|
On November 21 2025 20:12 RJBTVYOUTUBE wrote:Show nested quote +On November 21 2025 17:32 Kanzzer wrote:On November 20 2025 16:51 iopq wrote:
So we can see that cross spawn is overall balanced, close spawn is not quite. Why not just make more 3 player maps?
It would be a slightly longer distance than close, but shorter distance than cross. I believe besides the fact 3p maps are incredibly difficult to make (fitting an equilateral triangle inside a square - like NEMEC mentioned in the past), ZvZ can also become luck-based, which may not be good for tournament play. they can bring back Apocalypse into the map pool. statistically it is the most balanced map together with Radeon. Apocalypse also produced a lot of very high quality entertaining matches.
I wasnt around when Monopoly was used in the ASL mp. But i looked at the picture of it and is surprising that protoss actually dominates zerg and terran in such map. I mean not by a huge extent. I honestly could believe such stats are actually balanced. But it is a map that never comeback.
|
On November 21 2025 17:32 Kanzzer wrote:Show nested quote +On November 20 2025 16:51 iopq wrote:
So we can see that cross spawn is overall balanced, close spawn is not quite. Why not just make more 3 player maps?
It would be a slightly longer distance than close, but shorter distance than cross. I believe besides the fact 3p maps are incredibly difficult to make (fitting an equilateral triangle inside a square - like NEMEC mentioned in the past), ZvZ can also become luck-based, which may not be good for tournament play.
IDK about difficult but i would say the best maps made in SC remastered are proly.Sylphid / Ascension / Apocalypse / Dominator / all of them 3 players map.It seems map makers figured out the formule to get bangers out of 3 players while the 4 players formule is right now stuck in those boring clones we are getting. With the exception of Tempest that imo should have made few changes and be put again into the MP.
|
On November 21 2025 20:12 RJBTVYOUTUBE wrote:
they can bring back Apocalypse into the map pool. statistically it is the most balanced map together with Radeon. Apocalypse also produced a lot of very high quality entertaining matches.
On November 21 2025 20:20 [sc1f]eonzerg wrote:
IDK about difficult but i would say the best maps made in SC remastered are proly.Sylphid / Ascension / Apocalypse / Dominator / all of them 3 players map.It seems map makers figured out the formule to get bangers out of 3 players while the 4 players formule is right now stuck in those boring clones we are getting. With the exception of Tempest that imo should have made few changes and be put again into the MP.
I'll just quote NEMEC (he was responding to you on that thread actually, RJBTV. You were posting on that thread too, EonZerg):
On June 10 2024 04:35 POPsNemec wrote:
3 player maps are actually harder to make than 4 player maps. Drawing an equilateral triangle inside a square makes the map feel much smaller. (It becomes even more challenging when considering air units.)
Additionally, in Zerg vs. Zerg matches, Overlord searching luck can decide the game's outcome, leading to games that rely too much on luck.
On January 20 2025 03:26 POPsNemec wrote:
Apocalypse was removed from consideration because Terran players disliked it.
I personally think LatiAs is excellent at designing 3-player maps
(Sylphid, Apocalypse, Dominator).
When I have balance concerns while designing a map,
I often seek his advice as he has a great understanding of map balance.
I remember Freakling the great foreign mapmaker also said somewhere that 3p maps are very challenging, and it's not something all mapmakers are confident with. Testbug is one of the rare foreign mappers to be very proficient with 3p maps.
All the maps you listed are all made by LatiAs, Eonzerg, so it looks like out of all ASL-approved mapmakers, he's your best bet for a new and nice 3p map.
|
I'm processing the full data sample right now, will update everything soon-ish.
current ladder maps included
how many games will be end before 10 minutes in zvp ? and how many around 3-5 mins ? from 19365 games
yeah I can add vs-time-plots for specific mmr-brackets with the number of game which ended at any point in the background, just as for the full player base. I previously looked at those and there was not much variation between player groups, so didn't include.
[...]provide us with pair-wise analysis for the spawn locations. Could be interesting to break it down by year too, if possible, to test your hypothesis.
no I'm not gonna do this :D this is like checking for ghosts under your bed or something
not trying to be rude, this is actually kinda funny. like, how is it even supposed to work? do you think blizzard messed up a random number generator? the game doesn't even know what side a spawn is on or what map is being used. or was it some bad actor which changed the source code of brood war to specifically only in some rare cases increase this probability when detecting fighting spirit? while keeping track globally of spawn rates to make it fine over the whole sample lol... I guess you could do it with a UMS-map, but we don't have these in the data sample
|
do you think blizzard messed up a random number generator?
Most likely happened LMAO. Tell me something blizzard didnt freakin messed hahahahahhahahahahaha
dont feel pressure to do shit. We already grateful with what you have contributed. Like you said this could be chasing a ghost.
|
everything updated to use the full data set, including current ladder maps
On November 21 2025 07:56 sas.Sziky wrote:Hio Kraekkling gj!  maybe just i dont find there is number how many games will be end before 10 minutes in zvp ? and how many around 3-5 mins ? from 19365 games. i am just curious about this ( on high mmr )
overall numbers (19365 -> 23520) changed a bit because the data set became bigger.
tbh not too much insight in these plots, its mostly the same distributions with increasingly less data as we move up the MMR brackets
pvt + Show Spoiler +
pvz + Show Spoiler +
tvz + Show Spoiler +
|
On November 20 2025 23:56 Kraekkling wrote:
I switched to a different image hosting site - does it work now?
Yes, thank you! It works now
|
On November 22 2025 03:25 Kraekkling wrote:everything updated to use the full data set, including current ladder mapsShow nested quote +On November 21 2025 07:56 sas.Sziky wrote:Hio Kraekkling gj! maybe just i dont find there is number how many games will be end before 10 minutes in zvp ? and how many around 3-5 mins ? from 19365 games. i am just curious about this ( on high mmr ) overall numbers (19365 -> 23520) changed a bit because the data set became bigger. tbh not too much insight in these plots, its mostly the same distributions with increasingly less data as we move up the MMR brackets pvt + Show Spoiler +pvz + Show Spoiler +tvz + Show Spoiler +
Thank you i was curious about this because it was because of the lag or its just a normal ladder things. I mean my zvp s i think around 30% are Nex first or proxy 2 gw. So if i read correctly around 30-35 % end before 10 mins and almost half before 5 mins. so maybe the lag is a little bit forced them for playing this.( in my case )
|
On November 22 2025 02:21 Kraekkling wrote:I'm processing the full data sample right now, will update everything soon-ish. current ladder maps included Show nested quote + how many games will be end before 10 minutes in zvp ? and how many around 3-5 mins ? from 19365 games yeah I can add vs-time-plots for specific mmr-brackets with the number of game which ended at any point in the background, just as for the full player base. I previously looked at those and there was not much variation between player groups, so didn't include. Show nested quote +[...]provide us with pair-wise analysis for the spawn locations. Could be interesting to break it down by year too, if possible, to test your hypothesis. no I'm not gonna do this :D this is like checking for ghosts under your bed or something not trying to be rude, this is actually kinda funny. like, how is it even supposed to work? do you think blizzard messed up a random number generator? the game doesn't even know what side a spawn is on or what map is being used. or was it some bad actor which changed the source code of brood war to specifically only in some rare cases increase this probability when detecting fighting spirit? while keeping track globally of spawn rates to make it fine over the whole sample lol... I guess you could do it with a UMS-map, but we don't have these in the data sample
I mean, I would argue that them having it be TR vs TL on ladder for a period of more than a month IIRC is effectively the POC for this analysis, but as eon said, I respect what you've done already and am not making demands + I get your logic. Thanks again
|
Calgary26004 Posts
This is awesome! I don't know why I've never thought about match-up winrate vs time. Very interesting.
|
added bin labels which tell us which fraction of all games end before a given time
+ Show Spoiler +
|
Awesome work, thanks a lot mate!
|
On November 22 2025 04:27 Chill wrote:
This is awesome! I don't know why I've never thought about match-up winrate vs time. Very interesting.
This is an analysis I've seen quite a bit in AoE2
|
On November 22 2025 04:45 Kraekkling wrote:added bin labels which tell us which fraction of all games end before a given time + Show Spoiler +
I don't quite get this graph. It seems the later the game goes, the higher the P winrate is (from the numbers) but at the same time, the line starts dipping below 50%. How's that possible? Or do the percentage numbers just sum up the total number of games that lastet the specified time on the x-axis or shorter?
|
On November 22 2025 21:48 Miragee wrote:I don't quite get this graph. It seems the later the game goes, the higher the P winrate is (from the numbers) but at the same time, the line starts dipping below 50%. How's that possible? Or do the percentage numbers just sum up the total number of games that lastet the specified time on the x-axis or shorter?
No, protoss only wins more ( winrate over 50%) if game last between 9 and 17 minutes - yellow line.
Blue bars are cumulative and say how many of total games are ended before that time.
|
On November 22 2025 02:21 Kraekkling wrote:
do you think blizzard messed up a random number generator?
it's actually broken in single player, just reload in single player, you get MUCH more of the same spawn than other spawns when you restart mission
|
On November 22 2025 22:13 Soft_General_5023 wrote:Show nested quote +On November 22 2025 21:48 Miragee wrote:On November 22 2025 04:45 Kraekkling wrote:added bin labels which tell us which fraction of all games end before a given time + Show Spoiler + I don't quite get this graph. It seems the later the game goes, the higher the P winrate is (from the numbers) but at the same time, the line starts dipping below 50%. How's that possible? Or do the percentage numbers just sum up the total number of games that lastet the specified time on the x-axis or shorter? No, protoss only wins more ( winrate over 50%) if game last between 9 and 17 minutes - yellow line. Blue bars are cumulative and say how many of total games are ended before that time.
Yeah ok that makes sense, thank you.
|
I always knew Zerg was weak in the midgame
|
just give scv 40 hp already
how many more decades we gotta keep doing this?
|
On November 23 2025 09:47 tankgirl wrote:
just give scv 40 hp already
how many more decades we gotta keep doing this?
Just buff Scout and Ghost :D
|
|
|
|
|
|
EPT![[image loading]](https://i.ibb.co/yc0Y3X18/racial-win-rate-vs-time-pvt-match-MMR-ge-1900.png)
![[image loading]](https://i.ibb.co/DgP8Dp9J/racial-win-rate-vs-time-pvt-match-MMR-ge-2400.png)
![[image loading]](https://i.ibb.co/wFwc87JK/racial-win-rate-vs-time-pvz-match-MMR-ge-1900.png)
![[image loading]](https://i.ibb.co/Y4B9t0hT/racial-win-rate-vs-time-pvz-match-MMR-ge-2400.png)
![[image loading]](https://i.ibb.co/67Rd2kJq/racial-win-rate-vs-time-tvz-match-MMR-ge-1900.png)
![[image loading]](https://i.ibb.co/0vp4w7h/racial-win-rate-vs-time-tvz-match-MMR-ge-2400.png)
![[image loading]](https://i.ibb.co/v4MWjs4T/racial-win-rate-vs-time-pvt-match-MMR-ge-1900.png)
![[image loading]](https://i.ibb.co/DhJDVhK/racial-win-rate-vs-time-pvt-match-MMR-ge-2100.png)
![[image loading]](https://i.ibb.co/s8nxw0B/racial-win-rate-vs-time-pvt-match-MMR-ge-2300.png)
![[image loading]](https://i.ibb.co/fzMDbXX2/racial-win-rate-vs-time-pvt-match-MMR-ge-2400.png)
![[image loading]](https://i.ibb.co/9mHJTXJs/racial-win-rate-vs-time-pvz-match-MMR-ge-1900.png)
![[image loading]](https://i.ibb.co/BK7ghxsd/racial-win-rate-vs-time-pvz-match-MMR-ge-2100.png)
![[image loading]](https://i.ibb.co/B7WD14M/racial-win-rate-vs-time-pvz-match-MMR-ge-2300.png)
![[image loading]](https://i.ibb.co/BKc68Gyq/racial-win-rate-vs-time-pvz-match-MMR-ge-2400.png)
![[image loading]](https://i.ibb.co/XZwKNYRM/racial-win-rate-vs-time-tvz-match-MMR-ge-1900.png)
![[image loading]](https://i.ibb.co/RGz8sykV/racial-win-rate-vs-time-tvz-match-MMR-ge-2100.png)
![[image loading]](https://i.ibb.co/Q7y24c8b/racial-win-rate-vs-time-tvz-match-MMR-ge-2300.png)
![[image loading]](https://i.ibb.co/gZwf3Y1F/racial-win-rate-vs-time-tvz-match-MMR-ge-2400.png)
