Due to technical reasons (BG rank reward delays) and some human issues (jetlag) I was only able to finalize my analysis, or rather my initial analysis of the Battleground Blitz today. So a day late, here's where we ended up.
Final score: 1,193,487,854, just short of the 1.2 billion milestone, about where we were projected to land. If we eliminate the Stark donations, actual player effort looked like this:
Adjusted final: 1,023,487,854.
I should say, that removes the 170 million donations that were obvious: according to Kabam Crashed, there were 200 million total donations. I have a hunch where they are, but they are small enough that it would be difficult to prove where they were injected, and this is compounded by the fact I spent half the event in Japan, so the gaps in my data due to sleep encompass the entirety of when donations would be made (if you don't collect data bracketing a donation closely enough and the donation is small enough, it can get "averaged out" of the data). But if we remove an additional 30 million points from the player total, we get a final player contribution of 993,487,854, just under one billion points.
Here's approximately how many points we put up per day:
Day 24 is when we reached the one billion milestone. On average, points per day trended upward until that day, then dropped off somewhat. I think if those other 30 million points were not included, players would have still reached one billion given that. Whether players would have reached one billion with *none* of the donations is harder to say, because the donations did not just add points, they improved sentiment. We can see how player activity jumps upward dramatically in the lead up to one billion on days 22, 23, and 24.
Because daily activity bounces up and down it can be difficult to see overall trends. One thing we can do is remove one source of daily fluctuation: objective reset days. On days where BG objectives reset activity was always higher on average than on the alternating days. We can calculate the average improvement in scores and remove it from the data to try to see what the overall trends were outside of that signal:
Some interesting patterns emerge. We see a relatively low level of activity in week one, until there's a sudden dramatic jump upwards in activity. Then things settle back down more or less in week three, with scores slowly trending upwards, and then another jump upwards peaking on day 24 when we hit one billion, then a significant drop off again. My guess is that we're seeing four separate things going on here. First, an overall slow rise in activity from week one to near the end of week four, corresponding to roughly how most BG seasons go, with activity rising throughout the season. Second, a big burst of activity in week two as players start accumulating large amounts of elder marks from unlocked milestones they want to burn off, that lasts until most players have used most of them. And then a big push near the end when the one billion milestone is in sight. And finally, once it is achieved, players start slowing down or coasting from that point as they reach the 5k minimum to unlock all of the milestones.
So just exactly how many players participated overall? Well, I specifically put up 4 points (the bare minimum) on an alt to try to get the lowest possible rank. That rank was: 314500. That implies at least 314500 players participated (here I'm using "players" and "accounts" interchangeably for reasons I've discussed previously). It could be slightly more, because it is possible there were other players who scored 4 points and ranked lower than my alt, but I doubt it was many of them. This number is slightly higher than what I believe to be the previous highest participation level in BG (~300k).
Of these, at least 119,723 players scored at least 5k points and earned all of the unlocked milestones (5011 points received that rank). That's way more than I thought would score that high, and implies that it wasn't that hard to so do: even lower progress players had to be doing it, because I don't think we have 120k Paragons and Valiants grinding out BG. Especially considering how many players made it to GC this season:
98311.
The number of players in Uru 3 at the end of the season was 89611. As there are 8500 players in the ranks above that, this means 98311 players made it to GC. As far as I am aware, this completely blows away all previous GC brackets. I don't recall seeing more than 40k in GC ever, and it is typically closer to 30k. And from my experience in GC with both main and alt, it was a wide range of account progress in there. My main made it in about half way through the season, and my TB alt marathon grinded into it from zero in the last week, getting in two days before season end. I'd say the average account in Vibranium and Uru 3 was somewhere around mid tier Cav strength.
There was some discussion on the forums if people would continue to grind in GC, or if they would grind in VT and then just park in GC. We can actually determine to some degree if that was true or not by looking at ELO distributions:
This shows how many players had each ELO rating in the lower tiers of GC: mainly Uru 3 and a bit in Uru 2. An interesting thing to note is that when a player promotes from Vibranium 2 to Vibranium 1, they actually have to play at least one match in GC to qualify for Uru 3 rewards. So one is the minimum number of matches a player should play in GC (some do play zero, but then they don't qualify for GC rewards, so most don't make this mistake). If you play exactly one match, there are only two possibilities for your final rating: zero, or sixteen. So of course, these are the two highest ratings populations. The next most common rating is 32, which most players get for two wins in a row.. Then 48, and then it quickly gets blurry because some wins don't generate exactly 16 ratings points and some losses don't cost that amount, and the scores start to smear out as players play lots of matches.
But the number of players who play exactly one match *must* be less than the total number of players with zero and 16 rating. In fact, it must be less than that because there are more ways to score 16 than just win one (as well as zero). If we look at all the ratings around 16 we see there's a bump at 17 and 18 with about 2300 players in each bucket. They are players that won two and lost one, or similar, and ended up off by one or two ratings points. We can assume that there are probably a similar number of players who arrived at 16 by similar multiple matches, which means of the 7086 players with rating 16, about 4700 got there winning one match and the rest got there by more complicated paths. We then assume that for every one of those winners there was a loser at zero, and we end up with a rough estimate of about 9400 players that played one match in GC and then quit.
9400 out of 98000 means only a tiny fraction of players did this. Probably because many of them were not at 5k points yet, so had to keep grinding, but the implication is the playerbase did benefit significantly from the GC point bonus, because there were a significant number of matches being played in GC. In fact the rate at which the ELO curve quickly smooths out suggests the average number of matches being played by GC players was not small. The more matches being played, the more quickly ratings peaks get smoothed out in the data. And in fact by ELO 64 - roughly equivalent to four wins - the "resonances" around 16 basically disappear:
Overall, the data allows us to answer some basic questions about participation.
Did the realm event encourage more players to participate in BG? Yes, it seems so. But probably only slightly more than has participated in the highest participation seasons to date. Regardless, 314500 players is a lot of players, especially considering the game has perhaps just under a million players, and fewer than 700,000 players are likely to qualify to play BGs (you need to be uncollected or higher, and we have on the order of 785k players Proven and higher). Out of all the players currently playing the game about two thirds qualify to participate in BG, and the event drew in about half of them. We may simply be running into the practical limits of how many player Kabam can encourage to do anything. We have a diverse player population, they aren't all going to want to do the same stuff, and some will actively refuse to do certain things.
Did it encourage participants to play more? Yes, a lot more. More than twice as many players entered GC than at any time in the past as far as I am aware, and that's a strong proxy for overall playing and scoring. The average player scored 3159 points. This is a pretty high level of activity compared to previous seasons. Even if we hand the average player all 700 points from 48 hour objectives, that still is about 55 matches in VT at 50% win rate (2/3rds of the players remained in VT for the entire season).
It is interesting to me that the average player scoring can be replicated by a player playing about four matches every other day and scooping up most or all of the periodic objectives. I'm not saying the average player did that, just that such a player would fall about where the average player performance was. The average player performance also did not come close to exhausting elder marks: 3159 would have unlocked milestone 13, and with it 5200 elder marks. That's enough for about 58 matches. This is more than the average matches played. This suggests to me the bottleneck to increasing player participation is not rewards, but time. Players were spending only so much time on BG, even to the point of throwing free marks - and free rewards - away.
Would I consider the event a success? Ah, that's the question. But the real question is: success at what? What was the event even trying to do? We can't judge if the event was a success or not, without first answering the question: what was it trying to do in the first place. That's not a data question, so I will leave that one for another post.
