----------------------------------------------------------
Happy Vs Zanster Game 1. Effective usage of Ramps to hold pressure from zerg. There is a way to beat ultras!!!!
________________________________
First off, for the choice of map pool, I question a bit of blizzards decisions. They seemed somewhat extreme. Terran did not perform well on Ruins of Seras nor central protocol going 1-16, and performed best on dusk towers going 6-4. Dusk towers has one attack path; Ruins of Seras and Central Protocol have large space outside the third and many paths to be countered. This is consistent with snowballing as terran where when an attack takes place the rally is moved with the army as is often done by the korean players TY and Bomber, and was only found on one foreigner replay with Bunny. The koreans also used the drops much more for scouting than doing damage. They did less damage with the first drop for the most part despite having higher win rates.
Now I notice that Terrans did not make it to the finals of dreamhack consistent with the theme of HOTS . In fact, according to the wikipedia page, terran had a 28.1% and 48.1% win rate against zerg and protoss, respectively. The game has changed significantly and the sample size is relatively small. Its hard to generalize so much. I watched every single game to try and figure out what caused the loss and determine whether some trends could be gleamed from the game. Throughout this I will always show win-loss records as wins first for terran and losses for terran.
I want to ask if terrans are losing for the same reason as hots. So let's start with the premise that terran has to end it early or its nearly an unwinnable situation. I'll look at the replays from dreamhack and see if this trend occurs. If it does, then I'll ask if this trend should continue and then ask how to fix it if this trend should change. Keep in mind that the low number of games sways this trend. (Particularly since bomber and zanster were in a lot of these games.) With a win rate at 6-5 in favor of terran over the other races in games lasting longer than 16 minutes it doesn't appear that this is a major issue at the moment and is a major improvement over HOTS.
The second are of interest would be looking at the units in each matchup. The cyclones were not used TvZ. Cyclones were a staple of TvP. In games with banshees, terrans went 5-3. (that includes a win vs protoss.) Hellbats saw little success going 1-2. Liberators in both TvZ and TvP were effective in numbers less than 12 but in both games where the count was higher terran lost. One of those was a TvP Happy vs PetitDrogo where tempests and storm destroyed high liberator counts. Another intersting statistic is the usage of the advanced ballistics upgrade on the liberators terran only won 1 game and lost in 4 others. Ultralisks had a lot of success except in the Happy games I'll discuss in depth later. Surprisingly vipers were not popular in the TvZ games. Ravager+infestor compositions were much more effective than the ravager only compositions hinting at an importance to transition to them. Disruptors went 1-3 vs Koreans and faired much better against the foreigner implying some bit of usage disparity. Adepts were more useful. Interestingly, warp prisms were hardly killed by cyclones. Given the lack of larvae increased queen counts seem to have made banshee and hellion plays less common and less effective.
TvZ disconcerting trends. I do want to say many of the games that terrans lost were 4 scenarios. First is high reaper count openers. (6+) This build did not appear to carry any weight. Given the little success of the build its surprising it was used in 5 of the games with only 1 win. The second area where high number of loses occurred were to ling ravagar and roach ravager midgame timings. Interestingly when tanks were not built holding the push effectively worked 4-1 where nearly every time there were high tank counts the pushes were not held.
The third area where a high number of losses were seen were 3 gas nydus games. Games which were ling bane muta went 3-1. Two styles did fairly well vs roach ravager, fast banshee and aggressive multipronged drop play. Interestingly enough, the tank play did not fair very well vs roach ravager. In every case, once 2 tanks died against ravagers the game basically ended in a loss for terran. Also, pure bio did beat roach ravager. Tank drops were only done in 1 game as an opener which resulted in a loss for bunny. There were no mech games.
TvZ promising results. Many of the games were biomech, or marine tank. Happy did show a semi-mech style vs Zanster. If you are a mech player who is struggling in lotv, you should probably watch his games. They were probably as close to mech as were in the tournament. I'll discuss them a bit more in a revised version of the post with a table of games. Also, zerg drops did not work in either or the 2 games that they were tried by Elazer. Terrans struggled significantly to scout hives. When the player scouted the hive while building also the win rates were high 3-1.
Warp prisms speed seems to be off, it moves as fast as a speed medivac all the time. This seems too fast especially with pickup distances and make its extremely difficult for terran to catch them causing several snowballing scenarios in TvP. It appears that there is almost no risk at all to using warp prisms and disruptors in the dreamhack games. There still are mechanical issues to be addressed. The building lifting caused problems in 2 of the games where the player was trying to land the building and instead all of the others lift off. Also TY and Bunny appeared to have problems with the new alt-separation technique. I don't use it often enough to know if that's an actual problem.
TvT the player defending and who was more greedy had the bigger advantage in general which is in contrast to what was seen in hots. Almost none of the TvT games had vikings and were purely marine tank. I'm really not sure at all why this is. Also, the player who build more sensor towers won every single game. I'm not sure if this is just coincidence, but a similar scenario came up where games with sensor towers vs zerg were also much more effective with 3-1.
Thus, with the premise that terran cannot win into the late late game, it does not seem to be the case here given the statistics. (Notice though that a lot of the statistics come from very passive games, or a terran lead) Terrans struggled to get to the late game, and lost to certain builds everytime(3 gas nydus). Also, terran performed poorly on the maps you would expect. (small drop area, many counterable paths) The most difficult thing to deal with TvP was warp prisms. (I heard they are already nerfing it) There are some obvious problems. Alt separation doesn't seem to work well, and multiple buildings lift for an addon or landing. The liberators have an obvious weakness of having to move and siege which means almost mandatory use of sensor towers, which i believe would benefit liberators more in the late game.
I can post standard builds if people want them. There were 3 different common ones use TvP and 4 different ones TvZ.
----------------------------------------------------------
I think it helps to have some pictures to show some of the things that happened during the games.
Liberator usage (maybe advanced ballistics isn't very good right now)
+ Show Spoiler +
Happy Vs Puck Game 3- Effective liberator play.
Happy Vs Petit Drogo Game 1- Effective usage of liberators to force an engagment into a choke.
Happy Vs Zanster Game 1- Run by positioning. Not effective for team fights.
Bomber vs Petit Drogo Game 1- Uneffective Liberator positioning.
Building Lift off Problem-
+ Show Spoiler +
Bunny Vs Majestic Game 2- Serious problem where buildings lift when they shouldn't.
Tanks prove ineffective to Roach-Ravager pressure-
+ Show Spoiler +
Happy Vs Firecake Game 2- Tanks prove ineffective vs Roach-ravager pressure.
Psuedo-Mech, biomech from Happy against Zanster to beat ultraviper compositions
+ Show Spoiler +
'
Happy Vs Zanster Game 1. Effective Usage of Ghosts and ramps to channel Ultralisks.
Happy Vs Zanster Game 1. Effective usage of Ramps to hold pressure from zerg.
Happy VS Zanster Game 3. Usage of Ghosts Sensor towers and Liberators to hold ultra viper builds.
Happy Vs Zanster Game 1. Effective Usage of Ghosts and ramps to channel Ultralisks.
Happy Vs Zanster Game 1. Effective usage of Ramps to hold pressure from zerg.
Happy VS Zanster Game 3. Usage of Ghosts Sensor towers and Liberators to hold ultra viper builds.
The importance of sensor towers in TvT and TvZ
+ Show Spoiler +
Happy Vs Zanster Game 2. Forgetting to build Sensor Towers Results in a loss.
Keep your depots up!
+ Show Spoiler +
Lilbow vs TY Game 2. Keep your depots up vs Adepts
Simcity to build chokes vs ling-ravager
+ Show Spoiler +
TY vs TLO Game 2. Using a simcity at a third to deal with ling ravager builds.
---------------------------------------------------------
Shortened summary-
---------------------------------------------------------
1. Avoid tanks vs ravagers. Tanks did not appear so effective given ravager side(lack of splash) and the corrosive bile.
There are interesting articles about Ravagers, ravagers killing mech, and tanks here.
2. Reaper builds(6+) not super successful in tournaments. Practice alt seperation.
3. Rally army forward when attacking to snowball. its better to scout with a drop then fight with it.
4. Ghosts and Liberators together with chokes prove useful for dealing with ultralisks. The Liberator upgrade does not appear useful.
5. Drop play is effective at dealing with roach-ravager. Banshees are not as useful given higher queen counts. Given the map pool not have much room for drops, koreans seem to not be attempting as much damage and scouting more.
6. Pray Zerg's don't nydus.
7. Target infestors. Lings and ravagers do not work as well without them.
8. Don't be afraid to build sensor towers.
9. Scouting for a hive will save your life. See one coming go fight.
10. Don't let adepts in. You'll never catch the speedy warp prisms. Watch your army to split against disruptors.
And last but not least- Balance doesn't look so bad other than nydus and warp prism speed.
11. Hope the Map pool changes soon