EPTAll done! The tables are pretty ghetto, but I'm happy how they came out.
In my previous post (link), I described the process of breaking down the TvP matchup into a set of definable skills, which were ranked in terms of their; a) ability to introduce a player to the flow of the game; b) relevance to a novice-level match; c) relevance to winning the game. I was able to come up with a list of four major skills, the most important being:
#1 – Learning one build order that is standard for this matchup, that transitions into the mid- and late-game well. (In my case, the FD build)
Today I wanted to share a method I’ve been using to approach #1 (learning the FD build), along with the results I’ve seen in my own games.
The Build, and the ‘Standard’ Method
FD Opening -> 5 Fact Timing Push @ 9:05
Details:+ Show Spoiler +
9 supply
11 rax
11 refinery
16 factory
16 supply
20 supply
31 cc
32 supply
34 ebay
39 2-3 turrets
46 supply
47 factory
50 refinery
50 armory
50 academy
54 supply
60 3 x facts
11 rax
11 refinery
16 factory
16 supply
20 supply
31 cc
32 supply
34 ebay
39 2-3 turrets
46 supply
47 factory
50 refinery
50 armory
50 academy
54 supply
60 3 x facts
Since I ranked ‘learning the FD build’ as my top priority, I’ve been spending most of my time working on that. For a reference, I’m using Stylish’s 5 Fact Push Build (link), which uses a standard FD opening followed by a timing push at just over 9 minutes. Though I intend on learning other types of openings and mid-game timings in the future, I thought learning a 5 Fact push would be a good starting position, as it can be scaled back to 4 Fact or lengthened to 6 Fact.
I began practicing, thinking that the easiest way to learn the build was to open up a local melee game, play through until the push (9:05), then repeat. This way I could at least learn the ‘form’ of the build and hopefully drill it enough to be able to use it fluidly in-game.
Using this method, I was able to learn the order of the build fairly well, but when I tried applying it to actual games I found myself out-macro’d by the Protoss almost every time, even if all conditions were right (i.e. Toss taking third expansion fast). Looking back at my replays, it wasn’t hard to spot the problem: my push was coming out at an average time of 11:07, instead of 9:05.
New Thinking
My initial response to discovering this delay was to practice more, in the same way, but this time with a timer (ADVLoader with the APM Live plugin – APM off). This way I could have a target to work towards, and could keep track of my progress.
But then I thought: “Well, 9 minutes is quite a long period of time, and I really don’t want to practice in 9 minute chunks. Can I cut this down somehow?”
Followed by: “You know, I think my timing is pretty good until about 20 supply – so my problem must be somewhere else.”
Then finally: “If I break this build down into segments, I can determine exactly where my timing is off, and focus on improving that part”
New Method
1 – Using a replay or VOD of the selected build, record the timing of two-three key actions before the large action (push).
I decided to go with the following:
Marker................Stylish
---------------------------------------------------------------------------------------------------------
Interval A...............---
31 CC...................4:36
Interval B...............---
50 Gas..................6:32
Interval C...............---
60 Fact.................7:24
Interval D...............---
Push.....................9:02
* Note: that the ‘Interval’ rows represent the amount of time lost in each interval.
I chose these markers since I felt they were a) major events in the build and b) since they were spaced out well considering the number of actions within each interval. The later intervals contain more actions, and therefore I set my markers closer.
2 – Using replays of real games and/or practice builds, record when you hit each of those key actions.
It’s best to choose games where you weren’t cheesed (obviously), and where you felt you worked straight through until the push without any huge disturbances. A larger sample of replays helps to determine more precise time that reflects your current ability.
Marker................Stylish..............My Old Avg.
---------------------------------------------------------------------------------------------------------
Interval A...............---......................-0:21
31 CC...................4:36....................4:57
Interval B...............---......................-1:03
50 Gas..................6:32....................7:56
Interval C...............---......................-0:38
60 Fact.................7:24....................9:26
Interval D...............---......................-0:03
Push.....................9:02...................11:07
* Note: that the ‘Interval’ rows represent the amount of time lost in each interval.
The way I approached this data was this: each marker represents the end of a build interval. What we need to determine is the interval(s) in which we are wasting the most time. Taking the differences in time between Stylish’s play and my own, it’s easy to see exactly where my timing was off.
- I hit ’31 CC’ :21 late (-:21 seconds for interval A)
- I hit ’50 Gas’ 1:24 late (-1:03 for interval B)
- I hit ’60 3 Fact’ 2:02 late (-:38 for interval C)
- I hit ‘Push’ 2:05 late (-:03 for interval D)
3 – Using a timer (ADVLoader with APM Live, or an alternative), play through the build as perfectly as you can, and save the game (in a separate slot) at each time marker.
You do this to create save-game positions which you can load up to practice a specific interval of the build. If you can’t pull off the build timings accurately yet, you might want to have a more experienced player help you. For most pushes it would only take about 10-15 minutes to go through the build and capture all the save-game positions.
I named my saved games: 5 Fact Push @ 4:36; 5 Fact Push @ 6:32; etc…
4 – Repeatedly practice the build interval where you are losing the most time, keeping track of the time you reach the next key action.
Repeating the specific segment of the build will allow you to focus your efforts where your timing is weak, while not wasting it where it is strong. It will also allow you to identify complex pieces of the build that were holding you up.
For example, I was losing the most amount of time with interval B, and when I started practicing it I realized why: within the same 10 second period at the end of the interval, you’re supposed to build a depot, a second factory, your second gas, a tank, an academy , an armory and SCVs from two command centers. Whoa.
My Results
I spent two practice sessions (approx one hour each) drilling both segments B and C, after which I played a series of games. Not all of the games allowed me to work through the build uninterrupted, but I was able to get a sample of four ‘smooth’ games. These aren’t just the games with the best results – I lost one of these, and almost fell apart after the push in another.
Marker................Stylish...............My Old Avg..........My New Avg.
---------------------------------------------------------------------------------------------------------
Interval A...............---......................-0:21...................-0:21
31 CC...................4:36....................4:57....................4:57
Interval B...............---......................-1:03...................-0:24
50 Gas..................6:32....................7:56....................7:17
Interval C...............---......................-0:38...................-0:13
60 Fact.................7:24....................9:26....................8:22
Interval D...............---......................-0:03...................-0:04
Push.....................9:02...................11:07..................10:04
I was able to significantly reduce the time I lost during both intervals B and C, and most importantly my overall push timing was reduced by almost a full minute. It’s also worth noting that this was my result in-game – in practice I can now consistently hit a 9:05 push (+/- 5 seconds or so).
Some Questions
Though I presented this as a scientific report of sorts, there are quite a few factors which I didn’t address. I thought I’d try to deal with a questions I thought of, even if only theoretically:
Q: How does the practice of individual segments of a build help you perform that build as a cohesive whole? Won’t there be hiccups during the transition periods between each?
A: My personal experience is that performing the build in its entirety does become smoother after this method, though you will experience instantaneous moments of ‘whoa – what do I do now?’ at the transition points. But those are insignificant losses (maybe 1-2 seconds) compared to what you lose during the segments.
I’ve also considered shifting the time markers in later practice sessions, in order to remedy the ‘whoa’ transition moments. We’ll see how that works out.
Q: How do you actually practice the segments? How do you take into account in-game distractions and multi-tasking?
Hopefully I can cover this in a future post. There is a lot of room for experiments here too, and I’ve been working on a few.
Conclusion
Breaking down the build like this helped me improve my push tremendously in a very short period of time. I will continue to practice, of course, since there’s still a lot of room for improvement. But hopefully by using this method I’ll continue to see positive results, both in practice and in my games.
I hope these experiments are helpful, even if just to help players think differently about their training methods. Thanks for all the feedback, and I hope to continue posting as often as I can.
Cheers!
Denotate
p.s. Sorry for the screwy formatting on the tables. I'm new to posting with BB Code and don't have somewhere to host images. Thanks to Stylish for his great builds and VODs!
