Graphic summary of this season's placements for my old main, long stuck in silver (
ZeroGravitas#2272), verses my newer alt in high gold (
Z3R0gravitas#2420). 27% verses 60% win rate, respectively (across the 15 matches on each). Seems backwards...
This is just a snippet of evidence of a larger picture, where my silver comp games have been
consistently harder and less winnable than games on my alt, for a long time. [
Edit:
note, I also placed my alt for the first time in season 20, then played an additional 5 games per role, for a total of 45 matches on that account to date.]
Despite my alt being 402 to 873 SR higher, this season, I felt the games were more fair and easier, overall. I got more wins despite lower stats, sometimes (dps role, particularly). My occasional mistakes having a clear impact without being totally devastating.
The standard basic advice from this sub (and elsewhere)
actually being fully relevant, on my alt. E.g. stay alive, get value, synergise and counter pick, analyse your mistakes, etc. (I've watched a very large amount of educational videos and streams in the last few years too, of course.)
Whereas my main has hopeless "
gg go next" for about 2/3 matches. I examine all my losses in the replay viewer (a fantastic tool) to see what big mistakes I've made and what was happening overall. They are predominantly a clear case of team diff and/or enemy dps 'smurf' carry. (I'm fairly confident at recognising high plat+ rank snipeDoom, etc, play.)
I'll explain more about my personal experiences further down. First I'm going to run through what I've been able to find out about the OW competitive matchmaker, because I think this might explain the
main reason for my situation while also being relevant for many other players. But
not most - this sub definitely provides a whole lot of useful education info that clearly helps a lot of players. Good job! But many here can be a little
too dogmatic in assuming that
only personal failings can
ever be a major issue preventing rank climb. I want to question that assumption, and maybe some others (with evidence).
MMR-SR disparity rank trap hypothesis:
[
Edit:
please check out Kaawumba's more detailed explaination of season 24 matchmaking and SR adjustments, with many more relevant source links, on the official forums.]
From what the devs are prepared to tell us, I quoted and summarised the bits I thought important at the end of
this exchange, on a thread in this sub a week ago. So I'll only link the sources here. Only points 1 to 5 (and maybe 6) are fully relevant to this theory, but I've included more, because I'm verbose...:
What we know for sure about the competitive matchmaker:
- Jeff and the devs have emphasised that it primarily uses hidden MMR to match players together. [A 2016, B 2016]
- MMR is not equal to SR. It's a somewhat complex 'player quality' metric, but it maps skill on a -3 to +3 scale (for the majority of players). [C 2019]
- It tries its best to make every match's predicted outcome as close to 50/50 odds as possible. The cut-off for making a match is 40/60 percentage odds. More SR is given for a win overcoming bad odds. [D 2018]
- We all see that the average SR of each team is
always usually close. These will also generally be fairly close to your personal SR, if you are solo queuing. (This is more commonly divergent for players at the rarefied extremes of the rankings, with fewer peers to match with.) [Edit: But I now fully believe the devs when they say they don't use this for matchmaking, after a detailed look at some SR adjustments following decay. Credit u/Preloa.] - MMR [can sometimes] change more rapidly than SR usually can. And "will fluxate more rapidly the less frequently you play". Faster for band new players. [E 2020] (I've heard claims MMR used to be averaged across player stats for all time, but I'm dubious that this was ever so, or is still the case, if it ever was.)
- It considers your MMR relative to the enemy team's average MMR ("strength of schedule"). [E 2020] I think this means it will award less SR gain for winning matches below your rank (I've not seen this explicitly stated). Or rather, below your hidden MMR, however good it secretly thinks you are.
- First time comp placements have accelerated SR change, from a starting point of mid-gold. Accelerated SR change continues for the next 5 (?) games after first placements (at roughly 5 times the normal amount, ~125 instead of ~25).
- There's a minor performance based SR contribution (plat and below), of course. [E 2020] So hard carry players will gain a little more where they greatly outperformed the average for players on their chosen character(s), in matches around that rank.
- Grouping is considered advantageous, so it always tries to match same sized groups, with similar skill distribution, where possible. [D 2018] I think I've heard it hinted/admitted that other player quality factors might be used to counter-balance unequal group sizes? Please link me if you have a source.
- At the high end of the ladder, wins gain less SR the closer one gets to 5000 SR. Grouping is more limited and between season placement results are currently capped at 3900SR.
- Bonus SR gain for win streaks was essentially removed in season 11 [F 2018]. It may still exist, within mid-ranks, with raised requirement and lower rewards, so is not very relevant.
- Blizzard never analyse audio or text comms, despite their patent describing such. [G 2020]
- In season 3, the devs included an SR win/loss adjustment factor that often looked extreme and arbitrary (win 10 SR, lose 40 for no obvious reason). It was there specifically to realign the 'normal' curve of player SR distribution back towards gold. They apparently cocked up the maths at the start of season 2, making the range of SRs too compact and too high up towards plat. [H 2016]It seems to me that this one instance of arbitrary shifting of SRs may have created an ongoing misunderstanding, e.g. via this post. I doubt that, in general, SR is systematically shifted towards a player's hidden MMR with a biasing factor, like that. Not beyond fixing that specific issue, 4 years ago. (I've heard that the distribution's currently centred near 2350 as an average, rather than exactly mid-gold...?)
TL;DR - Crux of the trap theory: because the competitive matchmaker is
simultaneously solving to balance teams by visible SR *
and* hidden MMR, to create a predicted 50/50 odds outcome, it will
have to counterbalance higher MMR players on one team with higher MMR players on the other team. Or lower MMR players on the same team. I believe MMR is able to change at a much faster rate than SR, e.g. to balance QP within a few games.
[
Edit:
I no longer think SR is used for matchmaking. Nor that MMR could possibly be shared between QP and comp. But I still suspect QP MMR could be a secondary consideration for team balancing of smurfs, etc.]
I think it's set up this way because the devs had to accept smurfing as a reality. (BTW, I think 'smurfing' is mostly
account sharing in lower ranks.) There are a substantial number of players who deliberately
avoid being moving fully towards an SR corresponding to their true player skill. These players would *
entirely* dominate the outcome of matches if the system didn't try to counter-balance them. So it makes sense that any player performing a little better (or worst) should be able to tip the balance of a match, and thus climb (or fall). Even if they are not the carry.
The problem (I think) is that this could also create a trap, biasing matchmaking against players who are
legitimately under-ranked. With MMR significantly above their SR. (Though not massively so - I'm only claiming ~500SR, due to my known limitations.)
This trap definitely isn't the case most of the time, so when is it relevant?:
- Not for an account that's always playing a lot of comp - SR rank will gradually increase in line with actual skill (and MMR).
- But maybe leaving comp alone for a long time (after initial placements), improving significantly in QP, then returning to comp.
- Especially if one's skill is not many ranks higher. I.e. a GM could always carry in mid/low ranks. (Although, Forgiving lost 2 of his Reinhart unranked-to-GM placements. And I watched Tesla have to go full try-hard on Moira in low gold, after losing 2 games in a row that he felt were highly stacked against him.)
- For those strong on support/team play, but mechanically weaker, and less able to hard carry. Especially if they are counterbalancing under-ranked players who are more dps and mechanically oriented (getting picks).
- Player distribution is a normal (bell shaped) curve centred on mid-gold: Because new players are dropped in at the middle, like sand into the bottom of an hourglass, there's a gentle, persistent force pushing players outwards, towards high ranks and low. Top 500 players have often complained about getting clueless masters tanks in their games, some claiming this inflates their SR (gaining more from wins and losing less from losses - point 6, above). Surely a similar, but reversed, effect happens at the other end, drawing tanks (and supports) downwards...?
Sub-hyppothesis: the same MMR metric is shared between QP and competitive matchmakers (on a per-role basis). [
Edit:
I now doubt this is plausible. See end note.]
I feel my comp matchmaking, on my main, has been more manageable when I've used my alt account for all my quickplay, instead. Conversely, it was particularly hard this season, after playing a fair bit socially with my game friends, who average around diamond/plat. But this is harder for me to be absolutely certain about, because my capabilities
do vary a little, and the times I play at vary a lot, too.
Although, I see other players with under-ranked accounts talking about systematic bias against ranking up. Like this masters player,
here yesterday, reporting a gold account with 67% win-rate failing to gain SR. I think their case may be more about MMR being a *
lot* higher than rank, so it's expecting that they carry every game (point 6, above).
If the comp matchmaker were using a separate MMR stat, this would be far more limited to how far out of whack it could get. It could only project your skill based on performance differentials verses players of low skill. A top 2% player in silver is still just exceptionally good for silver; it can't say if they below in plat, diamond, higher. For accounts only playing comp (like streamers doing 'bronze to GM'), it con only go on comp performance. But I've not seen devs hint at any kind of MMR duality anywhere, so I expect it's shared (fitting with my experience, too). I'm keeping in mind that much referenced talk about the MMR system comes from back before competitive was first introduced [
e.g. 2016].
Note that the scale of your MMR-SR disparity (how under-ranked you are) would not necessarily correlate with the scale of smurf skill you will be balanced against. A super-smurf (say, MMR 3 ranks above SR) on one team, would often be counter-balanced by multiple under-ranked players on the other side (say, 3 who are 1 rank out), so the total MMRs sum up. If something like this happens, then moderately under-ranked players could expect to hit super-smurf enemies a lot more than players who's SR is close to their MMR. I feel I've been seeing high plat and above skill enemy DPS players in ~1/3 of my matches (in addition to major imbalance in a further 1/3).
Alternative/complementary hypothesis: high silver truly has higher skilled players than high gold. But then they'd need to be trapped there, by this or other factors. Or maybe 'skill' is too one dimensional; they may be (for example) better mechanically while much worst overall, due to entrenched misunderstandings. I think there's some truth here, but it's not most significant.
Back to my situation and personal experiences:
Quick biography - I'm 37 and studied degrees in physics, then systems engineering. Since then I've been chronically ill with
ME/CFS, and so have spent years in various game communities exploring and explaining counter-intuitive technical aspects. Hence I find this topic academically, as well as personally, interesting...
Anyway, I made my alt account at the end of last year, after being fed up of being stuck in silver (or below) for years. I placed my main season 1, when I was only playing socially, very casual. I was sure that I was now a gold level player, maybe a little higher on my main roles - Moira/main tank (from filling for years). I could see that, for one source, from looking at very many profiles of the players I matched against while solo queueing in quickplay (getting numerous plats and even some diamonds at various points). Yes, that's difficult to assess, with only 1/6 profiles visible, or so (and there might be bias towards higher ranks having more visible?).
But placing my alt totally vindicated me, to my satisfaction, at least. The matches on it were a breath of fresh air. Of course there was variability with some unfortunate match-ups. But it's a
different order of magnitude, with
most games more balanced than
any on my silver main. I think people don't appreciate the scale difference of the issue, when trapped players complain of things that they also see, just to a much lesser extent.
There's definitely an 'ELO hell' aspect, for me, too - I have strong comms and team play and relatively weaker, slower mechanics. So higher ranks boost my contributions a little more, too (e.g. flankers I call out sometimes get dealt with!). My main limitation is slow visual information processing - I'm dyslexic with a reading speed half what it should be. I just
can not read the kill feed while engaged in a fight and it takes me a little longer to see what's going on generally, especially in more chaotic situations. With ADHD-PI (and a little 'brain fog' from ME/CFS), my working memory can be weaker, so struggling to keep track what's behind me, too, when under pressure...
But I'm certain these deficits aren't the core issue with my main account's rank, because:
- As well as reviewing my play, I've previously looked repeatedly at my Overbuff stats, which were usually high or very high percentiles. (I've not linked it here because they seem substantially broken since role queue came out.) Below average only on Moira damage, as I play her balanced, not flanking or dps-ing as much as some, probably.
- I also placed high gold in the new open queue comp mode (on my old account). Where I ironically have better matches, despite hating the setup; losing games in the spawn room because it's impossible to figure out what team mates are good at, then having toxicity from people stuck on off-roles, etc. (Role Q was the best game change!)
- I know what it feels like to be out of my depth in a match, from playing QP with my friends who are all higher ranked (centred on high plat/diamond, one masters-ish). Also giving me good reference points for appraising under-ranked enemy dps in my games. (I've been having fun playing along with Jayne's new guess the SR segments.)
- My SR gains/losses are usually a reasonable 20-something; performance adjustments aren't my issue, just match outcome.
Note that all my matches have been solo queue, spread across many different days. Various times of day, due to my broken 26 hour long circadian rhythm with my ME/CFS (chronic fatigue). Avoiding peak times for kids (because their higher mechanics and lower team play may be unfavourable for me). Always warming up in Mystery Heroes and QP first, never playing comp unless I'm functioning well enough. Definitely never play drunk or tilted, stopping after 2 bad losses. Using the same, light, informative, positive comms (where there's anyone at all to talk to). No toxicity or ordering others about. I even tried one season (last year) of staying out of comms altogether, to be sure I wasn't distracting teammates (particularly with the EU language barrier). I still slowly dropped.
The only times I've consistently climbed have been:
- During placements, back when they were more/predominantly based on personal performance (I think?).
- When I've duo'd for a few days with a substantially under-ranked player. One who returned to game and quickly rose from gold/plat to diamond. Another, previously, who was a self confessed masters (tank) ranked smurf, who I believe was account sharing with a younger sibling. Both friending me, impressed with my comms and team play.
- When I avoid playing any quickplay on my main, only comp (see above), which seems to start giving me somewhat more balanced games (after a few). But this isn't really acceptable - I want to be able to use my main account for mostly casual games, with it's myriad unlocks, some now unobtainable, and three hard won golden guns.
Anyway. Please note that I'm not really looking for tips on playing better, currently. I've not included any replay codes, partly because I don't currently have any (after the last patch wiped them). Also, it would be near impossible to *prove* that my silver games are unfairly balanced. That would require you looking at a lot of games, to build up a full picture of the situation. Or that my play deserves to let me climb; I know I make plenty of mistakes, as everyone does. Because of my biological limitations I'm never climbing myself up as high as diamond, I'm pretty sure the pace and precision for that is well beyond me.
Edit 2020-10-24: as noted above, after comment discussion about decay, plus finding and reading more of thi
s fantastic forum post,
I no longer think SR is used for matchmaking. Nor that MMR could possibly be shared between QP and comp. But I still suspect QP MMR could be a secondary consideration for team balancing of smurfs, etc. Or there is some other secret analysis of play within comp, alone, to balance this kind of thing.