Hot take, but I don't see any reason to say Cross Spawn Nexus First is busted based on the data I've explained on the following video.
There is a reddit thread about this already but I'll also copy paste stuff from there to here
Q: If 1/3 of the games should be cross spawn, why we have 14k games total on 4 player map but only 4k games for cross spawn? Also some of the numbers don't add up.
A: I have eliminated games that are less than 60 seconds (dodges and free wins)
Without this restriction, I have about 16,529 games in PvT total and 5,411 of them are cross spawn. Which is very close to 1/3 of the game.
There maybe cases where people scout the nearest spawn location and just quit so that's why the cross spawn result is screwed a bit.
For the 35 game and the 41 games difference, those were undetermined meaning no clear winners. Those can be because of disconnects or hacks etc.
Q: I'm curious if there is any trends as you go up in MMR? Do terrans get better against 9 gate? Do they get better against Nexus first? Or do the builds not care about MMR since they're all S rank.
A: at 2600+ mmr
The nexus first results are abysmal. They don't do nexus first at all pretty much.
9 Gate - too few data points to make a conclusion. Only about 27 games total
at 2500+ mmr,
The nexus first is still bad but a bit better than 2600 data
- cross spawn - 42.5% win rate (120 games)
- not cross spawn - 40.7% win rate (226 games)
- overall win rate - 41.3%
9 Gate
- cross spawn - 38.2% win rate (34 games)
- not cross spawn - 45.5% win rate (55 games)
- overall win rate - 42.7%
So looks like 9 gate is a bit better than nexus first. Gateway at 10 looks like it has higher win rate though: 45.3% (1170 games for 2500+ mmr)
But we have to note that 2500+ mmr has only 44.2% winrate (1670 games) on Protoss side in general so 9 gate does look like above average build order lol
at 2600+ mmr, Protoss only has 35.2% winrate in general (508 games). Sadge
Note: All of these stats are for 4 player maps
Q: I noticed great on 9 is only a couple hundred occurrences, could that be just a few players who use that build who are just better at pvt? Nexus first seems to dominate the build orders, could the MMR of players going nexus first be inflated since it’s the most frequent build?
Seems like too many moving parts using the ladder to make a full conclusion.
A: So there were total of 14,605 games played on a 4 player map PvT
Nexus first build count - 2,681 (so no, it does not dominate the build order - 18.3%)
9 Gateway build count - 866 (so not a couple of hundred games - 5.93%)
10 Gateway build count - 8,480 (most common build order 58.1%)
And we take into account only 2600+ mmr matches, the results are even worse for nexus first. Therefore, I don't believe there are too many moving parts
Q: Okay then explain to me how the incredibly economically strong build that basically cannot be punished loses most of the time.
A: Well pros can answer that more than i can. I guess i can investigate how much economic advantage does nexus 1st give.
My guess is that this is a 4 player map issue. High level TVP had 35% winrate for Toss in 4 player maps.
We all know Terran scales better than protoss as the game drags on
I'm afraid at 2500+ and 2600+ mmr the player pool becomes just a handful of players. It pretty much just includes pros (when they ladder) and some high level amateurs from Korea.
2600+ is just top 20 and 2500+ is top 70. And how many among them are hackers? We know Best Bisu and Mini don't ladder. Snow does sometimes but he doesn't even have an ID above 2500 at the moment. Terran players ladder much more especially some ladder monster like Speed who is probably the main reason for the skewed win rate of TvP at this mmr.
Wouls be better if we can count the 12 Nexus win rate in non ladder pro matches but the only way to do that is watch their stream and count manually. Would take a team of many people and do that for a month or so to gather big enough sample size.
Good work, there's some nice insights here. IMO any discussion is better when the arguments are grounded in data like here.
technical comments ---
First of all, IMO it might be easier to digest the results if you'd put the numbers in some sort of graph? It would also make it easier in the future, if people want to share or reference your findings.
Then, I think you ought to evaluate the statistical significance of your findings. Basically, when the sample is as small as 50 or 100 games, the statistical errors are bigger than the effect you're trying to measure - i.e., you're likely just seeing noise.
With data samples this small, you're also likely mostly measuring the performance a handful of players; e.g. there's probably only a few players above some MMR, who really like doing this or that specific build. Your results are thus likely to be biased in this way, which you'd also need to evaluate.
I hate to be the guy to comment this, but I think this is relevant for the 2600+ mmr ladder, although 2500+ would probably have been a really good alternative as 2600+ is really jusy a very small sample size yet the same players occupy 2500-2600 as 2600+.
Ladder at 2500+ consists largely of pro terrans, pro zergs, and way less pro protoss. The protoss who are at that MMR are toss players whom are without a doubt tier 2-4 on pro level. Think of Rich, Motive, Ruin(hardly plays), nOOB, Brain, and Paralyze, although it can be argued they are a cut below the zergs and terrans who exist in this MMR. Airfou and Movie are the next best protoss and are very active in this rating range, and of all these players in this rating range, Airfou is single handedly so active he accounts for half of the protoss games in this MMR range.
For comparison, the terrans active within this range are Scan, Light, Royal, Speed, Shinee, sSak, Midas, Ample, BishOp, Rush, Mind, HBQ, Byul. Half of these players are stronger than the above mentioned protoss, and the other half is just as strong, but also stronger than Airfou and Movie.
For Zerg we see a similar trend. Soulkey, Queen, SoMa, BTS, HerO, Sacsri, Action, Shine, Soo, Kante, ddasik, Jaedong, Larva. Again half of these players are stronger than the strongest active ladder playing protoss, and the other half just as strong.
The protoss counterparts to the above players dont ladder, or they random on ladder. YSC, Best, and SnOw generally random. Mini rarely ladders. Stork has bursts of high ladder activity and then stops. Rain plays a lot of ladder when active so he is a viable counterpart, but he is not often active.
As a result we have an over representation of the strongest Zergs and Terrans vs mid tier protoss pros on the 2500+ part of the ladder.
Note: i have to add the effect this has on the outcome is probably a little but smaller than it impliesm
On June 05 2026 22:14 TMNT wrote: I'm afraid at 2500+ and 2600+ mmr the player pool becomes just a handful of players. It pretty much just includes pros (when they ladder) and some high level amateurs from Korea.
2600+ is just top 20 and 2500+ is top 70. And how many among them are hackers? We know Best Bisu and Mini don't ladder. Snow does sometimes but he doesn't even have an ID above 2500 at the moment. Terran players ladder much more especially some ladder monster like Speed who is probably the main reason for the skewed win rate of TvP at this mmr.
Wouls be better if we can count the 12 Nexus win rate in non ladder pro matches but the only way to do that is watch their stream and count manually. Would take a team of many people and do that for a month or so to gather big enough sample size.
For 2500, that is quite right. The 14,605 is for 2300+ MMR. I can also do it for 2000+ but that pool has tons of replays and maybe it will take months to analyze those. But it is possible
On June 06 2026 00:21 RJBTVYOUTUBE wrote: I hate to be the guy to comment this, but I think this is relevant for the 2600+ mmr ladder, although 2500+ would probably have been a really good alternative as 2600+ is really jusy a very small sample size yet the same players occupy 2500-2600 as 2600+.
Ladder at 2500+ consists largely of pro terrans, pro zergs, and way less pro protoss. The protoss who are at that MMR are toss players whom are without a doubt tier 2-4 on pro level. Think of Rich, Motive, Ruin(hardly plays), nOOB, Brain, and Paralyze, although it can be argued they are a cut below the zergs and terrans who exist in this MMR. Airfou and Movie are the next best protoss and are very active in this rating range, and of all these players in this rating range, Airfou is single handedly so active he accounts for half of the protoss games in this MMR range.
For comparison, the terrans active within this range are Scan, Light, Royal, Speed, Shinee, sSak, Midas, Ample, BishOp, Rush, Mind, HBQ, Byul. Half of these players are stronger than the above mentioned protoss, and the other half is just as strong, but also stronger than Airfou and Movie.
For Zerg we see a similar trend. Soulkey, Queen, SoMa, BTS, HerO, Sacsri, Action, Shine, Soo, Kante, ddasik, Jaedong, Larva. Again half of these players are stronger than the strongest active ladder playing protoss, and the other half just as strong.
The protoss counterparts to the above players dont ladder, or they random on ladder. YSC, Best, and SnOw generally random. Mini rarely ladders. Stork has bursts of high ladder activity and then stops. Rain plays a lot of ladder when active so he is a viable counterpart, but he is not often active.
As a result we have an over representation of the strongest Zergs and Terrans vs mid tier protoss pros on the 2500+ part of the ladder.
Note: i have to add the effect this has on the outcome is probably a little but smaller than it impliesm
That is great insight. I can filter this result for min MMR and max MMR. Is 2300 to 2500 going to be better?
I can also do analysis for 2000 to 2300 but that will take a month of computing time i think
On June 06 2026 00:21 RJBTVYOUTUBE wrote: I hate to be the guy to comment this, but I think this is relevant for the 2600+ mmr ladder, although 2500+ would probably have been a really good alternative as 2600+ is really jusy a very small sample size yet the same players occupy 2500-2600 as 2600+.
Ladder at 2500+ consists largely of pro terrans, pro zergs, and way less pro protoss. The protoss who are at that MMR are toss players whom are without a doubt tier 2-4 on pro level. Think of Rich, Motive, Ruin(hardly plays), nOOB, Brain, and Paralyze, although it can be argued they are a cut below the zergs and terrans who exist in this MMR. Airfou and Movie are the next best protoss and are very active in this rating range, and of all these players in this rating range, Airfou is single handedly so active he accounts for half of the protoss games in this MMR range.
For comparison, the terrans active within this range are Scan, Light, Royal, Speed, Shinee, sSak, Midas, Ample, BishOp, Rush, Mind, HBQ, Byul. Half of these players are stronger than the above mentioned protoss, and the other half is just as strong, but also stronger than Airfou and Movie.
For Zerg we see a similar trend. Soulkey, Queen, SoMa, BTS, HerO, Sacsri, Action, Shine, Soo, Kante, ddasik, Jaedong, Larva. Again half of these players are stronger than the strongest active ladder playing protoss, and the other half just as strong.
The protoss counterparts to the above players dont ladder, or they random on ladder. YSC, Best, and SnOw generally random. Mini rarely ladders. Stork has bursts of high ladder activity and then stops. Rain plays a lot of ladder when active so he is a viable counterpart, but he is not often active.
As a result we have an over representation of the strongest Zergs and Terrans vs mid tier protoss pros on the 2500+ part of the ladder.
Note: i have to add the effect this has on the outcome is probably a little but smaller than it impliesm
That is great insight. I can filter this result for min MMR and max MMR. Is 2300 to 2500 going to be better?
I can also do analysis for 2000 to 2300 but that will take a month of computing time i think
I say go for 2350 or 2400+, and filter out games shorter than 2 minutes, because quite often players leave randomly after more than 1 minute.
I think you need a larger pool size, any stats using less than a 100 samples isn't representative enough from my perspective, but the results on nexus first build appear to be significative already so I don't think it will deviate a lot. Maybe take all S rank reps.
On June 06 2026 15:01 iFU.pauline wrote: I think you need a larger pool size, any stats using less than a 100 samples isn't representative enough from my perspective, but the results on nexus first build appear to be significative already so I don't think it will deviate a lot. Maybe take all S rank reps.
The number of games going from 2600+ to 2500+ to 2400+ should multiply the entries sufficiently I would say
On June 06 2026 00:21 RJBTVYOUTUBE wrote: I hate to be the guy to comment this, but I think this is relevant for the 2600+ mmr ladder, although 2500+ would probably have been a really good alternative as 2600+ is really jusy a very small sample size yet the same players occupy 2500-2600 as 2600+.
Ladder at 2500+ consists largely of pro terrans, pro zergs, and way less pro protoss. The protoss who are at that MMR are toss players whom are without a doubt tier 2-4 on pro level. Think of Rich, Motive, Ruin(hardly plays), nOOB, Brain, and Paralyze, although it can be argued they are a cut below the zergs and terrans who exist in this MMR. Airfou and Movie are the next best protoss and are very active in this rating range, and of all these players in this rating range, Airfou is single handedly so active he accounts for half of the protoss games in this MMR range.
For comparison, the terrans active within this range are Scan, Light, Royal, Speed, Shinee, sSak, Midas, Ample, BishOp, Rush, Mind, HBQ, Byul. Half of these players are stronger than the above mentioned protoss, and the other half is just as strong, but also stronger than Airfou and Movie.
For Zerg we see a similar trend. Soulkey, Queen, SoMa, BTS, HerO, Sacsri, Action, Shine, Soo, Kante, ddasik, Jaedong, Larva. Again half of these players are stronger than the strongest active ladder playing protoss, and the other half just as strong.
The protoss counterparts to the above players dont ladder, or they random on ladder. YSC, Best, and SnOw generally random. Mini rarely ladders. Stork has bursts of high ladder activity and then stops. Rain plays a lot of ladder when active so he is a viable counterpart, but he is not often active.
As a result we have an over representation of the strongest Zergs and Terrans vs mid tier protoss pros on the 2500+ part of the ladder.
Note: i have to add the effect this has on the outcome is probably a little but smaller than it impliesm
That is great insight. I can filter this result for min MMR and max MMR. Is 2300 to 2500 going to be better?
I can also do analysis for 2000 to 2300 but that will take a month of computing time i think
I say go for 2350 or 2400+, and filter out games shorter than 2 minutes, because quite often players leave randomly after more than 1 minute.
I'd say go for even more than 2 minutes. What's the timing of the earliest bunker rush after Terran first scouts a close spawn 12 Nexus? Go for the time just before that , because some T may rage quit after scouting a 12 Nexus too.
EPT
