return until I replace both my units and my pfs. Lyyna doctrine states that you can't lose if you're spending efficiently, but he'd force me to replace pfs and units. These are battles I could not possibly prevent or improve. There's no way to position or engage better vs these. It's not like I can sit in a single spot with a perfect spread of tanks-- his army is more mobile, he can suicide where my army isn't to kill stuff I'd prefer not die if I don't respond.

You can't be maximally efficient if you can't fight near a pf-- the pf gives you extra splash/hp/and is a great repair target. If you aren't efficient the fact that protoss is outmoneying you really starts to matter. I can't stay even in expos since his heavy immortal composition means he can move faster than me with obs vision and quickly and easily stop poorly defended expansion attempts, and he can outmaneuver my army. So if I just sit at my 4th base I can't stop him from hitting my nat for example. So it's similar to the scenario against welmu's colos, except I can't just allow myself to be contained and defend-- immortal dps against buildings and my main damage-dealing units means he's guaranteed to do high amounts of damage.

This creates a war of attrition I can't win: because whatever vulnerabilities the immortals can't exploit, his mass blink stalkers can. I can't cover all of my bases at once-- my forces aren't as effective as his split up, and he's not even forced to split his up. Since I can't attack or harass or counter at all all I can do is react to his army and rebuild what he kills. Over time this has him winning no matter how inefficiently he uses his units.

The situation isn't hopeless, I just need to adjust my build accordingly. Battlecruisers are a big no-no against this composition. The immortals knock out your supporting units, and the giant stalker waves are actually effective since you lack the support (pfs and tanks primarily vs stalkers). That's ok though, since the only thing we really need bcs against are carriers. With proper support they're good vs everything protoss can throw at us, which is why the lyyna style focuses on accumulating them-- but that doesn't mean we have to make them. I had to delay bcs against the immortal/archon attacks to get fast +3, to accumulate a higher ghost count, and even to make a few extra tanks as well if protoss did choose to use that composition and poised to be aggressive. I've also uploaded many replays of very scrappy tvps where my income made bcs impossible anyway as evidence that a bc-less composition can beat protoss in the endgame.

So I can stop really early immortal aggression without trouble; the problems just start after I take 3rd base. With my old build I was getting unit production after cc, so there was a significant delay between taking 3rd and actually having new units from the new production buildings to defend that 3rd. Protoss can decide on the spot if they want to expand or warp-in and attack, so leaving vulnerabilities such as those is unwise.

So ok, why not lyyna's build? Currently he's getting a reactor fact, techlab port, and techlab rax (for ghosts) before 3rd cc. Hellions help scout, identify and eliminate or prevent proxy pylons-- nullifying the warp-in advantage, scout unit composition and possible protoss attack intentions/timings, etc, and their presence may dissuade protoss from leaving his base at all. Banshees compound this, giving new harass opportunities, and with good control can really mess up his attacks. EMP's uses at defending immortals are obvious. Well first of all, this is delaying the 3rd cc by quite a bit just to have the potential to defend potential aggression, giving us less money in the near future to defend whatever comes. Staying alive still requires income that exceeds maintenance expenditures.

Secondly 1port banshee production doesn't give that useful a number. They're good vs early colos for sure, which I believe is lyyna's primary reason for making them. However they don't give very good harass potential-- protoss can easily defend them in the numbers they're coming at the times they're hitting. But that's fine, we don't need to do damage anyway. They help stop proxy pylons and are airborne dps if nothing else, and of course, immortals can't hit them. Well, they can't kill immortals either when there are a great number of immortals. The classical lyyna tactic of scan, kill obs, mop up with banshees doesn't work when by the time the mopping up part is complete you've lost everything else, such that you can't defend the stalker waves that follow. Banshees are great against a small number of key targets, not so good vs masses.

So how do we actually stop the immortals? Well my experiments suggest that a strong ground army behind/next to a pf is the best solution. So we already knew pfs were good, and were already made aware of the need to secure pfs in key locations to help defend locations far from each other, so this isn't news. That a ground army does well is fine since he can't mass carriers if he's massing immortal/stalker. So we have a new major objective: pushing pfs. Why? Because it lets us defend what we've got and secure new expansions. Why? The former is necessary to fulfil our primary objective: staying alive; the latter gives us the means to win eventually. Without money you can never actually win the game unless you bore your opponent into leaving.

So what does it mean to push pfs? Make a cc, float it to a key location (often a xel naga watch tower), and then move your army to defend it while it upgrades into a pf. This is actually more involved than it sounds. To be able to do this you need an army which can actually move to defend the constructing-pf. This means you need to be able to fight his army away from a pf if need be, or have some kind of a deterrent. So when playing vs my immortal/stalker friend on ohana I found that I needed a pf in front of my nat, that the tower was too far initially. The pf in front of my nat helps secure the 4th though, since the terrain around the 4th lets you defend initially without a pf. That pf does defend your nat, helps defend your 3rd, and can be used to fall back on if you lose your 5th later-- so it's ok to need to make that pf. The tower can help secure your 5th, but it also helps you defend your 4th. So per pf cc you're getting a lot of utility-- it's not wasteful. If you're ahead you can make bolder plays with a proper pf perimeter, and if you're behind you can turtle behind them. Mass immortal may pop them with ease, but I'd rather lose a pf than 1500 hp worth of units which costs more and takes more time to produce; and if you're on top of scouting and moving your army you should be able to emp his immortals before he pops the pf with your tanks in range, so he's going to lose some immortals to get the pf anyway.

My first solution to all of this was a simple modification to my build: Make 2nd and 3rd factories before cc. You don't lose income to them until after they make addons, which means you can afford your cc at only a linear mineral-setback. I was getting +1 before facts still, and ghost acad after cc. Still no port. I was getting 2 techlabs 1 reactor on my 3fact-- same production capacity as with my old build, just a little sooner, and to protoss' observer it probably looked much more troublesome than a 3rd cc on 1fact. They had to think about their defense potential after seeing pretty fast 2/3 fact (no port and raven delaying them). Earlier hellion production gives either some accumulation, or some increased harass/scout potential. Faster tank production lets me easily defend my 3rd from 2base aggression. I was even getting 4/5 fact (both techlabs) before +3 for even greater production capacity. I figured I needed to max asap to ensure pf and expo securing.

As far as actually maxing goes, turns out that doesn't work well. I wasn't actually getting any supply lead, and it was at the expense of ghost production, expo timing, and oc count. But it did let me safely secure pfs. So in what it was trying to do it succeeded. However it was hurting my lategame in the way that pure tank/hellion does. My gas was being sucked up into only marginally useful units, and I didn't have a good hellion count of course.

So notably what it didn't accomplish was stop carriers. My still fairly insignificant hellion count wasn't giving me the information needed to realize my opponent was going carriers in time to make bcs. The production capacity still only facilitates defense really, so I can't guarantee to have stopped him. The problem with focusing my build around defending immortal/stalker is that it made me even more vulnerable to protoss taking advantage of my passivity, as described earlier.

Next thought was getting a 4th fact before 3rd cc for a 2nd reactor to get hellion production going. This worked better, but didn't quite leave me feeling satisfied. This was the best build by far since my old build though, and probably is viable.

Long story short I came up with a new idea, a modification of an antiga build I'd found unsuccessful a while ago. 2port 2reactor 1techlab thor ghost. Well having made myself more comfortable with tanks, I replaced the thors with tanks. This build does a few things my 4fact build doesn't (in light of this build, maybe the 4fact build would be better doing 3fact 2 reactor cc fact)-- namely gives raven production. If stalkers are part of the problem, why not make myself more prepared for them? Ravens are useful no matter what anyway and have a lot of utility. However it simply wasn't efficient to make them before-- gas needed to be devoted to ghost ups and production.

The build itself can open with whatever gas opening you want. I still use my thor expand (modified to get siege/tank instead of thor after cc if I scout nexus, though thor after cc is still viable) personally. Reactor hellion or hellion marauder are both good openings. Cloaked banshee might be a good opening, but I find it delays cc too much for having not that much initial potential. Numbers need refined, but get 3rd gas when you land cc, make 2-3 tanks and start +1 attack, then 2port, then another tank, then 2banshee cloak and lift fact to reactor. Make 3/4 banshee and then start cc. 2/3 facts shouldn't be needed until after cc since you have 2port before cc instead. Cut hellion production as is useful if your opponent isn't taking a 3rd.

+1 asap is still important, because +2 is very important. +3 is actually relatively unimportant if you aren't relying on defending with a relatively small number of units backed by relatively low production capacity. Ghost ups/production are still necessary asap after facts after cc. Sooner you accumulate energy and get moebius reactor, cloak, and nuke, sooner you can harass with nukes. If gas is tough you can stop at 4 ghosts until your income allows better production.

Upgrades: +2 needed for tanks to 3shot stalkers, +1 needed to 4shot emp'd immortals, +3 needed to 4shot zlots but splash will kill at +2, +3 needed to 6shot full-shield archons but splash will kill at +2, +2 doesn't help tanks vs colos relative to +1 but +3 6shots-- emp helps a lot; +2 lets thors 3shot stalkers, +3 lets thors 2shot zlots after splash, +3 lets thors 3shot immortals after emp, +3 lets thors 5shot full-shield archons, every up helps hellions.
So get +2 as soon as possible, and get +3 before any other upgrade.

After 3/4 banshee, make 5th banshee and raven, possibly keeping up these production cycles. Depending on the game circumstances further banshee production may not be necessary. I don't think 3rd techlab fact is necessary, or at least not until later in game.

So why is 2port banshee production ok, if 1port wasn't? Well by every moment in game I have more banshees than I would have otherwise. This is a more serious threat to protoss. They have to react to this with increased stalker/cannon/obs production. My 2reactor hellions will quickly exacerbate the situation. I still don't need to get kills. Forcing warp-ins is damage alone. What these units really give me is constant scouting of protoss. He can't go carrier because he's spending money on defense, and I'd see it.

Additionally compare my build to lyyna's. Instead of a 2nd fact I'm getting a 2nd port. Instead of a ghost I'm making a banshee, instead of +1 armour and ghost acad and a thor I'm getting cloak and a tank. By getting 4 tanks instead of 3 thors I save 100 gas in the same amount of time, and 300 minerals-- which pay for the next 2 banshees. If 3 tanks suffice, I'm even better off. My hellion/banshee are a deterrent from aggression anyway, and can defend if needed.

Now let's look at how this build addresses the vulnerabilities of the previous. I already mentioned that carrier rush is impossible vs this. Immortal/archon shouldn't have a particularly strong timing window after my harass. Immortal/stalker will also be set back. If his initial stalker count is too low as he moves out with his immortals, I can start killing them before the fight happens by engaging with a pdd. My heavy hellion count will help vs his immortals as well-- their splash and body-block and being-light are all good at supporting tanks against immortals. Colossi are no threat while I keep my banshee count high. Reactive phoenixes won't be useful either-- I'll have marines to defend initially, and can just make a thor to be safe. Trying phoenixes may even make himself vulnerable to an attack. If the phoenix numbers get out of control somehow I can just put my 2ports on reactors and my facts on techlabs for viking/thor production as is needed. Under cover of my harass I should be able to push pfs, secure expos, build up my ghost count/energy, and add nukes to the mix. What all I can accomplish will come down to how well I'm using my units, as it should be. I'm not sitting at my pf hoping protoss doesn't do anything scary. What beats this style? Superior play of course, as it should be. If my opponent can match my multi-tasking defensively and then raise me his multi-tasking offensively it'll be on me to defend his harass.

Your goal is to stay alive, push pfs to secure expos, build a good oc count (13 is the most you'll ever need on most maps, fewer depending on the scenario), outmax protoss and start tearing him apart with your superior efficiency/army-size/units. I would rather actually kill my opponent if possible than to wait for him to exhaust all resources futilely.

So how do bcs fit into this? Currently I see them as being a coup d'etat rather than being the actual goal. You make them to finish your opponent if you can't otherwise end the game. If you can't guarantee to win without them, make them. I'm also going to look into some timing after which I'll just make 1 bc at a time to start the accumulation process without hurting my army too badly.