Quote:
Originally Posted by John Dewey
But a crummy fighter is still going to end up being a crummy fighter no matter who rates him.
|
Yes and no. It's hard if not impossible to rate lower end fighters with accuracy. Two things to consider:
1) The top (a 15 rated fighter) should be the best ever to lace them up. the bottom, a zeroed out rating should represent the absolute bottom. The guy that fought as a pro just once or twice and was blasted into oblivion each time. That guy (the very bottom) should litterally never beat the absolute best fighter representation. In real life, things happen but not this. The outcome of pitting these two should always have the 15 rating come out on top. The problem is that this isn't the case, run a 1,000 fights and that bottom rating will beat the top a few times. Not many, but the number should be zero (save a sniper in the crowd). If the top represents the top of the sport and the bottom demostrably doesn't represent the true bottom then the scale is off on a systematic level.
2) If you create 'average' ratings at each rating level, from 2 up to 15, each average level will beat the next lower average level a certain percentage of the time - it is quite consistant and makes 2 through 15 each act like true levels. There is a bit more of a spread between the bottom and top of what constitutes the level '1'. Not enough to squeeze a second 'true' level in but close. The zero rating has between 12 and 17 levels as measured by the winning percentage that represents a true level from 2 to 15. I've tried this out and tested it over hundreds and hundreds of thousands of test fights with template zero ratings. I find that at 12 levels within the zero rating, each has a little more than one true level separation and with 17 it's a shade under one true level separation. The difficulty is that the deliniated template ratings within the zero rating level are not as consistant in testing as the 2-15 ratings are. So, it's really hard to rate fighters that fall into this category with any sort of accuracy.
All that is to say that there are not 16 rating levels. There really are 27 to 32 of them. That means that the average fighter (the guy who sits at the mid point of all possible fighters in his class with half better and half worse) is likely a rating in the high zero rating range. That means that the vast majority of historical fighters out there that have yet to be rated probably deserve a rating just below the zero - one rating level divide and southward; that's a range where accuracy is nearly impossible unfortunately.
It took a long time for me to realize this about the way this game's rating's work unfortunately (Will did a lot of the testing leading to this understanding and I just probed the zero rating in the tests I did) - it does however explain some of the frustration I found when I used to try making ratngs myself.
The bottom rating isn't the real bottom when referenced against the top and half the fighters out there to rate fall into a range (high zero and down) that acts somewhat inconsistantly in the game engine and doesn't lend itself to accurate rating.