Street Fighter V: Competitive Design Review

A quick reminder of how we review competitive gaming design. It’s all about the 1, 2, 3:

1. A level playing field.

2. Effective skill-based matchmaking.

3. Accurate and fair skill & result-based ranking.

Street Fighter V seems to do a very good job at all of these on the surface. In fact, despite everything I’m about to criticise it for, it’s still the best competitive design I’ve seen for (at least non-Early Access…) fighting games to date. The fact it has a large audience, courtesy of it being the powerful IP and brand of “Street Fighter” of course, and the fact they were incredibly smart to add cross-platform play between the PS4 and PC/Mac Steam players, really does allow them to “paper over the cracks” as it were when it comes to its competitive gaming design. Simply put, the large playerbase smooths out a lot of these problems.


Single Ranked Score per account:

SFV’s first glaring problem is that you only have a single Ranked score per player account, regardless of what character you pick to use. Exacerbating this problem is that this is also the ONLY score you are also matchmade on. This means, if you are very good and practiced with one character, but are not as proficient with others, as soon as you (or anyone else) changes characters away from your best one, you’ve immediately partially broken both the ranking and matchmaking systems. This means for example for myself, if I want to get at all valid matchmaking or ranking, SFV’s character select screen essentially looks like this:

Now, select your best Bipson!
Now, select your best Bipson!

In practice, because most people use their best character most of the time in matchmade modes (Casual and Ranked battles), and because of the aforementioned large playerbase, it doesn’t end up too bad a lot of the time, but it’s still a huge failing of the design, and one that all fighting game devs (who’re primarily Japanese) have continually got wrong for multiple generations of video games now.

This horribly bad design also pushes people towards the non-matchmade (& very poor in other ways) Battle Lounge fight ‘lobby’ system, especially when they want to play a new character. This is a ‘general bad’ for the whole community as it removes people from the matchmaking pool and reduces the chances that good skill & connection matches can be made with every player who’s playing in a lobby instead of a matchmade queue.

And, even worse, it also incentives players to make to create extra accounts just to play as new characters (if they can do; Steam limits this option somewhat more than PSN). These accounts then completely screw the matchmaking and ranking for everyone else, as you now essentially have a “smurf” account, whether intentional or not. Whenever an expert overall player creates a new account, they will start from a Ranked score of 0 (Rookie rank) and be matchmade with, and be totally crushing, far weaker opponents .. and, as we’ll get to later, thanks to the flaws in the ranking system, this may go on for a large number of opponents. Plus there’s even another very easily exploitable smurf loophole here: if you don’t play enough (or any) Ranked mode, you will be playing lesser skilled opponents forever in Casual match, the system can never adjust for this (other than opponents blacklisting to avoid you!). This is even evident for my own account. Despite the fact I try to keep my ranked score as high and as accurate to my current skills as I possibly can do (I’ve always had around a 50% win rate in Ranked), my win rate % is much higher in Casual match mode. This is mostly because many people choose to sometimes play their secondary or new characters in the Casual mode (rather than Ranked, where they would de-rank by playing as them), and so I am accidentally exploiting them and picking up ‘extra wins’ vs an opponent who is clearly lesser skilled than me when I am playing with my main, and they are not.

Incredibly, given their sales model for SFV, this bad design even puts people off trying or even buying new characters! So this isn’t a good thing for their business, nor for online “fun factor” of getting the chance to play as or against a lot of different characters. As noted, I myself only ever play my main character on my main account in the matchmade modes, and reserve practising a new or secondary character vs my friends in a Battle Lounge. Conversely if SFV had a better designed Ranked score per character I would’ve played as almost EVERY character in the ranked mode (& there’s already some in-game currency fight money incentive to do this, just as there is to deliberately smurf), just to see how well I could do with them.


Semi-grindable ranking system:

The second notable problem for SFVs competitive gaming design is the fact that the Ranking system is semi-grindable. What I mean by this, instead of starting at some kind of ‘average’ skill rating, such as with ELO or TrueSkill, and then adjusting from there based on a player’s performance, this system starts you at zero (Rookie rank in the game) and you have to work up from there getting wins, at a potentially very high percentage for possibly a very long time before the system can find your “true” rank. No matter how good you are, you simply HAVE to play a lot of matches (grind) to rank your account up. I say “semi-” though because the system does mean you lose points when you lose, so it’s not a pure grind or a mere experience point system as such. This means that the Ranked score is meaningful for an SFV account and you do have to improve your skill in order to attain higher ranks in the game, but it’s only in combination of improving your skill AND playing a lot of matches to prove it, that your Ranked score will reflect your skills accurately.

semi-truck
Here’s a semi… ready to grind.. or something! 😉

This kind of ranking system does have some nice features to it – especially the sense of progress it can provide to players, and the incentive to play more – but it also has the obvious huge downside that under-ranked players will cause to matchmaking. This is exacerbated due to the prevalence of intentional and unintentional smurf and extra accounts in SFV for the reasons covered already. Now every ranking system of course has this problem of ‘finding a player’s true skill’ and also accurately reflecting their skills over time as they improve (or even decline) in skill at a game. But SFV is actively working to make this as bad a problem as possible by its design! There’s no way to self-rank yourself at a higher level on a new account, by choice or by doing a single-player skill test of some kind. There’s no “placement matches” or other such things that other ranked systems have used to more quickly sort the skills of their players. Usually the best ranking systems use multiple score values (often some of them hidden to the players themselves) to find a way to both accurately and quickly reflect player skill at least for their matchmaking, whilst simultaneously providing a sense of progress for players and offering incentives for those who play a lot.

Whilst I noted just now that you need to improve your skills to improve your ranking score in SFV, there’s one way this is not strictly true. That’s because the system is designed to gradually inflate the rank of all players over time. As all new accounts start at zero, and gain points much more quickly than they lose them (even with a 50:50 win/loss at lower ranks, you’ll be rising), the overall points within the system are not zero-sum. This is why if you continually play SFV you’ll notice that the rank needed to actually obtain, for example, the top 20,000 accounts in the world is gradually rising. And if you stay static in rank score you’ll find your actual ranking amongst the total number of accounts will slowly decline. This is also why at the very top end of the ranking system they keep adding new leagues beyond the previous top rank.

Here’s a couple of good quotes from players that reflect this semi-grind design:

“The ranked system really does seem to reward tenacity slightly more than skill. You still need to be good in order to get to Platinum, but the majority of your rank-up points will be from lower league players.” (Source)

“Anything that uses more of an ELO or TrueSkill type of points system would work if the goal is for points to better reflect player skill. The design purpose is presumably for players to grind, though, and get new titles and other stuff from playing and eventually ranking up with the gradual points inflation, rather than more accurately trying to measure skill. If that gets more people to play more often, then mission accomplished as far as I’m concerned.” (Source)


Demarcation between leagues (used for matchmaking) does not match the playerbase:

The third huge flaw is in SFVs system is this. The system is good in that it doesn’t use the precise ranked score (called “League Points” or LP in SFV) for matchmaking, but allows you to fight vs any player around your ‘league’ in skill. The leagues or skill bands it has are set at arbitrary ranking point values, for example, 500 points gets you out of the initial Rookie league mentioned earlier and into Bronze. 4000 points gets you into Gold league, complete with a new in-game title and Achievement. SFV’s matchmaking will usually (with an exception at the top end leagues) allow you to be matched with any players within your league band, and a league band above or below you. We’ve mentioned before on this site why this can be a good thing, to allow you to face some easier and harder matches, and the system doesn’t try to ‘force’ a 50:50 win/loss rate at all, which can be demoralising for players. The boundaries between league bands also become bigger as you rise up through the rankings, for example someone in Super Platinum rank faces opponents varying up to 4500 LP in skill, whereas someone in Super Silver faces only a 1500 LP difference range.

So far so good in general, but the problem is that the designers either vastly overrated the competency of the general SFV player base, or vastly overrated the amount most players would play Ranked – and likely a combination of both.

This graph is from stats collected from V-League before Capcom annoyingly shut them down, and so is now years out of date (which is an issue with the ever inflating overall rank scores), but still tells an important story:

Basically, in mid-2017 at least 80% of the player base who had played the Ranked mode were still in the Bronze League or below.

These figures from this thread mid-2016 show it was the same problem earlier in the games life too.

Some other figures available related to this are the public achievements on Steam (this is at January 2019):

  • 38% of players on Steam have 1+ ranked wins.
  • 27.8% have 10+ ranked matches.
  • 12.9% have 100+ ranked wins.
  • 12.9% have 300+ online match-made matches played

However this only really shows how little most players play online in SFV, which does muddy all of these results (although players that don’t play Ranked at all aren’t counted above).

Really what we would want to measure is the figures across the leagues of actually relatively active player accounts, to see where the active online playerbase really is across the various league bands: Sadly there are no figures for this available publicly. But all of these figures shown above, plus the anecdotal experience of myself and many other players, all point to the fact that the distribution of the playerbase in SFV is a bit of a mess overall, with way too many players clustered at the lower end of the league bands. Despite this probably slowly improving over time due to the general rank inflation, it has happened far too slowly to be healthy for the games matchmaking. If you think about the playerbase as a bell curve of skills, it means you need the most accurate matchmaking around the mid-point of that bell curve where the majority of players are. It appears SFV attempted to do something like that with their league boundaries growing as players rise the ranks as noted above, but because of how slow their semi-grindable system is to accurately reflect the skill of a player, instead they’ve got a huge variety of skills of players all clustered within the lower end of the leagues, all suffering from pretty poor matchmaking.

“The population drops off so heavily at silver that the chance of getting matched unfairly skyrockets at that point I think.” (-FightingRoadhog)

“I’m platinum and most of the time I’m just farming low league players with a balanced match here and there.” (Source)

“Yup, when I reach Gold 4000LP which I do every now and again, I then get matched with people up to 7500LP who are easily like, double my skill, and I’m gaining/losing pretty much equal points win/loss vs them since I’m then in the same ‘division’ as them. Basically the divisions dividers (heh) are in the wrong place for the playerbase.” (me, a few years ago)


Other points:

There are other good things about SFV’s gaming design. The Fight Request system introduced in SFIV is still used and it’s still a great idea to allow the player to play the actual game in other modes whilst waiting for a match. Sadly the implementation isn’t as good in SFV since the fight request system for some reason is forced to pause during the many loading pages or menu screens between fights which means you aren’t in the queue at all at these times, and it can even interrupt a matchmade game from even happening at all because the game happens to load at the wrong time.

The netcode, whilst still being utter rubbish compared to the best stuff like GGPO, is at least a huge improvement over SFIV due to it being rollback netcode rather than input delay code. Sadly the netcode seems to have issues consistently matching you well with someone with a good connection to you. It seems to be unable to handle (or detect) wifi users, or router compatibility (see our port-forwarding guide to help) or cross-platform issues, and on top of this is one-sided rollback which means bad games are going to be REALLY bad for one of the players, but still, it’s at least an improvement on the sad state of netcode that still passes in most mainstream fighting game titles (despite the fact smaller indie games still in Early Access like Fantasy Strike have far better netcode than any big budget game).


Competitive Design Review:

starstarstar

Overall, I like Street Fighter V quite a lot as a game, and it’s a much more fun game than Street Fighter IV for me. I held off writing this review for a few years of the game’s life in the hopes that some of these problems would be addressed, but they have not. So, despite the improvements it’s made in some areas of its design, these flaws mentioned here, and the fact they all exacerbate each other, means that I can’t score it any higher than SFIV in terms of its overall competitive gaming design.

2 thoughts on “Street Fighter V: Competitive Design Review

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.