I mean even removing just one players stats, the outlier company from the terror list, from the total win loss ratios of axis and allies. This changes the balance from 58/42 to 54/46. A single player had an 8 point W/L ratio impact on the entire mod. This means that we have to take great care in how we use what numbers we have, and not quote them like scripture blindly.
These stats are fairly worthless without any proper context or sourcing. Also, once again, given the small number of players in the community, individuals can have a huge impact on W/L ratio of each side. It is also incredibly irresponsible for a member of the Dev team to post such statistics in support of an argument they are making to people who do not have such access to the raw data. Especially given that the dev in question has been identified as a major statistical outlier in his own data set. I would even go as far as to call it an abuse of power. As to the points Robie made. First, you cannot compare a game that has, on a good day, about a dozen to two dozen matches, to a game that has literally thousands of matches a day. Matches that are auto-matched to boot. The margin of error is just too great because of the low data set. I mean even removing just one players stats, the outlier company from the terror list, from the total win loss ratios of axis and allies. This changes the balance from 58/42 to 54/46. A single player had an 8 point W/L ratio impact on the entire mod. This means that we have to take great care in how we use what numbers we have, and not quote them like scripture blindly.
This is an unbelievably high win rate for this many games played. Past 100-150 range it should start to even out. In League of Legends if a champion has a 42% win rate its considered a dead champion.
Quote from: robieman on May 08, 2019, 07:39:37 AMThis is an unbelievably high win rate for this many games played. Past 100-150 range it should start to even out. In League of Legends if a champion has a 42% win rate its considered a dead champion.If you plot the cumulative wins over the war, allies were winning at 150 games. a meta developed from around ~175 games in, that has seen the 100 or so games yield epic axis winrates around 70:30, its a very recent development. The range of cumulative axis win rate this war has ranged from 44% to its current 58%. I would take any numbers you can derive here some scepticism.*All based on Razers sheet.
How exactly are we supposed to use a set of data that is heavily skewed, massively incomplete and potentially misleading? Would you get on board a 737 if your pilot told you that someone had smashed his instruments the night before, but its okay, hes sure he can manage the flight with a compass and a road map from 1954?If you would not, then why would you say - for sure and for certain - that these stats are our best bet? That we should cling to them and follow their wisdom. Even and especially when it is clear, for instance, that there are some massive outlier users in every category of the data presented alone (for instance, 5 out of 12 Terror doctrine companies that exist right now have statistically exceptional performance - that is, a more than 20% deviation from a 50% WLR - some for the worse, some for the better.)