Overwatch Data Analysis-Stats for Wins vs. Stats for SR- Support Edition

ANA Category Correlation Value Rank 1 2 3 1 2 3 Positive SR Off. Assists Healing Def. Assists 0.65 0.55 0.49 Positive Win% Def. Assists Off. Assists Healing 0.28 0.28 0.26 Negative SR N/A Negative Win% Deaths -0.20 SR v. Win% Off. Assists Healing Deaths 0.37 0.29 0.25

So let me explain how this works. I am listing the top 3 in game stats for each of the categories. I also have the correlation values for reference so you know if it is strong or weaker relative to the other stats listed. Ana's stats are pretty self explanatory in that assists and healing are reasonable to deduce as important to winning and in fact they line up to high SR players! Yay, score one for Blizzard.

However note the final column which shows the difference between win% vs. SR. It looks like SR is too strongly tied to these stats. When we combine this with the win% for Diamond and above players (for my sample set) you will find an abysmal win rate of 46%.

Player Rank Count of Hero Average of hero win% Diamond 44 46.0% Master 40 47.2% GrandMaster 25 46.8% Ana Total 109 46.6%

This suggests to me that SR gains and losses should be toned down a bit as it is a little to easy to get SR as her. But at least it appears to be balanced towards winning stats.

Lucio Category Correlation Value Rank 1 2 3 1 2 3 Positive SR Off. Assists Eliminations Def. Assists 0.58 0.45 0.39 Positive Win% Objective Kills Eliminations Time on Fire 0.57 0.53 0.43 Negative SR Self Healing Objective Time Healing -0.36 -0.28 -0.20 Negative Win% N/A SR v. Win% Self Healing Objective Time Objective Kills -0.55 -0.54 -0.51

Now we can start to see the problems, what happens if the stat that leads to winning games is not really getting captured or encouraged by SR gains? Not seen on this chart, defensive assists are still important for winning games, but it is not as important as objective kills. Also not seen is that offensive assists don't really seem to have a correlation to winning, so why is it being rewarded so much? My theory is that the SR gains may have been balanced to the old Lucio that had a humongous aura. Back then, the speed was more important to the healing. This new meta calls for a better balance between the two.

If we look at the SR v. Win% you will see Objective as something very important for winning, but completely ignored by SR. So you spent 3 hours holding the objective and scoring kills to win the game? Welp, that's too bad you better hope you left speed on the entire time and that your teammates were killing everything for you.

Looking at the negatives, players are penalized in SR for self-healing. What appears to be happening is that self-healing interferes with your ability to give out juicy offensive assists, since this stat is so important to SR, there is a penalty. This really makes me sad.

Player Rank Count of Hero Average of hero win% Diamond 43 52.8% Master 32 52.9% GrandMaster 27 53.7% Lucio Total 102 53.1%

You can see that Lucio's have a good win rate that isn't too bad. I would suggest that the number be a little bit closer to 50, but there are always going to be heroes performing at different rates compared to the meta.

Ok, now it's time for the controversy... Let's be honest, you scrolled straight down to this section to see what's happening with our favorite healer right?

Mercy Category Correlation Value Rank 1 2 3 1 2 3 Positive SR Time on Fire Off. Assists Def. Assists 0.66 0.43 0.36 Positive Win% Healing Def. Assists 0.24 0.23 Negative SR Objective Time Self Healing Objective Kills -0.53 -0.46 -0.27 Negative Win% Deaths -0.43 SR v. Win% Time on Fire Objective Time Self Healing 0.52 -0.49 -0.44

So there is a ton to talk about, but hopefully by now you can read this table and understand what is presented. We see here that Time on Fire is a very strong metric for gaining SR. How do you get fire? A 5-man rez is guaranteed to get you enough fire to put you over the line and if you can sustain it, it will be beneficial to your SR. Unfortunately the biggest delta between SR and Win% is also this same category. Essentially time on fire doesn't correlate strongly to winning. The data suggests that healing and def. assists are what leads to wins.

Looking at the negatives, SR penalizes Mercy's that play the objective and get large self-healing numbers. Honestly this one infuriates me. SR is actively encourage Mercy's to hide in the corner and wait for rezzes even though the correlation to win% is tied to healing instead. From an outside view, I would rather see the exact opposite happen to encourage more interesting gameplay (but that is just my opinion).

Looking at the deltas, Time on Fire is strongly tied to SR despite it not contributing to wins. Objective time and self healing are penalized. So now you know why Mercy players are hiding so much.

Player Rank Count of Hero Average of hero win% Diamond 67 50.7% Master 44 51.3% GrandMaster 18 49.7% Mercy Total 129 50.8%

Surprisingly enough, Mercy doesn't seem out of balance based on win rate. Unfortunately, the problem lies in the amount of players at high ranks. Based on my sample, there is roughly 30% more high level Mercy's then the other support heroes. This wouldn't be a problem necessarily, but when you combine it with the wrong stats being prioritized, it's easy to get matches with Mercy's who have been trained wrong by the system in place. But don't fear, that also means there are Mercy's who know whats up and are helping their teams win game. Give Mercy players a chance, they can win you games

Honestly, I got upset after doing this analysis. Something is very wrong with how SR works for Symmetra players.

Symmetra Category Correlation Value Rank 1 2 3 1 2 3 Positive SR N/A Positive Win% Eliminations Damage Objective Kills 0.47 0.42 0.32 Negative SR Solo Kills Final Blows Time on Fire -0.57 -0.54 -0.49 Negative Win% Deaths -0.28 SR v. Win% Eliminations Final Blows Objective Kills -0.87 -0.82 -0.74

This is just bad, out of the basic stats I tracked, there was no stat that correlated to gaining SR. Potentially there is a stat like turret kills or players teleported that helps out Symmetras, but my basic stats couldn't find anything. Meanwhile there are a few stats that show good correlations to winning like Elims damage and obj. kills. So then what is the SR system doing?

Symmetra players are penalized SR for getting solo kills final blows and for time spent on fire. These are very strong negative correlations that make me wonder... How in the world can players on fire get penalized???

What's worse is when you look at the deltas. Eliminations, final blows and objective kills were nearly guaranteed to wreck your SR even though it can contribute to winning. One potential answer for this is that high level symmetra play just doesn't work as well. But let's look at the win rates.

Player Rank Count of Hero Average of hero win% Diamond 42 61.4% Master 22 59.7% GrandMaster 8 55.5% Symmetra Total 72 60.2%

60% overall plus 55% at GM level seems pretty good. What isn't good is the player count, so many Symmetras are stuck at low level. Is that an effect of the system or is Symmetra no good? I am leaning toward the first one.

Zenyatta Category Correlation Value Rank 1 2 3 1 2 3 Positive SR Def. Assists Eliminations Self Healing 0.50 0.43 0.43 Positive Win% Eliminations Time on Fire Off. Assists 0.53 0.41 0.40 Negative SR N/A Negative Win% N/A SR v. Win% Objective Kills Healing Deaths -0.40 0.24 0.23

So this one was a little bit interesting, but the good news is that this hero wasn't as crazy as the rest of them. Eliminations ended up as important for winning and for SR which is great. Self healing also makes sense for SR gains as ultimates are the only way for him to self-heal (shield healing or picking up med packs doesn't count).

Nothing in the negative category, potentially I need to look at more stats, but that was a bit odd.

It appears again that despite Zenyatta's best efforts to save the point with objective kills, he is not rewarded with SR for doing that. Not shown on the chart, objectives kills had a correlation of .37 which is strong enough to show that it leads to winning.

Player Rank Count of Hero Average of hero win% Diamond 37 56.3% Master 20 59.0% GrandMaster 16 57.5% Zenyatta Total 73 57.3%

More research on this is needed for other heroes and with different player lists.

Ultimates usually had a strong impact on SR even if it did not really lead to winning.

Objective kills and objective time almost always were neutral or negative impacts to your SR.

Stats related to winning rarely lined up with stats related to high SR.

So that's it... hopefully you made it through to the end. But some final thoughts