For the best experience use full HD.

Friday, January 8, 2016

Fleet Friday: A World of Warship Statistics Nerd Site



For Fleet Friday today, I want to talk about statistics. One of the things I enjoy about World of Warships if the ability to track how I am doing. To facilitate this Wargaming.net provides numerous statistics concerning every captain's game play. If you want to know how good someone who dissed you in game is you can look it up. You'd be surprised how many of them have statistics worse than yours.

But there is a problem with the statistics provided by Wargaming.net. They are a snap shot in time. There is no history. They don't tell you how you are doing over time. And frankly, I really want to know if I am getting any better or not. It helps to know such things on those frustrating days when nothing seems to go my way.

Fortunately, there is a web site that will provide historical context to your World of Warship gaming. It is WarshipStats.com, and it's plugged directly into Wargaming.net using the World of Warship API via your game login. It will show you many things about your game in either graphical or tabular form. It is a statistic geek's fantasy come true. Here are three charts I reference at least once a week.



Two of these charts clearly show that I am in deed getting better. The average damage chart can't. The big dip you see around the beginning of November was when I began playing Russian destroyers. Lower tier ships just don't have the damage potential of higher tier ships. So average damage to a player who plays a variety of ships as I do is not as useful.

So one thing that WarshipStats.com does not track is Ranked games. You can see from my WarshipStats.com generated signature at the top of the post I only played three games last night. That is not true. I played six. I don't know if the team will add tanked statistics later. I know the site Ship Comrade is working on a ranked leader board, so it is at least accessible through the API. Time will tell. The only bummer about this is it makes me statistics somewhat inaccurate. But then again, the World of Warships official statistics don't seem to take them into account either. I suppose it really isn't necessary. You can see how someone is doing just by looking at their rank. Right?

Anyway, personal statistics aren't the only thing you can get from WarshipStats.com. You can get tons of information about individual ships as well. I have recently started diving into this treasure trove of data, and it has been fascinating. And though there is no functionality to download the raw data, you can still use copy and paste to bring it over into a spreadsheet program like Excel. I did just that on Wednesday this week for the North American server data. With some minor tweaking, here's what I got.
Oh boy! What fun a statistics nerd like me can have with this! For example, have you ever wanted to know what the best ship in a particular class is? I know I have. Of course, you can be content with anecdotal information like your favorite World of Warships YouTuber stating the Mogami is the best cruiser in the game, but to me that's not actually knowing something. And doesn't such a judgement depend on what your measuring criteria is? Do you use average damage or win rate for example? Of course the answer is both according to what you think is important.

So is the Mogami the best cruiser in the game? Let's have a look. This is where you're going to need to learn the art of the Pivot Table. I will admit it's not one of my strong points, but as a statistics nerd I feel it is my duty to struggle onward. And frankly, the newer versions of Excel make it so damn easy. So first, let's isolate all non-premium cruisers by tier and compare their average damage. Here is the result.
Looking at the Tier 8 cruiser results, the Mogami is over 9000 HP average damage above the Admiral Hipper and over 15,000 (!) average damage above the New Orleans. Nope, no balancing issues there.

But how does that translate into victories? Just because a ship has an inherently higher raw potential for damage doesn't mean it wins more often. The skill of the player makes affects that a lot. The only way to know if the higher damage potential equates into more wins is to take in a whole lot of data from all captains so as much noise is eliminated as possible and look at the average win rate. This is what that looks like.

In the case of the Mogami, it seems higher potential damage does correlate to a higher win rate. This does not always hold true. The Cleveland has the highest average damage or it's tier, but the N├╝rnberg has a slightly better win rate. I think we can say the Mogami is the best Tier 8 cruiser in the game.

And this is just the tip of the iceberg with what you can do with the raw data provided. Have you ever wondered what the most popular ship is in World of Warships? I have. Mind you, it doesn't keep me up at night, but I have on occasion been distracted wondering what gets played most. Is it the Tirpitz which seems to show up in every match? Or the Cleveland for the same reason? Of course, my view of that is tainted by the tiers I play. But with data on all ships I can figure it out.

To do so I split the ships into standard ships and premium ships. I did this because I needed to pair down the data rows and I did not want to break them out by class or tier. I want to see them all at once, and separating the purchased ships from the free ships wasn't such a bad thing in that regard. Even with that done, the chart for the standard ships was freakishly enormous. I have a 30" 2560x1600 monitor and it was still a struggle to get it to fit and be legible. But I think I got it. Here's my ginormous chart (click here to see the full size, full resolution version,) with the much smaller premium ship chart below it.

These two charts show two things. I figured there were two measures of how popular a ship is, and one of them is directly affected by how many players actually achieve that tier of ship. The bars show the total number of games played on the North American server by all players for each ship. That is the one affected by the number of players who get the ship. It would be possible to normalize the data across tiers as the total number of players with the ship is provided, and doing that is on my list, but I've a day job too. So for now, the total games played is really only relevant within tiers, though won can see just how popular the OP Cleveland really is. The diamonds show the average number of games played in that ship by all players. This measure isn't affected by how many people have the ship as it is an average of those who do. It is a direct measurement of how much a specific ship wants to be played, and that in itself is an indirect measurement of how popular it is. The one caveat to that is the number is affected by how long the ship as been available. The German and Russian ships will not have as high an average as they have only been out half (less?) as long.

So, of the standard ships, looking across the diamonds, it appears the most popular ship in World of Warships for those who have earned it is the U.S. Midway carrier. Of course, that could be because it's so OP, but hey, popularity is popularity. Tied for second place is the Japanese carrier Hiryu and the Japanese destroyer Shimakaze. Third place goes to the Japanese carrier Taiho.

The most popular premium ship by average number of games played is the Japanese cruiser Atago. A far distance behind it is the U.S. cruiser Atlanta, and right behind the Atlanta in third place is the Russian cruiser Murmansk. The Tirpitz only comes in at fourth place. YMMV.

Anyway, I hope you've enjoyed the charts. There will likely be more in the future. One of the charts I'm working on now is potential income earners. It would be ships with a high victory percentage as well as a high survival rate. Average damage and other statistics that earn credits and experience also need taken into account. That chart will be so busy I'll have to break it out by tier or class or both! I can't wait.

Lastly, if you play World of Warships please consider sending a donation to the WarshipStats.com team. They have done a lot of work for the community and I am very appreciative of their efforts. Their effort deserves to be rewarded. Also, if you have specific charts you'd like to see but don't have the time or software to create them, let me know in the comments. I'll see what I can do to provide anything you're interested in seeing. And until we meet in battle, sail carefully and have a wonderful day.

No comments:

Post a Comment

Be civil, be responsible and most of all be kind. I will not tolerate poor form. There will be no James Hooks here. We are all better than that.