Starcraft AI blog

Locutus’s game records are in almost the same format as Steamhammer’s and can be summarized by the same script. I expect it will also work for DaQin and AITP.

Where Steamhammer was set to keep 100 game records per opponent, Locutus was set to keep 200. Since there were 100 rounds in the tournament, game counts over 100 mean that pre-learned data is included in the table alongside the tournament data. If Locutus was not trained on a near-final version of the opponent, then the two could be significantly different.

After seeing a few Locutus-PurpleWave games I got the impression that PurpleWave reacted adequately to Locutus’s trick strategy of cannoning the ramp and then dropping zealots. So I was surprised that Locutus considered it the best choice. But the overall win rate is high compared to the tournament results, so I suspect it is influenced by pre-learned data from games against a weaker version of PurpleWave.

The numbers in the “recognized” columns of the plan table show how widely Steamhammer cast its net for a solution to Locutus.

Locutus never tried to steal the gas of a zerg. Objectively, that makes sense. In the context of bot play, I’m not so sure; many bots of all races mess up their builds in the face of a gas steal.

Locutus declared an enemy-specific strategy against Iron. I’m not sure why it also had pre-learned data.

Proxy DT rush. That tends to confirm my picture of XiaoYi as vulnerable to tricks.

Locutus configured an enemy-specific strategy against UAlbertaBot. Openings other than CannonFirst4GateGoon are from pre-learned data, which was ignored in making the opening decision.

Today is Steamhammer. With a mid-rank finish and the widest range of builds, plus informative game records, Steamhammer may give us the best insight into how other bots played.

The tournament was 100 rounds, and Steamhammer was configured to remember the previous 100 game records, because in play there is no reason to remember more (earlier records are increasingly discounted). Steamhammer also had pre-learned game records for many opponents, so when the game record count reached 100, new records added caused old pre-learned records to drop away. Not all 100 tournament games happened for each opponent, but the pre-learned games filled in the small gaps so that Steamhammer ended up with exactly 100 game records per opponent in every case.

The “opening” table counts Steamhammer’s opening choices. The “plan” table shows the plan that Steamhammer first predicted that the opponent would play, then recognized that the opponent was playing. Both prediction and recognition can be wrong. The timing table is new this year, an attempt to get a little more information out of Steamhammer’s rich game records. For some events, it gives the count of games in which the event occurred, and the median time, earliest time, and latest time it occurred in those games when it did. The times are given under the assumption that 1 second of game time is exactly 24 frames, a simplification.

• gas steal attempt - When Steamhammer sent out the drone to steal gas (if it did).
• gas steal success - Whether the gas steal attempt succeeded in taking the opponent’s gas. Steamhammer doesn’t record the time it happens, so this is only a success count.
• enemy scout - When the enemy scout first reached Streamhammer’s base.
• enemy combat units - When the first enemy combat unit was seen.
• enemy air units - When the enemy is first known to have tech for flying units (except overlords).
• enemy cloaked units - When the enemy is first known to have tech for cloaked units.

It looks like Locutus opened forge-expand every game. It worked. Steamhammer desperately tried everything, including ZvZ builds and island builds, and finally squeezed 3 wins with a risky extreme macro opening, 4 hatcheries before spawning pool, which was able to win one game in five. I should add 5 and 6 hatch before pool and see if they help.

Locutus rarely made corsairs or dark templar. I wonder what its criteria are? Maybe it won before it got that far. The scout was usually quite early, and the first combat unit was seen late, as expected for a cannon-first opener.

I played over the 3 wins. They were in rounds 65, 70, and 73; after that, I expect that Locutus found a way to win. In 2 games, Steamhammer pulled ahead in early economy with its greedy opening, then struggled to defend and fell into a losing position. But Locutus got most of its units stuck in its base, and Steamhammer was able to turn it around and win after a hard fight with critical defiler support. In the third win, Locutus chose a zealot-archon unit mix that Steamhammer knows how to cope with, and zerg powered through.

PurpleWave in contrast went with mostly 2 gate openings against Steamhammer; that’s what “heavy rush” means for protoss. Steamhammer countered with early sunkens plus hydras or, less successfully, mutalisks (this version had a bug that weakened mutalisk play). There are also 3 wins with a ZvZ fast mutalisk opening. 2 gates should beat that, so protoss either played poorly or chose a different build in those games.

BananaBrain contrasts with both previous opponents in that it played a variety of builds. Steamhammer was unable to predict what was coming. It looks strange that the best reaction was an opening designed to counter terran factory-first builds that include a vulture runby, but in fact it is a mildly specialized 2 hatch mutalisk variant and not so surprising. BananaBrain made corsairs and dark templar in most games.

DaQin played forge-expand and has similar timings to Locutus, for the same reasons. The fast scout is to allow adjustment of the cannon count and timing, and the late combat units are due to getting a gateway later. Steamhammer couldn’t find any better reaction than to try to bust with zerglings, either early or late, and it was not particularly successful.

ZZZKBot mostly played a turtle-into-mutalisks strategy against Steamhammer, and was somewhat successful. You can read the idea straight out of the tables above. The 2:25 earliest timing but 4:06 median timing for combat units says that ZZZKBot sometimes rushed zerglings, but usually not.

Microwave played a 9 pool speed build into expansion and then spire, which you cannot read out of the plan table because Steamhammer didn’t recognize it accurately. But in the timing table you can see that combat units (zerglings) were early and air units (mutalisks) were not late.

Steamhammer was not able to steal Microwave’s gas. It probably should have stopped trying.

Look at that huge range of scout timings! 0:34 means that the scout SCV was sent immediately at the start of the game and went directly to the zerg base. 15:09 probably means that no enemy unit got into the base until the end of the game when Steamhammer lost (Steamhammer is on BWAPI 4.1.2 and cannot detect scans). Steamhammer prevented the scout entirely in 13 out of the 100 games by its own count; 15:09 is probably the same. Steamhammer was not able to steal Iron’s gas, and did eventually give up trying.

Steamhammer liked 7 pool against XiaoYi, just as Microwave did, but also liked its dawn hydra rush.

Thanks to pre-learning, I expected Steamhammer to play its overpool turtle build every game. I’m not sure why it didn’t. I also don’t know how it hit on the 5 scout build, which means send out a drone at 5 supply to scout very early, then leave all decisions to the strategy boss. It’s a logical try against a random opponent, especially one that has a single strategy for each race, and it was fairly successful. But it did not appear in the pre-learned data.

AITP scored zip against both mass zerglings (9HatchExpo9Pool9Gas) and against fast mutalisks (AntiFact_13Pool). And it successfully scouted Steamhammer’s base only 25% of the time. If you don’t scout reliably, it will be hard to withstand rushes.

Steamhammer was not able to judge whether BunkerBoxeR was playing a proxy (with its proxy bunker) or a worker rush (since it sent SCVs in support). But it didn’t matter. The reactions are nearly the same. Since BunkerBoxeR never wants gas, stealing its gas was a waste.

Steamhammer knows 142 different openings. In the whole tournament, it was only able to try 111 of them! It tried the most openings versus protoss, since it was looking everywhere for an escape from the overwhelming top protoss bots. Most openings were tried only a few times and lost every game, which means that Steamhammer would have performed better without them. That’s expected and even intentional; my plan is to add smarts until it is able to make good guesses about what to try. The work is underway.

Microwave keeps its result files in the same format as UAlbertaBot: A file for each opponent, and in the file a list of strategies tried, each with win and loss counts. But Microwave independently restricts the win count and the loss count to not exceed 10. This amounts to intentionally forgetting older history when there has been a lot of it. The advantage is that Microwave adapts its strategies more quickly if the opponent shifts its play. The disadvantage, of course, is that information is thrown away. As a side effect, the numbers in the “total” and “overall” cells of my tables are not too informative.

This post looks at Microwave’s result_*.txt files for each opponent, since it’s what I’ve done before and I already had a script to parse them. This year Microwave also kept history_*.txt files with a record of each game. I could get a fuller picture of what Microwave did from the history files. I’m not sure whether Microwave uses the history files to make decisions. Still, if this is about “what Microwave learned,” then the result files are what Microwave learned, at least in large part.

Microwave has pre-learned data files for a number of opponents. Data from those files survived to be included in these tables. In other words, the tables here include not only tournament games, but in some cases preparation games played before the tournament.

Microwave’s author MicroDK commented that Microwave might have a bug in keeping its learning files, since the numbers did not always agree with official tournament results. As explained yesterday for UAlbertaBot, the player cannot always know what the tournament manager decides is the outcome of the game. Between that and the pre-learned data, I see reason to doubt that Microwave had a bug. But I didn’t look into details.

This year Microwave recorded a total of 32 strategies, compared to 19 last year. I tried to keep the tables tractable by breaking them down by opponent race, since not all strategies were tried against all races. Nevertheless, prepare to scroll right!

terran

It seems that most of Microwave’s openings worked about equally poorly against Iron, which is interesting and hard for me to explain. 2 hatch muta equals 3 hatch hydra bust equals 7 pool—among others? XiaoYi was vulnerable to rushes, and Microwave settled on 7 pool. After seeing the learning files of only 2 bots, I’m already getting a picture of XiaoYi as strong but apparently not robust against tricky strategies; UAlbertaBot chose DT rush against it. AITP and BunkerBoxeR were easy opponents and seem to have been vulnerable to the first thing Microwave tried, so that zerg never felt the need to vary.

The tournament had 100 rounds. Totals of more than 100 games versus an opponent, as versus Iron here, are a sign that pre-learned data was carried over. Microwave did not have time during the competition to try this many strategies so many times each.

#	bot	total	10Hatch9Pool9gas	2HatchHydra	2HatchLurker	2HatchLurkerAllIn	2HatchMuta	3HatchHydraBust	3HatchHydraExpo	3HatchLingBust	3HatchPoolHydra	4HatchBeforeGas	4PoolHard	4PoolSoft	5Pool	5PoolSpeed	6Pool	6PoolSpeed	7Pool	8Pool	9Pool	9PoolLurker	9PoolSpeed	9PoolSpeedLing	OverpoolTurtle	ZvZ_Overpool11Gas
8	iron	63-175  26%	1-5 17%	1-5 17%	1-5 17%	2-7 22%	5-10 33%	5-10 33%	1-5 17%	0-5 0%	1-5 17%	2-7 22%	5-10 33%	5-10 33%	1-5 17%	5-10 33%	1-5 17%	3-8 27%	5-10 33%	5-10 33%	3-8 27%	4-10 29%	2-7 22%	2-7 22%	3-8 27%	0-3 0%
9	xiaoyi	22-13  63%	0-2 0%	-	0-1 0%	-	0-2 0%	-	-	-	-	-	-	10-4 71%	1-1 50%	-	-	-	10-0 100%	-	1-1 50%	-	-	0-2 0%	-	-
12	aitp	10-0  100%	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	10-0 100%	-	-
13	bunkerboxer	10-0  100%	-	-	-	-	-	-	-	-	-	-	-	-	10-0 100%	-	-	-	-	-	-	-	-	-	-	-
overall		-  36%	1-7 12%	1-5 17%	1-6 14%	2-7 22%	5-12 29%	5-10 33%	1-5 17%	0-5 0%	1-5 17%	2-7 22%	5-10 33%	15-14 52%	12-6 67%	5-10 33%	1-5 17%	3-8 27%	15-10 60%	5-10 33%	4-9 31%	4-10 29%	2-7 22%	12-9 57%	3-8 27%	0-3 0%

protoss

Aha, Bananabrain had a weakness against Microwave’s 3 hatch zergling bust! I’ve seen the same on BASIL, e.g. Microwave-BananaBrain on Empire of the Sun, where BananaBrain was negligent in setting up the defense of its natural. In general, Bananabrain showed sensitivity to the opponent’s strategy; the other top protoss bots were more consistent.

#	bot	total	10Hatch9Pool9gas	2HatchHydra	2HatchLurker	2HatchLurkerAllIn	2HatchMuta	3HatchHydraBust	3HatchHydraExpo	3HatchLingBust	3HatchPoolHydra	4HatchBeforeGas	4PoolHard	4PoolSoft	5Pool	5PoolSpeed	6Pool	6PoolSpeed	7Pool	8Pool	9Pool	9PoolLurker	9PoolSpeed	9PoolSpeedLing	ZvP_10Hatch9Pool	ZvZ_Overgas11Pool	ZvZ_Overgas9Pool	ZvZ_Overpool11Gas
1	locutus	0-202  0%	0-10 0%	0-10 0%	0-10 0%	0-8 0%	0-10 0%	0-10 0%	-	0-8 0%	0-8 0%	0-8 0%	0-10 0%	0-10 0%	0-10 0%	0-8 0%	0-8 0%	0-8 0%	0-10 0%	0-8 0%	0-10 0%	0-7 0%	0-10 0%	0-7 0%	0-7 0%	-	-	0-7 0%
2	purplewave	26-154  14%	2-10 17%	2-9 18%	0-6 0%	0-5 0%	2-10 17%	2-10 17%	0-4 0%	3-10 23%	0-4 0%	0-4 0%	0-3 0%	3-10 23%	3-10 23%	0-3 0%	0-3 0%	0-3 0%	3-10 23%	2-9 18%	0-3 0%	0-5 0%	1-7 12%	3-10 23%	0-3 0%	-	-	0-3 0%
3	bananabrain	85-130  40%	1-3 25%	2-4 33%	2-5 29%	1-5 17%	2-4 33%	0-2 0%	1-4 20%	10-4 71%	1-4 20%	0-5 0%	6-7 46%	10-9 53%	10-9 53%	5-7 42%	3-5 38%	4-6 40%	1-3 25%	4-6 40%	4-6 40%	0-5 0%	2-4 33%	4-6 40%	0-2 0%	-	7-8 47%	5-7 42%
4	daqin	8-64  11%	2-9 18%	0-4 0%	0-4 0%	-	3-10 23%	0-3 0%	-	3-10 23%	0-3 0%	-	-	0-3 0%	0-3 0%	-	-	-	0-3 0%	-	0-3 0%	-	0-3 0%	0-3 0%	0-3 0%	-	-	-
10	mcrave	95-78  55%	2-2 50%	2-3 40%	5-3 62%	2-3 40%	10-4 71%	10-0 100%	0-4 0%	2-3 40%	1-2 33%	1-4 20%	3-2 60%	10-4 71%	1-4 20%	2-3 40%	5-4 56%	10-5 67%	4-5 44%	2-3 40%	7-4 64%	0-3 0%	3-3 50%	3-2 60%	2-2 50%	8-4 67%	-	0-2 0%
overall		-  25%	7-34 17%	6-30 17%	7-28 20%	3-21 12%	17-38 31%	12-25 32%	1-12 8%	18-35 34%	2-21 9%	1-21 5%	9-22 29%	23-36 39%	14-36 28%	7-21 25%	8-20 29%	14-22 39%	8-31 21%	8-26 24%	11-26 30%	0-20 0%	6-27 18%	10-28 26%	2-17 11%	8-4 67%	7-8 47%	5-19 21%

zerg

Steamhammer and ZZZKBot are opposite opponents, from Microwave’s point of view. Whatever worked against one did not work against the other. Most of the numbers in Steamhammer’s row, by the way, are from preparation games. I see the same numbers in the pre-learned data file. According to the history file, Microwave played its 9 pool speed opening in every game.

#	bot	total	12Pool	4PoolHard	4PoolSoft	5Pool	5PoolSpeed	6Pool	6PoolSpeed	7Pool	8Pool	9HatchMain8Pool8Gas	9PoolHatch	9PoolSpeed	9PoolSunken	OverpoolSpeed	OverpoolTurtle	ZvZ_Overgas11Pool	ZvZ_Overpool11Gas
5	steamhammer	26-16  62%	2-2 50%	-	-	1-2 33%	-	-	-	-	-	2-2 50%	1-2 33%	10-0 100%	1-2 33%	4-3 57%	-	-	5-3 62%
6	zzzkbot	67-70  49%	0-2 0%	1-5 17%	0-2 0%	10-7 59%	0-2 0%	0-2 0%	5-10 33%	1-5 17%	7-10 41%	10-7 59%	10-4 71%	0-2 0%	0-2 0%	3-4 43%	10-4 71%	10-0 100%	0-2 0%
overall		-  52%	2-4 33%	1-5 17%	0-2 0%	11-9 55%	0-2 0%	0-2 0%	5-10 33%	1-5 17%	7-10 41%	12-9 57%	11-6 65%	10-2 83%	1-4 20%	7-7 50%	10-4 71%	10-0 100%	5-5 50%

random

UAlbertaBot was the only random participant. It’s striking how similar openings can have different outcomes, though the numbers are noisy because the game counts intentionally limited and an opening that makes a bad first impression may not be repeated.

#	bot	total	4PoolHard	4PoolSoft	5Pool	5PoolSpeed	6Pool	6PoolSpeed	7Pool	8Pool	9PoolSpeedLing
11	ualbertabot	34-21  62%	5-4 56%	10-0 100%	4-4 50%	10-3 77%	4-3 57%	1-2 33%	0-2 0%	0-2 0%	0-1 0%
overall		-  62%	5-4 56%	10-0 100%	4-4 50%	10-3 77%	4-3 57%	1-2 33%	0-2 0%	0-2 0%	0-1 0%

#11 UAlbertaBot was one of the weaker participants, but no player shut it out. Even against #1 Locutus, UAlbertaBot scored 1 win and learned a little bit about its opponent. That also tells us something about each opponent.

The “total” column gives UAlbertaBot’s view of how many games it won and lost, which does not always line up with the tournament results. The results give UAlbertaBot 6 crashes, when it presumably could not record any information. Also if one side overstepped the frame time limit (UAlbertaBot never did), or if the game timed out and was decided on points (12 instances for UAlbertaBot), the player has no way to know what the tournament manager decided, and the two may disagree about who won. Something like that must explain why UAlbertaBot recorded 3 wins for itself against #2 PurpleWave when officially it won only 2 games. These issues cause difficulties for learning, but as long as most games finish normally it shouldn’t be serious.

#	bot	total	Terran	Terran	Terran	Terran	Protoss	Protoss	Protoss	Zerg	Zerg	Zerg	Zerg
			4RaxMarines	MarineRush	TankPush	VultureRush	DTRush	DragoonRush	ZealotRush	2HatchHydra	3HatchMuta	3HatchScourge	ZerglingRush
1	locutus	1-99  1%	0-9 0%	0-8 0%	0-8 0%	0-8 0%	0-11 0%	1-15 6%	0-10 0%	0-8 0%	0-8 0%	0-7 0%	0-7 0%
2	purplewave	3-93  3%	0-8 0%	0-8 0%	0-8 0%	0-8 0%	3-18 14%	0-8 0%	0-7 0%	0-7 0%	0-7 0%	0-7 0%	0-7 0%
3	bananabrain	16-82  16%	0-7 0%	1-10 9%	0-6 0%	0-6 0%	0-4 0%	0-4 0%	9-20 31%	0-4 0%	0-4 0%	0-4 0%	6-13 32%
4	daqin	21-77  21%	0-10 0%	0-9 0%	0-9 0%	0-9 0%	0-3 0%	3-8 27%	5-8 38%	0-2 0%	0-2 0%	0-2 0%	13-15 46%
5	steamhammer	9-89  9%	0-4 0%	8-9 47%	0-4 0%	0-4 0%	0-6 0%	1-9 10%	0-5 0%	0-12 0%	0-12 0%	0-12 0%	0-12 0%
6	zzzkbot	10-89  10%	0-8 0%	0-8 0%	0-8 0%	0-7 0%	0-4 0%	0-4 0%	8-16 33%	0-7 0%	0-7 0%	0-7 0%	2-13 13%
7	microwave	18-81  18%	0-7 0%	2-12 14%	0-7 0%	0-7 0%	0-3 0%	0-3 0%	13-11 54%	1-9 10%	0-5 0%	1-9 10%	1-8 11%
8	iron	9-90  9%	1-9 10%	1-8 11%	0-5 0%	0-5 0%	0-8 0%	0-8 0%	3-18 14%	0-6 0%	0-6 0%	0-6 0%	4-11 27%
9	xiaoyi	26-68  28%	0-4 0%	0-5 0%	4-13 24%	0-4 0%	21-5 81%	0-2 0%	0-5 0%	0-7 0%	0-7 0%	0-6 0%	1-10 9%
10	mcrave	56-44  56%	-	22-9 71%	-	-	0-4 0%	10-19 34%	0-5 0%	-	-	-	24-7 77%
12	aitp	71-23  76%	-	19-8 70%	-	-	-	-	24-5 83%	19-2 90%	7-2 78%	1-2 33%	1-4 20%
13	bunkerboxer	88-12  88%	-	34-4 89%	-	-	-	-	30-0 100%	-	-	-	24-8 75%
overall		-  28%	1-66 1%	87-98 47%	4-68 6%	0-58 0%	24-66 27%	15-80 16%	92-110 46%	20-64 24%	7-60 10%	2-62 3%	76-115 40%

UAlbertaBot was random. Its learning plan is to first play its best opening for each race (terran marine rush, protoss zealot rush, zerg zergling rush), and switch away only if it lost too often. If you are always losing, there is no harm in experimentation. Against strong opponents it tried everything, to little avail. Against weak opponents, the best opening might be reliable, so it did not try others.

UAlbertaBot’s configuration file has enemy-specific strategies defined for many historical opponents. In this tournament, 2 of them reappeared: Iron and ZZZKBot, and the declaration for ZZZKBot says “make the default choices.” I don’t see evidence in the table that UAlbertaBot paid attention to its Iron-specific strategies, so I watched replays to find out. It turned out as I expected, an enemy-specific strategy became the default strategy, the expected best opening, and if it failed severely enough (as it always did against Iron) then UAlbertaBot would try its other strategies.

The “overall” row across the bottom tells us that its best openings truly were the best. In most cases, it did no good to try alternatives. The notable exceptions are that the dark templar rush won against XiaoYi, while the 2 hatch hydra rush won against AITP (this suggests that AITP consistently followed a mech strategy). Of course, UAlbertaBot played random, which can confuse opponents that learn. It’s possible that a protoss bot that always rushed dark templar might do less well against XiaoYi, and so on.

Some openings were useless in the tournament, and UAlbertaBot would have done better without them. For example, the 3 hatchery scourge opening is designed to combat XIMP by Tomas Vajda, and scored miserably. The terran vulture rush made 58 losses and no wins at all, a weight pulling down the ranking.

There is more to learn from the table. Steamhammer had some trouble against the terran marine rush, but shut out the zealots and the zerglings. The other 2 zergs had more trouble against the hard zealot rush (which was historically difficult for zerg bots to cope with, at least zerg bots other than KillerBot by Marian Devecka). I think the difference ultimately reflects the skills of the different bots. Steamhammer has micro and defensive weaknesses against ranged units in general (the one loss against protoss was to the dragoon rush). Its opening learning is ingenious enough to cover the weakness, but only at the expense of losses against protoss and zerg. So instead Steamhammer’s learning converged on the idea of allowing the marines to win sometimes, and strictly controlling the other races. It’s counterintuitive but effective.

For the AIIDE 2019 main tournament, I looked in each bot’s write directory to see what files it wrote, if any, and in its AI directory to see if it had prepared data for any opponents. The usual disclaimers apply: A bot does not necessarily use the data it writes. Preparation for specific opponents is not necessarily in the form of data in the AI directory, it might be in code.

#	bot	info
1	Locutus	The learning files look just like last year’s. Locutus also has pre-learned data for all but 2 of its opponents (DaQin and AITP), plus a number of bots that did not compete.
2	PurpleWave	The learning files have an initially chosen strategy followed by a sequence of “fingerprinted” enemy strategies. (PurpleWave also has specific preparation for its opponents, but that’s in code rather than data.)
3	BananaBrain	A learning file for each opponent in the form of brief records of results. Each record consists of date+time, map, BananaBrain’s strategy (“PvZ_9/9gate”), the opponent’s recognized strategy (“Z_10hatch”), a floating point number that I haven’t discovered the meaning of, and the game result. Cool data, I want to analyze it. There is also pre-learned data for 7 of the tournament opponents.
4	DaQin	Learning files straight from its parent Locutus (very similar to Steamhammer files). There is no visible pre-learned data (in a quick check I also found no opponent-specific code).
5	Steamhammer	Opponent model files unchanged from last year. Also prepared data for 9 of the tournament opponents.
6	ZZZKBot	Carried over from last year. Learning files for each opponent with detailed but hard-to-interpret information about each game.
7	Microwave	One file per opponent listing Microwave’s strategies, with win and loss counts for each limited to a max of 10. Microwave deliberately forgets history. This part is the same as last year. New is a “history” file for each opponent, with a one-line record of data about each game. Also pre-learned files for 8 tournament opponents plus many other bots. The author said in a comment that there may be errors in the learning data due to code bugs.
8	Iron	Nothing.
9	XiaoYi	Learning files like its parent SAIDA, for each defeat, internal error, or timeout.
10	McRave	2 files for each opponent, one with strategy information for all games similar to last year, and an “info” file which seems to include some kind of build order or unit count information about only one game.
11	UAlbertaBot	Carried over from past years. For each opponent, a file listing strategies with win and loss counts for each.
12	AITP	Opponent model files straight from Steamhammer. AITP is a Steamhammer fork.
13	BunkerBoxeR	Nothing.

Virtually all bots recorded data about their games and opponents. Only holdover Iron and newcomer BunkerBoxeR chose to ignore history. Most bots from #7 Microwave on up also had prepared data about predictable opponents. Learning is now an expected feature. To aspire to a high finish, you pre-train against your opponents. And ideally come up with a trick that your opponents cannot prepare for, like Locutus’s defensive opening with zealot drop.

I already have scripts to parse most of these learning files. I’ll write a new one for BananaBrain.

Here are the per-player map tables for the AIIDE 2019 bonus tournament on “unknown” maps, meaning standard and well-known maps that weren’t announced beforehand. Since the competition is about maps, I want to look at the maps first.

With 10 opponents and 100 rounds, there are 10 * 9 * 100 / 2 = 4500 games total if all are played, meaning 900 games per player and 900 / 5 = 180 per player per map. Each player has 9 opponents, so there are 180/9 = 20 games per cell in these tables. The official results say that 4491 of the 4500 were successfully completed, so some cells have slightly fewer. Tail-ender UAlbertaBot participated in every game that did not complete (C0G has the same pattern).

Locutus	overall	Polari	Longin	Arcadi	Fighti	Roadki
PurpleWave	48%	75%	40%	50%	35%	40%
DaQin	73%	75%	75%	85%	75%	55%
BananaBrain	66%	55%	75%	70%	75%	55%
Microwave	98%	100%	90%	100%	100%	100%
Steamhammer	99%	100%	100%	95%	100%	100%
XiaoYi	97%	100%	90%	100%	95%	100%
McRave	96%	100%	95%	100%	100%	85%
Iron	100%	100%	100%	100%	100%	100%
UAlbertaBot	98%	100%	95%	95%	100%	100%
overall	86.11%	89%	84%	88%	87%	82%

PurpleWave	overall	Polari	Longin	Arcadi	Fighti	Roadki
Locutus	52%	25%	60%	50%	65%	60%
DaQin	76%	60%	90%	70%	95%	65%
BananaBrain	68%	80%	60%	70%	60%	70%
Microwave	96%	95%	100%	95%	95%	95%
Steamhammer	86%	70%	95%	80%	90%	95%
XiaoYi	90%	90%	80%	90%	90%	100%
McRave	93%	95%	95%	90%	95%	90%
Iron	98%	100%	100%	100%	90%	100%
UAlbertaBot	100%	100%	100%	100%	100%	100%
overall	84.28%	79%	87%	83%	87%	86%

DaQin	overall	Polari	Longin	Arcadi	Fighti	Roadki
Locutus	27%	25%	25%	15%	25%	45%
PurpleWave	24%	40%	10%	30%	5%	35%
BananaBrain	41%	60%	30%	35%	30%	50%
Microwave	89%	95%	95%	90%	80%	85%
Steamhammer	97%	95%	95%	100%	100%	95%
XiaoYi	74%	100%	65%	70%	60%	75%
McRave	37%	50%	45%	30%	30%	30%
Iron	98%	100%	95%	100%	95%	100%
UAlbertaBot	77%	75%	70%	89%	68%	80%
overall	62.58%	71%	59%	62%	55%	66%

BananaBrain	overall	Polari	Longin	Arcadi	Fighti	Roadki
Locutus	34%	45%	25%	30%	25%	45%
PurpleWave	32%	20%	40%	30%	40%	30%
DaQin	59%	40%	70%	65%	70%	50%
Microwave	74%	75%	70%	75%	75%	75%
Steamhammer	81%	70%	80%	80%	85%	90%
XiaoYi	62%	50%	70%	50%	70%	70%
McRave	69%	80%	70%	55%	65%	75%
Iron	56%	30%	55%	65%	45%	85%
UAlbertaBot	87%	100%	85%	95%	80%	75%
overall	61.56%	57%	63%	61%	62%	66%

Microwave	overall	Polari	Longin	Arcadi	Fighti	Roadki
Locutus	2%	0%	10%	0%	0%	0%
PurpleWave	4%	5%	0%	5%	5%	5%
DaQin	11%	5%	5%	10%	20%	15%
BananaBrain	26%	25%	30%	25%	25%	25%
Steamhammer	84%	90%	95%	75%	80%	80%
XiaoYi	82%	85%	90%	70%	75%	90%
McRave	50%	45%	35%	45%	55%	70%
Iron	33%	30%	15%	40%	15%	65%
UAlbertaBot	65%	70%	60%	79%	60%	55%
overall	39.60%	39%	38%	39%	37%	45%

Steamhammer	overall	Polari	Longin	Arcadi	Fighti	Roadki
Locutus	1%	0%	0%	5%	0%	0%
PurpleWave	14%	30%	5%	20%	10%	5%
DaQin	3%	5%	5%	0%	0%	5%
BananaBrain	19%	30%	20%	20%	15%	10%
Microwave	16%	10%	5%	25%	20%	20%
XiaoYi	50%	25%	80%	70%	45%	30%
McRave	89%	85%	90%	95%	90%	85%
Iron	59%	85%	40%	60%	55%	55%
UAlbertaBot	90%	95%	90%	90%	84%	90%
overall	37.75%	40%	37%	43%	35%	33%

XiaoYi	overall	Polari	Longin	Arcadi	Fighti	Roadki
Locutus	3%	0%	10%	0%	5%	0%
PurpleWave	10%	10%	20%	10%	10%	0%
DaQin	26%	0%	35%	30%	40%	25%
BananaBrain	38%	50%	30%	50%	30%	30%
Microwave	18%	15%	10%	30%	25%	10%
Steamhammer	50%	75%	20%	30%	55%	70%
McRave	29%	35%	40%	25%	20%	25%
Iron	80%	80%	85%	70%	90%	75%
UAlbertaBot	79%	60%	85%	85%	100%	65%
overall	37.00%	36%	37%	37%	42%	33%

McRave	overall	Polari	Longin	Arcadi	Fighti	Roadki
Locutus	4%	0%	5%	0%	0%	15%
PurpleWave	7%	5%	5%	10%	5%	10%
DaQin	63%	50%	55%	70%	70%	70%
BananaBrain	31%	20%	30%	45%	35%	25%
Microwave	50%	55%	65%	55%	45%	30%
Steamhammer	11%	15%	10%	5%	10%	15%
XiaoYi	71%	65%	60%	75%	80%	75%
Iron	41%	25%	30%	45%	40%	65%
UAlbertaBot	46%	20%	63%	50%	55%	45%
overall	36.04%	28%	36%	39%	38%	39%

Iron	overall	Polari	Longin	Arcadi	Fighti	Roadki
Locutus	0%	0%	0%	0%	0%	0%
PurpleWave	2%	0%	0%	0%	10%	0%
DaQin	2%	0%	5%	0%	5%	0%
BananaBrain	44%	70%	45%	35%	55%	15%
Microwave	67%	70%	85%	60%	85%	35%
Steamhammer	41%	15%	60%	40%	45%	45%
XiaoYi	20%	20%	15%	30%	10%	25%
McRave	59%	75%	70%	55%	60%	35%
UAlbertaBot	75%	90%	80%	75%	80%	50%
overall	34.44%	38%	40%	33%	39%	23%

UAlbertaBot	overall	Polari	Longin	Arcadi	Fighti	Roadki
Locutus	2%	0%	5%	5%	0%	0%
PurpleWave	0%	0%	0%	0%	0%	0%
DaQin	23%	25%	30%	11%	32%	20%
BananaBrain	13%	0%	15%	5%	20%	25%
Microwave	35%	30%	40%	21%	40%	45%
Steamhammer	10%	5%	10%	10%	16%	10%
XiaoYi	21%	40%	15%	15%	0%	35%
McRave	54%	80%	37%	50%	45%	55%
Iron	25%	10%	20%	25%	20%	50%
overall	20.43%	21%	19%	16%	19%	27%

Against its strong opponents, Locutus had trouble on the map Roadkill, possibly because of the low-ground main. If Locutus stuck with its cannons-at-the-ramp strategy, the cannons were weak on low ground. Iron also struggled on Roadkill. Polaris Rhapsody, the only 2-player map, also showed some extreme results—see BananaBrain versus Iron and McRave versus UAlbertaBot.

There are plenty more details in the tables.

The AIIDE 2019 unknown maps competition results are up. At first glance, the biggest surprise is that the ranking is extremely similar to the ranking in the main tournament. 10 bots chose to compete. The weakest players did not participate, so the winning rates for all bots are lower than in the main tournament.

The results for some reason don’t include a straight listing of the 5 maps used. They are (2) Polaris Rhapsody, (3) Longinus 2, (4) Arcadia 2, (4) Fighting Spirit, and (4) Roadkill. Fighting Spirit is of course familiar to SSCAIT participants. The first 4 maps are classics from the KESPA era (which ended in 2012) and Roadkill is a more recent design.

I know from test games that Steamhammer plays well on Arcadia. (I test Steamhammer on all kinds of maps as a regular thing.) I’m pleased to see that reflected in the map statistics. Other map preferences that stand out are that DaQin likes Polaris Rhapsody and dislikes Fighting Spirit, Microwave prefers Roadkill, and Iron had trouble on Roadkill.

I will analyze the unknown maps tournament, at least to some extent. I’m not sure exactly how. There’s a bit of an embarrassment of riches at the moment.

I’m pleased with this one. This is the same data as yesterday, how each bot did against each other on each map, but organized by player rather than by map. If you’re a bot author, I think this is a better way to find out about strengths and weaknesses.

For example, the first table is from the point of view of Locutus. The percentages are Locutus’s win rates. The upset by DaQin on Aztec immediately stands out amid Locutus’s otherwise consistent results. I imagine that Bruce @ Locutus will examine those 10 games and perhaps find a bug that DaQin exploited. (Locutus played its cannons at ramp into zealot drop strategy in these games. It lost because cannons at the ramp are a poor defense when the outside is on higher ground—Aztec has low-ground main bases. Maybe a weakness in learning or preparation?)

Locutus	overall	Benzen	Destin	Heartb	Aztec	TauCro	Androm	Circui	Empire	Fortre	Python
PurpleWave	45%	50%	50%	40%	40%	40%	60%	40%	30%	40%	60%
BananaBrain	89%	100%	100%	80%	100%	100%	80%	70%	100%	70%	90%
DaQin	83%	100%	80%	100%	20%	80%	100%	90%	90%	80%	90%
Steamhammer	97%	90%	100%	100%	90%	100%	90%	100%	100%	100%	100%
ZZZKBot	99%	100%	100%	100%	100%	100%	100%	100%	90%	100%	100%
Microwave	92%	100%	100%	80%	80%	100%	90%	90%	90%	90%	100%
Iron	99%	100%	100%	100%	100%	100%	90%	100%	100%	100%	100%
XiaoYi	96%	100%	100%	100%	90%	70%	100%	100%	100%	100%	100%
McRave	99%	100%	100%	100%	100%	100%	100%	100%	90%	100%	100%
UAlbertaBot	99%	100%	100%	100%	100%	100%	100%	100%	90%	100%	100%
AITP	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%
BunkerBoxeR	99%	100%	100%	100%	100%	100%	100%	100%	90%	100%	100%
overall	91.42%	95%	94%	92%	85%	91%	92%	91%	89%	90%	95%

PurpleWave	overall	Benzen	Destin	Heartb	Aztec	TauCro	Androm	Circui	Empire	Fortre	Python
Locutus	55%	50%	50%	60%	60%	60%	40%	60%	70%	60%	40%
BananaBrain	44%	80%	70%	80%	50%	60%	10%	40%	20%	0%	30%
DaQin	85%	80%	70%	100%	100%	70%	80%	100%	100%	70%	80%
Steamhammer	71%	90%	100%	70%	40%	60%	80%	60%	70%	90%	50%
ZZZKBot	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%
Microwave	93%	100%	90%	90%	100%	90%	90%	90%	90%	90%	100%
Iron	98%	100%	100%	90%	100%	100%	100%	100%	100%	100%	90%
XiaoYi	97%	90%	100%	100%	100%	100%	90%	100%	90%	100%	100%
McRave	89%	100%	90%	100%	100%	90%	80%	100%	70%	70%	90%
UAlbertaBot	98%	100%	100%	100%	100%	100%	100%	78%	100%	100%	100%
AITP	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%
BunkerBoxeR	97%	100%	90%	90%	100%	100%	100%	100%	90%	100%	100%
overall	85.54%	91%	88%	90%	88%	86%	81%	86%	83%	82%	82%

BananaBrain	overall	Benzen	Destin	Heartb	Aztec	TauCro	Androm	Circui	Empire	Fortre	Python
Locutus	11%	0%	0%	20%	0%	0%	20%	30%	0%	30%	10%
PurpleWave	56%	20%	30%	20%	50%	40%	90%	60%	80%	100%	70%
DaQin	51%	30%	40%	60%	80%	50%	30%	60%	60%	60%	40%
Steamhammer	85%	100%	70%	90%	80%	80%	70%	100%	100%	70%	90%
ZZZKBot	83%	90%	100%	70%	80%	70%	90%	100%	70%	90%	70%
Microwave	71%	90%	70%	90%	70%	60%	70%	60%	80%	60%	60%
Iron	59%	40%	70%	50%	60%	60%	70%	80%	50%	70%	40%
XiaoYi	57%	40%	80%	40%	50%	50%	40%	80%	40%	80%	70%
McRave	69%	70%	60%	90%	70%	50%	80%	70%	70%	70%	60%
UAlbertaBot	84%	80%	90%	80%	90%	80%	90%	80%	89%	90%	70%
AITP	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%
BunkerBoxeR	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%
overall	68.81%	63%	68%	68%	69%	62%	71%	77%	70%	77%	65%

DaQin	overall	Benzen	Destin	Heartb	Aztec	TauCro	Androm	Circui	Empire	Fortre	Python
Locutus	17%	0%	20%	0%	80%	20%	0%	10%	10%	20%	10%
PurpleWave	15%	20%	30%	0%	0%	30%	20%	0%	0%	30%	20%
BananaBrain	49%	70%	60%	40%	20%	50%	70%	40%	40%	40%	60%
Steamhammer	94%	100%	100%	90%	100%	100%	100%	100%	70%	90%	90%
ZZZKBot	10%	10%	20%	10%	10%	10%	20%	10%	0%	10%	0%
Microwave	83%	90%	60%	60%	90%	100%	80%	90%	100%	80%	80%
Iron	92%	100%	90%	100%	90%	100%	90%	100%	80%	70%	100%
XiaoYi	82%	100%	100%	90%	100%	90%	60%	70%	60%	60%	90%
McRave	41%	40%	60%	40%	40%	20%	50%	30%	40%	30%	60%
UAlbertaBot	78%	70%	100%	80%	90%	70%	80%	80%	70%	60%	80%
AITP	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%
BunkerBoxeR	99%	100%	100%	100%	90%	100%	100%	100%	100%	100%	100%
overall	63.33%	67%	70%	59%	68%	66%	64%	61%	56%	57%	66%

Steamhammer	overall	Benzen	Destin	Heartb	Aztec	TauCro	Androm	Circui	Empire	Fortre	Python
Locutus	3%	10%	0%	0%	10%	0%	10%	0%	0%	0%	0%
PurpleWave	29%	10%	0%	30%	60%	40%	20%	40%	30%	10%	50%
BananaBrain	15%	0%	30%	10%	20%	20%	30%	0%	0%	30%	10%
DaQin	6%	0%	0%	10%	0%	0%	0%	0%	30%	10%	10%
ZZZKBot	59%	40%	60%	80%	80%	40%	50%	40%	60%	50%	90%
Microwave	25%	10%	30%	50%	30%	30%	20%	40%	20%	10%	10%
Iron	67%	70%	90%	100%	40%	40%	90%	60%	60%	100%	20%
XiaoYi	50%	30%	60%	50%	20%	10%	80%	50%	80%	60%	60%
McRave	86%	100%	90%	100%	100%	70%	100%	90%	90%	40%	80%
UAlbertaBot	91%	90%	90%	90%	89%	80%	90%	89%	90%	100%	100%
AITP	97%	90%	100%	90%	100%	100%	90%	100%	100%	100%	100%
BunkerBoxeR	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%
overall	52.25%	46%	54%	59%	54%	44%	57%	50%	55%	51%	52%

ZZZKBot	overall	Benzen	Destin	Heartb	Aztec	TauCro	Androm	Circui	Empire	Fortre	Python
Locutus	1%	0%	0%	0%	0%	0%	0%	0%	10%	0%	0%
PurpleWave	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%
BananaBrain	17%	10%	0%	30%	20%	30%	10%	0%	30%	10%	30%
DaQin	90%	90%	80%	90%	90%	90%	80%	90%	100%	90%	100%
Steamhammer	41%	60%	40%	20%	20%	60%	50%	60%	40%	50%	10%
Microwave	44%	50%	0%	60%	40%	50%	50%	50%	60%	30%	50%
Iron	55%	0%	50%	60%	30%	50%	100%	50%	80%	50%	80%
XiaoYi	49%	60%	90%	70%	10%	20%	60%	40%	20%	70%	50%
McRave	67%	70%	70%	50%	80%	40%	30%	80%	80%	90%	80%
UAlbertaBot	90%	90%	80%	90%	100%	90%	90%	70%	100%	90%	100%
AITP	72%	70%	60%	60%	90%	80%	20%	70%	90%	90%	90%
BunkerBoxeR	99%	100%	100%	100%	100%	100%	90%	100%	100%	100%	100%
overall	52.08%	50%	48%	52%	48%	51%	48%	51%	59%	56%	57%

Microwave	overall	Benzen	Destin	Heartb	Aztec	TauCro	Androm	Circui	Empire	Fortre	Python
Locutus	8%	0%	0%	20%	20%	0%	10%	10%	10%	10%	0%
PurpleWave	7%	0%	10%	10%	0%	10%	10%	10%	10%	10%	0%
BananaBrain	29%	10%	30%	10%	30%	40%	30%	40%	20%	40%	40%
DaQin	17%	10%	40%	40%	10%	0%	20%	10%	0%	20%	20%
Steamhammer	75%	90%	70%	50%	70%	70%	80%	60%	80%	90%	90%
ZZZKBot	56%	50%	100%	40%	60%	50%	50%	50%	40%	70%	50%
Iron	13%	0%	0%	10%	10%	0%	20%	50%	20%	10%	10%
XiaoYi	65%	60%	80%	70%	60%	80%	40%	80%	60%	30%	90%
McRave	64%	60%	80%	100%	70%	50%	60%	70%	60%	30%	60%
UAlbertaBot	82%	60%	100%	90%	60%	80%	60%	80%	90%	100%	100%
AITP	93%	90%	90%	70%	100%	100%	90%	100%	100%	100%	90%
BunkerBoxeR	99%	100%	100%	100%	90%	100%	100%	100%	100%	100%	100%
overall	50.67%	44%	58%	51%	48%	48%	48%	55%	49%	51%	54%

Iron	overall	Benzen	Destin	Heartb	Aztec	TauCro	Androm	Circui	Empire	Fortre	Python
Locutus	1%	0%	0%	0%	0%	0%	10%	0%	0%	0%	0%
PurpleWave	2%	0%	0%	10%	0%	0%	0%	0%	0%	0%	10%
BananaBrain	41%	60%	30%	50%	40%	40%	30%	20%	50%	30%	60%
DaQin	8%	0%	10%	0%	10%	0%	10%	0%	20%	30%	0%
Steamhammer	33%	30%	10%	0%	60%	60%	10%	40%	40%	0%	80%
ZZZKBot	45%	100%	50%	40%	70%	50%	0%	50%	20%	50%	20%
Microwave	87%	100%	100%	90%	90%	100%	80%	50%	80%	90%	90%
XiaoYi	26%	10%	50%	0%	50%	20%	40%	20%	30%	40%	0%
McRave	65%	70%	80%	80%	80%	80%	70%	50%	40%	40%	60%
UAlbertaBot	90%	100%	90%	100%	90%	90%	90%	60%	90%	90%	100%
AITP	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%
BunkerBoxeR	93%	100%	100%	80%	80%	100%	90%	100%	90%	90%	100%
overall	49.25%	56%	52%	46%	56%	53%	44%	41%	47%	47%	52%

XiaoYi	overall	Benzen	Destin	Heartb	Aztec	TauCro	Androm	Circui	Empire	Fortre	Python
Locutus	4%	0%	0%	0%	10%	30%	0%	0%	0%	0%	0%
PurpleWave	3%	10%	0%	0%	0%	0%	10%	0%	10%	0%	0%
BananaBrain	43%	60%	20%	60%	50%	50%	60%	20%	60%	20%	30%
DaQin	18%	0%	0%	10%	0%	10%	40%	30%	40%	40%	10%
Steamhammer	50%	70%	40%	50%	80%	90%	20%	50%	20%	40%	40%
ZZZKBot	51%	40%	10%	30%	90%	80%	40%	60%	80%	30%	50%
Microwave	35%	40%	20%	30%	40%	20%	60%	20%	40%	70%	10%
Iron	74%	90%	50%	100%	50%	80%	60%	80%	70%	60%	100%
McRave	36%	40%	10%	50%	30%	40%	30%	40%	30%	60%	30%
UAlbertaBot	73%	44%	60%	90%	80%	80%	75%	60%	89%	89%	60%
AITP	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%	100%
BunkerBoxeR	98%	100%	100%	100%	90%	100%	100%	100%	90%	100%	100%
overall	48.62%	50%	34%	52%	52%	57%	49%	47%	52%	50%	44%

McRave	overall	Benzen	Destin	Heartb	Aztec	TauCro	Androm	Circui	Empire	Fortre	Python
Locutus	1%	0%	0%	0%	0%	0%	0%	0%	10%	0%	0%
PurpleWave	11%	0%	10%	0%	0%	10%	20%	0%	30%	30%	10%
BananaBrain	31%	30%	40%	10%	30%	50%	20%	30%	30%	30%	40%
DaQin	59%	60%	40%	60%	60%	80%	50%	70%	60%	70%	40%
Steamhammer	14%	0%	10%	0%	0%	30%	0%	10%	10%	60%	20%
ZZZKBot	33%	30%	30%	50%	20%	60%	70%	20%	20%	10%	20%
Microwave	36%	40%	20%	0%	30%	50%	40%	30%	40%	70%	40%
Iron	35%	30%	20%	20%	20%	20%	30%	50%	60%	60%	40%
XiaoYi	64%	60%	90%	50%	70%	60%	70%	60%	70%	40%	70%
UAlbertaBot	43%	40%	60%	50%	10%	40%	30%	20%	70%	40%	70%
AITP	82%	80%	70%	60%	100%	90%	80%	80%	100%	100%	60%
BunkerBoxeR	71%	70%	80%	70%	50%	80%	70%	70%	80%	70%	70%
overall	40.00%	37%	39%	31%	32%	48%	40%	37%	48%	48%	40%

UAlbertaBot	overall	Benzen	Destin	Heartb	Aztec	TauCro	Androm	Circui	Empire	Fortre	Python
Locutus	1%	0%	0%	0%	0%	0%	0%	0%	10%	0%	0%
PurpleWave	2%	0%	0%	0%	0%	0%	0%	22%	0%	0%	0%
BananaBrain	16%	20%	10%	20%	10%	20%	10%	20%	11%	10%	30%
DaQin	22%	30%	0%	20%	10%	30%	20%	20%	30%	40%	20%
Steamhammer	9%	10%	10%	10%	11%	20%	10%	11%	10%	0%	0%
ZZZKBot	10%	10%	20%	10%	0%	10%	10%	30%	0%	10%	0%
Microwave	18%	40%	0%	10%	40%	20%	40%	20%	10%	0%	0%
Iron	10%	0%	10%	0%	10%	10%	10%	40%	10%	10%	0%
XiaoYi	27%	56%	40%	10%	20%	20%	25%	40%	11%	11%	40%
McRave	57%	60%	40%	50%	90%	60%	70%	80%	30%	60%	30%
AITP	75%	89%	80%	50%	78%	90%	70%	33%	90%	70%	100%
BunkerBoxeR	89%	70%	90%	90%	80%	90%	100%	80%	100%	100%	90%
overall	28.04%	31%	25%	22%	29%	31%	31%	33%	27%	26%	25%

AITP	overall	Benzen	Destin	Heartb	Aztec	TauCro	Androm	Circui	Empire	Fortre	Python
Locutus	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%
PurpleWave	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%
BananaBrain	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%
DaQin	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%
Steamhammer	3%	10%	0%	10%	0%	0%	10%	0%	0%	0%	0%
ZZZKBot	28%	30%	40%	40%	10%	20%	80%	30%	10%	10%	10%
Microwave	7%	10%	10%	30%	0%	0%	10%	0%	0%	0%	10%
Iron	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%
XiaoYi	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%
McRave	18%	20%	30%	40%	0%	10%	20%	20%	0%	0%	40%
UAlbertaBot	25%	11%	20%	50%	22%	10%	30%	67%	10%	30%	0%
BunkerBoxeR	59%	30%	20%	70%	40%	70%	100%	80%	60%	50%	70%
overall	11.62%	9%	10%	20%	6%	9%	21%	16%	7%	8%	11%

BunkerBoxeR	overall	Benzen	Destin	Heartb	Aztec	TauCro	Androm	Circui	Empire	Fortre	Python
Locutus	1%	0%	0%	0%	0%	0%	0%	0%	10%	0%	0%
PurpleWave	3%	0%	10%	10%	0%	0%	0%	0%	10%	0%	0%
BananaBrain	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%
DaQin	1%	0%	0%	0%	10%	0%	0%	0%	0%	0%	0%
Steamhammer	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%	0%
ZZZKBot	1%	0%	0%	0%	0%	0%	10%	0%	0%	0%	0%
Microwave	1%	0%	0%	0%	10%	0%	0%	0%	0%	0%	0%
Iron	7%	0%	0%	20%	20%	0%	10%	0%	10%	10%	0%
XiaoYi	2%	0%	0%	0%	10%	0%	0%	0%	10%	0%	0%
McRave	29%	30%	20%	30%	50%	20%	30%	30%	20%	30%	30%
UAlbertaBot	11%	30%	10%	10%	20%	10%	0%	20%	0%	0%	10%
AITP	41%	70%	80%	30%	60%	30%	0%	20%	40%	50%	30%
overall	8.08%	11%	10%	8%	15%	5%	4%	6%	8%	8%	6%

The zergs and terrans seem more sensitive to the map than protoss. For example, Locutus vs PurpleWave win rates vary from 30% to 60%, which could be entirely due to statistical fluctuation, while ZZZKBot vs Iron ranges from 0% to 100%, which is not random. I imagine that ZZZKBot’s map selectivity has to do with its learning algorithm. But overall, there are many cases where the map seems to make a difference against opponents of similar strength. I think bots will benefit from more sensitivity to the map design.