This CPL, we collectively played a grand total of 1503 games (not counting preseason) over 13 weeks, across 4 tiers and 3 races.
Now that the dust has settled, I thought it'd be interesting to look through our replay database and try to extract some statistics from them. There's pressing questions to answer, such as: Which team had the best TvP? Which map was the most pusagi? Is CPL really the ZvZ mirror matchup league?
Well, we now have answers to all these questions and more.
Want to just see the raw statistics? Check them out here! For the executive summary, read on.
RANKINGS
First of all, let's start with the rankings. As you know, the winners of CPL S8 were Lettuce Attack, with 2 Fact 2 Furious being the runners-up. But how close were the teams this season?
In the regular season, 2 Fact 2 Furious reigned supreme, not only having the highest win rate by a few percentage points but beating rival Lettuce Attack in both their meets. The map win rate was a lot closer, but 2F2F was quite a lot better in converting close matches into a win.
After that comes Korean Mechanics, GOSU and UED, all amazingly close together. KM only barely beats them out in match win rate. 4 Pool Lifeguards were unfortunately quite a few points away, but they did win one of the weeks (week 2 against GOSU).
The playoffs stats, however, are quite different from the regular season. 2F2F had a fine semi-finals (in fact, scoring almost an identical 11-8 as Lettuce Attack) but then collapsed in the finals, losing 6-13. In fact, 2F2F had so much trouble that it tanked their match win rate below that of GOSU and KM!
By the way, even though 4 Pool Lifeguards ended last during the regular season, the team had a slightly better map win rate than UED during the playoffs. Very nice!
MATCHUP WIN RATES
Next up, let's look at the matchup win rates.
Note that the percentage indicates the win rate of the leftmost race (so 51.90% for ZvP means Zerg wins 51.90% of the time). This actually looks perfectly normal to me—the "active" race in the matchup is expected to win very slightly more than 50%, and all races hover at around 52% for this statistic.
Let's have a look at how this breaks down per tier.
The differences are a bit larger here, especially in Tier 3. The T3 Zergs have done an absolutely bang up job against Protoss in particular. A number that stands out in particular is the TvZ rate in Tier 2, which is a shocking 28.21% win rate for the Terran. Apparently T2 Zergs were just insanely strong this season.
Overall, it seems that the higher a player's skill, the closer they get to that 52/48 average.
Next, let's see how these stats break down per map.
Interestingly, each map has its own unique characteristics with a different distribution of win rates. What surprised me is that Butter was our most balanced map by far, since it was also the newest of all—it has an almost perfect 50%, 51% and 52% for the win rates per matchup.
Also, Neo Sylphid was surprisingly good for Terran considering how abusive Mutas can be on that map. Terran won a shocking 68% of the time there. Contrast that Eclipse, where the Zergs have totally figured things out, with Terran only winning 41% of the time.
We've also got win rates per matchup per team, which lets us see how various teams did in specific areas. Most of the matchups seem to be pretty balanced for all teams, but there are a few outliers.
Lettuce Attack had the win rate record, with ZvZ and PvP both having an identically astronomical 70.83%.
On the other end of the spectrum, GOSU had the worst matchup in PvT, winning only 5 of the 23 they played.
Now that we've looked at how the various different matchups break down, can we see a trend in how various races performed overall?
Actually, all races are amazingly close together. 21 years after the last balance patch and Brood War is still going strong in terms of balance.
GAME LENGTH
For the game length, I've calculated not the average, but the median game length: this is slightly more representative of most games, as outliers (unusually long or short games) don't have as big an effect on the final value. In contrast, the average times are a bit longer, as it's more common for games to go very long than very short.
Looking at how long our games were reveals that all matchups were more or less the length most people would expect. Except for one thing: TvT was not the longest matchup!
Actually, both ZvP and PvT were longer than our TvTs. Meanwhile, ZvZ and PvP were generally pretty short.
In terms of maps, it's interesting to see that the two-player map Eclipse had almost the same length as Polypoid and Good Night. Butter, meanwhile, had significantly shorter and more aggressive games. Neo Sylphid takes the cake, however, seeing the shortest games of all.
Finally, there's a bigger breakdown on the full statistics page, but the shortest and longest games were 2:36 and 1:26:20, respectively. Obviously the shortest was a ZvZ, but the longest? Again, not a TvT as you might expect. It was an astoundingly long ZvP that totally mined out Neo Sylphid and went to a spellcaster duel. You can watch it if you really want to.
MOST IMPROVED PLAYERS
At the end of the season we asked all coaches and team admins to come up with a list of four most improved players. These are the players who put in a lot of effort to really boost their game during the season.
Special shoutouts to all of these players. Looking at this list, I think most of them will go up a tier next season, especially if they keep practicing in the meantime.
MISCELLANEOUS
Finally, the real question everybody wants to know: which race has the highest APM?
It seems pretty clear that it's Terran, the ultimate tryhard race. Zerg is a close second, although Protoss actually consistently has a very slightly higher EAPM.
Most of the interesting stuff is out of the way now, but there's a few more small things to look at.
4 Pool Lifeguards and Lettuce Attack were the luckiest teams (or unluckiest, depending on your viewpoint?) in terms of getting their games cast. UED, meanwhile, only had 37.5% of their games get shown to the world.
Amazingly, we ended up casting exactly 50% of all games. Note that this only goes for the regular season, as during the playoffs we cast 100% of all games.
On that note, here's a statistic I collected purely because it's a pet peeve:
Remember, make your casters happy by playing on Top vs Bottom so we can switch the colors if necessary. Clearly, people who have been playing the game for longer just don't do this as much!
And lastly, let's have a look at who put in the hours to cast CPL games this season:
It's actually super impressive how many hours people put in. We've cast a total of 341 hours of content, over 14 days worth. For comparison, an entire ASL is around 45 hours on average. And keep in mind all that content is online for you to watch, on Youtube!
The top four casters are clearly in a league of their own. Shoutouts to Neblime, cold_warpgates, Saetzero and Scimo for being the most prolific casters of the season by far.
At the same time, remember that you can always just cast a single group if you want to try it out. We're always looking for new casters!
CLOSING WORDS
Well, that's about it. These stats were put together using data from our new web app and from data extracted from the replay files. There's some potential for errors if I aggregated the data wrong, so let me know if something looks off or incorrect to you.
Thanks for reading and see you all at CPL season 9!
–Dada78641
PS:
Want to join the CPL? We're a league designed to help lower level players learn and improve in a competitive format. Join our Discord!
Watch all the VODs of CPL season 8 on our Youtube playlist.
Interested in watching some CPL and want a quick recommendation? The final cast of the season, featuring our strongest Tier 0 players in the grand finals, is a great place to start!
edit 2022-08-09: fixed mistake in the caster statistics.