By Plexa, Saracen and Infinity21
Being able to analyse your own replays is an essential skill for improvement. Being able to identify your mistakes is the first step towards correcting them, and there’s no better way to do that than reviewing your replays. Unfortunately, analysis of replays where you won generally won’t yield as much useful information as analysis of replays where you lost. You can still get some important information out of them, but generally speaking you will want to focus on your losses if you really want to improve. This article will teach you the basics of analysing your replays. This is a very long process and takes quite a bit of time when you first start doing it, but the rewards will quickly pay off if you stick with it! In time, you'll be able to glance at a replay and analyse it in no time at all.
Step 1: Save your replays
This is a step that can be often overlooked because of the inbuilt replay autosave feature. However, after a certain amount of time those unsaved replays will delete themselves. It is good practice to save every single one of your replays - even if you don’t intend to watch them - you never know when they might become useful. The easiest way to do this is to download SC2 Gears by Dakota Fanning. This can be configured to automatically save your replays in whatever format you please. Here is a mini tutorial on how to set this up:
1) Open the settings tab on the menu
2) Select the “Enable Replay Auto-save” option
3) Open “Miscellaneous Settings”
4) The first tab should be the Replay Auto-save configuration, hit the edit button
5) You will be greeted by a few menus and some options to include in your replay name. You can configure this however you like. Use the “Test on last replay” option to make sure you are happy with the result. Here are a few basic options:
- /F4 /p1 /p2 /T /m - this generates a replay like 0324 LiquidNazgul LiquidJinro PvT Lost Temple
- /d/p1,/p2 /r1/r2 /m - this generates a replay like 2010-01-06oGsVINES,LiquidRet PZ Metalopolis
This is the auto-save configuration that Ret uses
6) After you’re happy with the formatting, hit “Okay” to return yourself back to the Misc. Settings menu.
7) Choose whether you want SC2 Gears to delete the replay in the “Unsaved Folder” after auto-saving it and whether you want SC2 Gears to open your last replay for analysis
8) Go to the “Folders” menu (number 7) and specify where you want SC2 Gears to check for new replays. A good idea is just to make it check the “Unsaved” folder in your Account directory.
9) Hit “Okay” and viola, every time SC2 Gears is running it will automatically save your replays. You can also start SC2 via SC2 Gears so you will never forget to have all your replays saved.
Step 2: Give it a ‘once through’
As you improve, analysing replays should take less and less time. But if you are just starting out, watching your replay through once is the first step to properly analysing your replay. Sometimes the reason you lost can be easily identified such as
- Not scouting a hidden expansion
- Horrific micro mistakes (e.g. 1 baneling killing 12+ lings)
- Missing a baneling bust
- Not macroing properly
There is a myth that is commonly believed by higher level players, and that is that macroing better is the be all and end all of improving at lower levels. While better macro would have won the game, that doesn’t mean there isn’t valuable information to be gained from deeply analysing your play such as proper decision making. Understanding why your macro failed you and identifying ways to fix it can all be understood via replay analysis and will lead to quicker improvement. Further, other mistakes you make should be fixed as you are improving your macro - this leads to a healthier overall game
Never be content with just identifying your mistake, you also need to be able to correct it. That will come later. Other things you should be looking for in your first run through should be basic things like unit compositions, how your worker count compares to his, keeping an eye on upgrades, keeping an eye on your army worth etc. Here are a few basic guidelines for these:
It’s best that you use the Unit Counting station rather than the resources tab to count this, since the number of harvesters in the resources tab has a small delay on it - and the resources tab doesn’t display the number of MULEs the Terran has.
Protoss vs Terran - MULEs translate roughly to 4 SCVs. An even worker count would be the number of Probes = number of SCV + 4 per MULE (or Orbital Command)
Protoss vs Zerg - Since Zerg have non-linear growth, an even worker count is roughly number of Probes +- 3 = Number of Drones.
Zerg vs Terran - An even worker count is roughly number of SCVs + 4 per mule = Number of Drones +- 3
Keep an eye on this when you engage. Generally speaking, the army with greater worth will always win (given the right unit composition). However there are a number of other factors which can influence the result of the battle (e.g. positioning, upgrades, micro). Note any battles you lose where you have superior army worth - these battles should be reviewed on your second play through of the replay until you understand why you lost the battle. Also, if you lose a battle when you have inferior army worth then that’s why you lost the battle - you didn’t have enough units.
Step 3: Understanding the game
After your first run through, you will have identified some of the basic areas where you need improvement. Your goal is now to understand how to take this information and correct your mistakes.
Inferior Worker Count
Say one of your problems was that you were behind in workers. Take a good look at your replay and keep an eye on your CC/Nexus for any breaks in peon production. If you are Zerg, you need to see if there are any times where it is safe to build drones but you were making army instead (or nothing at all). Juggling between drones and units is a key skill for Zergs and understanding when to drone or when to make units can be done through analysing replays and identifying signs for when it is safe to drone and when you need to make units. Don’t get distracted by the battles going on in the replay, and note whether you drop peon production during battles or times of micro. Try to think of ways for you to remember to keep producing these workers during these micro intensive situations (e.g. before you engage add a round of workers to your queue).
Poor Worker Saturation
You need about 30 peons per base - 6 to mine gas, 24 on minerals. If you have 90 workers and 2 bases then you have 30 workers too many and desperately need an expansion. You can check how many workers you have mining minerals by selecting all of the peons that are mining them (be careful not to select any that are mining gas!). If you have 3 full rows - then you’re at perfect saturation for that base, assuming you already have 6 workers on gas. If you have 60 workers but you have 40 at your main, 20 at your natural - then you have poor worker saturation and should have transferred 10 more to your natural. This is a very easy mistake to make and even progamers like oGsMC often have too many probes at one base. This can be fixed by selecting your workers in game and checking to see whether you have good saturation or not.
Lost an equal army worth/greater army worth battle
Look long and hard at the battle. Were you behind in upgrades? Was your unit composition bad? (e.g. pure hydralisk against 8 Colossus) Did your opponent have a better position? (e.g. his concave was better, so he had more ranged units firing at any given time. e.g. you weren’t able to get a surround on his Zealots with your Speedlings because he was holding a choke). Were you focus firing in small ranged battles? (very important in PvP). Did he micro so that he took out key units? (e.g. making Immortals target Tanks/Marauders and not Marines). Were your units falling over each other? (e.g. Marauders stuck in the middle of a Marine/Marauder ball against Zerg - leaving them unable to soak baneling hits, or letting your Stalkers get in front of your Immortal/Sentry/Zealots and screwing up their AI).
Try to work out what exactly went wrong, and then try to come up with a way to fix it. For instance, if you engaged his army poorly, know that next time you get into a similar situation to pick where you fight better. Pull your army back a little bit so you have the positional advantage instead of him, etc.
Losing to an early pressure build
Make sure the build you are using is one that you know is good. You should have most of it planned out from before the start of the game AND you should have planned out what to do if you see him doing something you didn’t expect. If your build has excessive minerals/gas stocking up for no reason, then your build needs refinement. It’s often a good idea just to watch progamer replays/VODs and copy their builds until you become more comfortable with the game. Yes, it’s not much fun, but it provides a solid base for you to work off of.
Failed timing attack
If you cut workers since you were going for a timing attack which failed, you should take a long hard look at what happened in the game. Particularly noting the build that your opponent used and trying to understand why the timing attack failed. Perhaps if you had hit 30 seconds earlier your attack might’ve been successful, perhaps you needed to wait for a round of units, perhaps his build was a straight up hard counter to what you did. The next time you see that build try out your new refined timing attack and evaluate it’s success, and rinse and repeat. Also try to find a good timing or indication that you need to resume peon production, since if you are not going to be able to kill him with your attack you’re going to need to have a good economy to compete well in the midgame.
Special Analysis Techniques:
There are a variety of techniques you can apply to successive rewatches of your replay. These can be used to further refine your play or work on specific errors in your game.
Analysing one region at a time
Watch the entire replay with your screen fixed (or close to fixed) on a single area, specifically places like your natural, key fight locations or areas of strategic importance. By forcing yourself to watch one place for the entire replay you are able to notice far more than if you are jumping around everywhere. Say you focus on your natural for the whole replay, then you are able to see when your natural went up, how well you saturated it, how quickly you saturated it, whether it was prepared for any drops and if there was any way you could have better prepared for those drops and even really simple things like how quickly did you get your expo up. By focusing on one region you get an unparalleled opportunity to analyse exactly how you used that section of the map which can often pick up both small and glaring errors in your play.
FPView can be used as a tool to see what your opponent or yourself could see at any given time. This is particularly useful when analysing high level replays and you are trying to master the build used in the replay. Here is an example, by watching Mana play his 3gate Robo build against a Terran you are able to see what he sees - specifically they tells he is able to pick up by scouting. Mana won’t play a poor build on purpose, he will see what his opponent is doing and react accordingly. By watching his FPview you can see the information he sees, so when you see that same information you can play the same response Mana did (assuming he won, of course!). You can also do this with your own play and see if there was anything which came into your vision which you should have noted and to see any holes in your macro (and where those wholes occur).
Using the production tab
The production tab is one of the basic menus on offer when watching a replay, however it is often ignored as a way to analyse your play. You can actively watch the production tab throughout the game to make sure you are building something at every point in the game. When you see gaps in your production, pause the replay and either check your FPView or what’s going on in the replay to find a reason why there was a gap. Perhaps you didn’t have any money because you just expanded, or perhaps you were just being lazy and missed a round of units - in either case you need to identify and fix any gaps in your production.
Step 4: Refining your game
With the big mistakes taken care of, you can look at the more subtle details of the game and refine any mistakes there.
Refining your build
Detailing exactly how to refine your build is beyond the scope of this article. Fortunately, Infinity21 has written an excellent guide on how to design and refine your builds (http://www.teamliquid.net/forum/viewmessage.php?topic_id=202592). However, when you are watching your replays you will be able to observe your build in action and be able to evaluate it’s effectiveness. You’re looking for things like having good timing with your warpgate, good timing on your Larvae Injects/Queen transfers, good timing on your addon completions, not having unnecessary resources lying around and so on. You should also be mindful of the build your opponent used and any timing windows in his build that you would be able to exploit with the build you used so that next time you come up against that build you can pressure at better times.
If you can get a good read on what your opponent is doing then most of the time you should be able to win by playing the counter build to that. Being able to get a lot of information from small tells is an essential part of Starcraft 2, and indeed makes this game very similar to Poker. There are two ways to refine your scouting:
- See what you saw in game and whether there were any things you saw that could have been indicators of what your opponent was up to:
-- e.g. I didn’t see him with many drones and he had his gas, but he had 6 lings denying scouting, perhaps he’s baneling busting
-- e.g. The Terran made only marines and a bunker at his ramp in TvP, this is a good sign that he is teching
-- e.g. The Protoss isn’t spending his Chronoboost, perhaps he’s 4warpgating
- Note the timing of when things should be happening and in future games send a scout at that time to see what’s up.
-- e.g. after a Protoss FE vs Zerg send a Probe or a Zealot to scout whether he’s going for a Baneling bust or Roach pressure
In general, when you send a scout make sure you’re sending it LOOKING for something. Too many players just send a scout and say “oh I scouted” - that doesn’t cut it. When you scout you are looking for tells in his build, worker count, unit composition etc to tell you what his next move is likely to be - and then you need to know how to react to it. For instance, you scout early in PvP because you need to see if he is going for any kind of Proxy, Cannon Rush or Korean 4 Warpgate and to see how his Chronoboost is being spent pre-stalker. All of this is incredibly valuable information which tells you how the rest of the game is going to play out. If you are unsure about reading your opponents build, you can ask in the strategy forum for advice or try and get a hold of some replays of the build in question and look for things that give it away (e.g. first addon is a Reactor not Tech Lab).
Picking your battles
As you watch the game, notice where your armies engage. Are you fighting in positions which are good for you? For instance, in PvZ you want to be engaging in choke points to increase the effectiveness of forcefield, conversely in ZvP you want to be engaging in wide open areas so you can get more units firing at once. Try to find any common theme in your poor engagements, and then try to correct it - remember, it is better to delay the fight and engage in a better position than to force the issue and waste your army.
Map awareness and map control
During the midgame try to evaluate the position every so often. Try to judge who has the initiative/map control and whether the player was able to capitalise on it or how the player lost it. Did he over commit to a battle? Did his macro slip up? Understanding how players lose advantages like this can be key during a game since you may be able to force him into a situation where he makes one of those mistakes (e.g. pressuring on multiple fronts to force bad macro or baiting him into engaging in a poor position).
Another things to be aware of is whether or not you knew what was going on on the map. Many players like to sneak expansions and due to poor scouting habits they often go unnoticed. As Zerg, are you keeping your Overlords above expansions? Are you regularly checking his base with Overseers? As Protoss, are you using observers to their full potential? Are you regularly checking his army composition with your observers? Are you checking for hidden expansions? Are you making good use of Hallucination in PvZ? As Terran, are you using your unused buildings (i.e. Factory TvP) wisely? Are you regularly checking expansions with units? There are many things to think about optimising your map awareness, these are all things which can have tangible benefits in game.
Some questions to ask yourself about your macro:
Are you always chronoboosting something when you have the energy or saving up chronos for a certain timing?
Are you always dropping MULEs if you don't need to be scanning?
Are you always spawning larvae?
Are you always spreading creep?
Are you always making workers (P or T)?
Are you droning as hard as you can when no pressure is being applied (Z)? (A nice thing about watching replays is you get to see exactly how much stuff your opponent has so you can figure out for yourself exactly what you should be doing).
Do you ever float minerals before you finish your build order?
Do you ever get supply blocked?
Do you have too few production facilities? Too many?
Step 5: Reflection
Now that you have analysed your replay in detail, it’s a good idea to take a minute to recap what you’ve analysed. This helps improve your understanding of Starcraft by letting the information sink in which makes it more ingrained into your memory. Eventually, it will become instinct or second nature to you and you will notice the vast improvement in your play. If you are unable to work out how to fix/improve certain areas of your play, post your analysis along with your question in the strategy forum and people will be more than happy to help you understand.
We have tried to keep this guide as general as possible. There are certainly things that each race should be looking for in replays and thinking about while watching. For example, Zergs should be thinking about ways to gauge the opponents army strength - weak armies mean drone, strong armies mean units. These things, while invaluable, are out of the depth of this basic introduction to replay analysis. Perhaps a more advanced guide will be written at a later date.
Further Advice on Developing Build Orders by infinity21 - http://www.teamliquid.net/forum/viewmessage.php?topic_id=202592