There's a lot of complaints in the Blizz Overwatch forum that seem unfounded.
With so few games played (10) and having wins and loses heavily weight your skill rank, that rank is going to fluctuate wildly at first and then stabilize as your total games played increases.
That's just math.
Winning 8 out of 10 placement matches is an 80% win/loss ratio. Immediately losing your next 5 matches in a row for a record of 8-7 will give you a 53.3% win/loss ratio. 80% to 53% in five games isn't the ranking system being overly harsh to you specifically, it's the result of having so few total games played.
If you had that same 80% ratio after your first 30 games (24-6), and then lost 5 straight games, you'd only fall to 68%. That's a near 28 point drop versus 12 points.
What you're not taking into account is that you're affected both ways. Winning a lot of games early on will give you an unusually high win/loss ratio (and therefore skill rank). People enjoy the effect then that but overvaluing their skill and undervaluing the instability that a low sample size causes.
Personal skill and win-loss record
Too many people discount one of these factors, leaning towards players with good win-loss placement records and disappointing rank, the result of poor personal play while being carried by teams that preformed well.
There are for groups worth noting:
1. Good skill rank and a good win-loss record.
2. Poor skill rank and a good win-loss record.
3. Good skill rank and a poor win-loss record.
4. Poor skill rank and a poor win-loss record.
Most complaints are coming from group #2, the "played poorly but got carried" group. Probably the second most vocal group is #3, they played very well but lost most of their games because of circumstances out of their control. Teammates who quit early forcing a handicap match (for whatever reason), players making poor Hero selection decisions, and players with lacking skill.
Determining your personal initial skill rank is not the only purpose of placement matches. They also exist to place /everyone else/ which will -- if the penalty and ranking system is working well -- force people who quit matches early and play poorly into lower ranks where you'll won't see them again this season, if you're playing well yourself.
Having wins and loses figure into your skill rank is necessary in any team game or sport, but it's still just number that pulls you in a certain direction. Over time as less skilled players and trouble makers fall in rank from consistently making their own teams lose, you'll either maintain your position with fewer undesirables around you, or you'll gain in rank, as you get to play more matches with more skilled and dedicated players.
What's happened so far
The average skill rank of people posting in the Overwatch forum out of the first 50 that I examined was 51.48, with a median of 50.00. The top 50 players placed between 66 and 50, and the bottom 50 between 50 to 37. /This is a biased sample since it only includes people who posted on this one forum and doesn't include everyone that did so./
What to expect
The data is still useful, though. Players in the bottom 50 tended to lose 50% or more of their placement matches, with some exceptions. There were a handful of 8-2 placements in the bottom 50 which indicate poor skill by that player.
That's good news, it shows that playing poorly while being carried by good players will harm your rank which eventually means you won't be fortunate to have good teammates later in the season.
A number of players with mediocre placements, 5-5 and 6-4, had high (for this sample) skill ranks. One 5-5 player initially ranked 55, showing very good personal skill while probably placing with bad teams. That's also good news, as most players around rank 55 had good records, meaning he/she will start off playing with better teams than they placed with almost immediately.
Things to ponder
Too many people overestimate their own skill and don't understanding how groups will form for rank. Just because possible rank ranges from 1 to 100 doesn't mean that low skilled players and trolls will be very low in the 10s and 20s while elite players will clump up in the 80s and 90s, with everyone else concentrating in the middle.
There will probably end up being two sets of groups with each group having three sub-groups as the season progresses. The first group will be players who don't have friends they can group with regularly and so end up playing mostly uncoordinated matches made up of truly random people.
The second group will be pre-made groups of friends and acquaintances of varying sizes with a strong likelihood of coordination. This group can be further divided by semi-professional groups who practice regularly and develop strategies, and professional groups that do nothing but play this game for a living with individual players who have already mastered multiple Heros.
You can expect professional teams to attain the highest ranks and you'll probably never see them in the middle and latter parts of the season in ranked play. Because of that, large pre-made groups of friends will probably run into them in quick-play while the professional teams are practicing for game time or testing potential recruits.
People playing alone will probably clump up in the high 30s and low 40s because pre-made groups that communicate will always have the advantage. Having something worth playing for (sort of) should make pre-made groups proliferate beyond their current numbers.
Where you'll land will be determined by your personal skill and what your situation is between those two extremes. Exceptionally skilled individual players should do pretty well and find themselves filling in with mediocre-to-good pre-made groups of less than 6 players. Unskilled players are going to rank poorly, probably no matter what, because good pre-mades will kick them out eventually for dragging them down.
Mediocre players, I'm guessing, will end up in the 30s, good players in the high 40s and mid-50s, and elite players in the 60s and higher.
The first season of competitive Overwatch won't matter much. We know D.Va is getting a DPS buff and other Heros will be changed before the fall. Next season won't feature sudden death as it exists now and there may even be new maps to play. There will be new maps for certain, and they'll be free. It's a matter of when they are release and the end of a competitive season is the best time to do it.
The deeper a talented player goes in the season, the easier it should be for them to steadily grind rank higher, while bad luck loses won't sting as much.
Players of all talent levels will benefit in rank by grouping up, but there will probably be a sweet spot in group size where decent players can reliably win without running up against more dedicated semi-pros. Groups of 5-6 that aren't serious about winning should stick to quick-play because of that. Individuals should find a group or also stick to quick-play.
Very few people will rank in the 90s or 10s the way things are calculated for the first season. High 60s is pretty damn good, so don't feel like you need to be in the mid-70s if you think you're really good. And don't despair about being in the low 40s and high 30s, especially if you've played less than 75-100 games. Seasonal play lasts months, if you're in the 30s near the end of the season, welp... ouch.
What I'd like changed
Ties instead of sudden death is the only fair and practical way to end stalemates short of playing best-of matches repeatedly until a definitive winner is determined. Professional teams might disagree, in which case custom games can offer a choice between ties and play-until-the-absolute-end.
A one-size-fits-all ranking algorithm strikes me as bad idea for a squad-based game that doesn't require actual teams. A single person teamed with a group of five will suffer more of a disadvantage than the group will. How much personal performance weighs in undermining your rank versus winning or losing the game could be a sliding scale. Put simply, being alone could result in personal performance weighing more heavily than winning or losing that particular match. Being in a group of four or more could result in winning or losing being more heavily weighted.
Since competitive matches ending always puts you back at the main game screen, the end of the match could (and should) be more involved than voting for player performance cards. Or those cards should carry weigh with rank. Getting at least one vote but less than five currently means nothing, in competitive play it could be used to positively affect a player's rank. Instead of letting players "rank" the map, let them rank the performance of their own team for the match (preferably on a scale of 1-5 or 1-10, certainly not just "good" or "bad").
Competitive matches are supposed to be taken seriously, so they should be treated seriously. If a player leaves and doesn't return after a certain amount of time, the match should end prematurely. A waste of time is preferable to pretty much any other solution. There is no way to continue the match without negatively affecting everyone who stayed at that point. It either has to end, or we find the least negative effect, which is probably allowing a ringer to join mid-game. It's the least awful solution that's still quite awful. The ringer could be significantly better than the player who left, or much worse. It's probably impossible to algorithmically find the perfect replacement. It is not inherently unfair to have substitutes mid-game in a 6v6 pre-made match, but only professional teams can enjoy that. For the rest of us, it really sucks, but the game just needs to end and we'll call the time we've lost playing that game the cost of doing business in a competitive, serious environment. Make the leaving penalty as harsh as you wish. It should be exceptionally harsh -- it's competitive play.
Show the existence of pre-made groups on the Hero selection screen, and on the "tab" game status overlay. If you're distracted at the wrong time it can be very easy to miss seeing the makeup of the teams when the game begins.
Leaderboards and overall rank by percentile. What good is knowing your rank if you don't know everyone else's?