Jump to content

Tinfoil hat study

Recommended Posts

So I've been keeping detailed records of battles for about 300 games now, just shy. I've had the impression for a long time something shifty is occurring with MM and the teams I'm being handed. The only way to deal with impressions is to compile the data and actually look it over to determine if something is really up or you're just crazy. These battles are all solo, a standard mixture of tanks, neither exceptionally good or bad overall. Tiers ranged from 2 to 9 with average probably running 6 or so.


Since I'm at another rage quit I thought it would be a good time to compile the numbers to see if anything is showing. I need to learn Excel anyway, so two birds one stone.


I recorded the tank, the tier of game, XVM's WT prediction, the actual battle result, and which team had an advantage in top tier heavies.


I simply averaged 3 numbers: MM tier, XVM prediction, and then the actual result.


XVM's win rate prediction: an odd 55.24%. On the one hand, this is what I expected the number to average to. My mathematical intuition had this as the number before I even ran them. This seems rather bad for a player of 63% WR. Cue paranoia.


My actual tier vs tank (counting sheltereds at -.5 and +.5) actually came out to a mild advantage of .104. Overall I've been getting higher tier about 10% of battles. I felt this was occurring as well, it seemed as if I were indeed getting a little luck on teiring lately.


The weird part is win rate. Actually averaging up my WR for this period I won a surprising 66.0% of games. My intuition had that number set more around 62%. The gap between XVM prediction and actuality is at a whopping ~11%.


~300 is only a small sampling, but I think I have a useful take away here. Lots of people say XVM win rate ticks them off and here it is: it's showing I'm going to lose when I'm not. I've been basing much of my perception of MM and teams on XVM's ratings, and thus it feels like I'm getting hosed more often than it appears I am. At least in this sampling there doesn't appear to be any bias occurring in MM, just a bias being planted in my perception by XVM's attempt to quantify. I'm actually surprised at how how my WR for this period has actually been compared to my impression.


Ok, so, no more XVM winrate prediction.

Link to post
Share on other sites

XVM Chance to Win was designed with the average player in mind. It's been tested a bunch of times, with the same reoccurring result. Accuracy is strongest with an average player, and weakens the more that becomes an outlier.

Link to post
Share on other sites

Creepy. My win chance so far has been 55,02%, for a much lower stats-rated player:




By contrast, for a 48% 780WN8 player, win chance is actually higher than his win rate... at 49,61%


WARNING, small sample size:


Link to post
Share on other sites

I turned off the win chance guess a while ago. It just isn't very accurate in my opinion.

Over the course of a thousand or so games if XVM was guessing less than ~55% chance to win it was overstating (so if XVM said the team had a 40% chance the actual chance was ~10%) and for more than 60% was understating (for example my team won ~80% of "60% guess" games).

Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

  • Recently Browsing   0 members

    No registered users viewing this page.

  • Create New...