This article was originally published on GameSpot's sister site onGamers.com, which was dedicated to esports coverage.

　　

Semifinal Preview

With the semifinals for the World Championship rapidly approaching, we compiled the stats of the four remaining playoff teams, SSW, SSB, SHR, and OMG. SSW faces off against SSB in the upper bracket semifinal while SHR and OMG face off in the other semifinal. This article is organized into three sections and shows basic team stats, objective stats, and early-mid-late game stats.

　　

Basic Team Stats

Samsung White has looked the most impressive at Worlds thus far. The basic team stats confirm this by showing how SSW's KDA nearly doubles the KDA of the other semifinalists. SSW averages the most kills per game and lowest deaths per game despite having the lowest average game time. SSW has an action-packed playstyle and seeing how this style matches up against the more-conservative style of its sister team, SSB, will be very interesting.

　　On the bottom half of the bracket we see a pretty closely matched SHR and OMG. OMG struggled the most in group stage going 3-3 but 3-0'd Najin White Shield in the quarterfinals with new support player Cloud. OMG played two fewer games than SHR to this point and has the lowest KDA of the remaining teams.

　　

Objective Stats

Seeing all four teams with >50% objective proportion is not too shocking given the win-loss records of the teams. SSW once again has the best stats, giving up the fewest global objectives and fewest towers per minute. The Korean teams averaged high towers per minute, whereas the Chinese teams are worse in this stat. OMG secured the fewest first bloods and first dragons, with regional counterpart SHR securing the most first dragons.

　　

Early-Mid-Late Game Stats

SSW and SSB both average large gold leads at 10 minutes, which is surprising for SSB given that many people think that SSB tends to fall behind early and rely on mid-late game decision making and team fighting to win games. Both Korean teams earned an early lead in seven of their ten games and converted 100% of their leads into wins. SHR and OMG have noticeably weaker early games, evidenced by lower average gold amounts and worse gold differences. Both Chinese teams convert a large percentage of their early leads into wins, yet have "thrown" leads throughout the tournament thus far.

　　The 10-20 minute stats show that SSW is perhaps the strongest team in this phase. SSW averages an insanely high 2031 GPM during this phase, nearly 400 GPM higher than the other three teams. The Korean teams have a gold lead at 20 minutes in a majority of their games, whereas the Chinese teams have a lead in roughly half of their games. All four teams convert 100% of their 20-minute leads into wins, however. SHR had two comebacks from 20-minute gold deficits, the most of any team in the tournament. SSB is praised for their late game strength, yet recorded only one comeback out of three chances thus far.

　　It's remarkable that only three of SSW's ten games lasted longer than 30 minutes. In all three games, SSW converted its lead into a win. SSW's only loss thus far was to TSM and lasted less than 30 minutes. All four teams average large gold leads at 30 minutes which is expected given their win-loss records.

　　

Conclusion

　　Based on some of the team stats, one could expect to see SSW get out to early gold leads over SSB in a majority of their games. The intriguing aspect of this matchup is whether SSB can go toe-to-toe in the early game with SSW or if SSB has to instead rely on their late game strength. Even though many people consider SSB to be a better late-game team than SSW, SSW's late game stats seem to be as strong on paper. In the SHR and OMG matchup, one could expect to see longer, bloody games with perhaps less objective focus. One might expect to see both teams even in gold at 10 minutes, with OMG perhaps out-performing SHR in the 10-20 minute phase. The chance of a late-game comeback appears to be higher on paper for the Chinese semifinal.

　　Thanks to Cuzimafish and TinyShrimp for gathering data.