I'm going to take a minute to wrap up what I started in January. It will include a pretty lengthy explanation of the math I used to construct the weights, and the reasoning behind the math. If you don't care about the math, you can skip the rest of the thread.
I don't care about math. Where are the weights?Here:
https://drive.google.com/open?id=1MxHipSv4rfFQr-zpzWoiJN4r8a4QYhUJWhat is this math all about?Starting with one of the 0.4.5 release, there are at least two attributes that are are calculated in the game mechanics based on a parabolic rather than linear equation – Speed and Passing Accuracy.
Yes, I know that's a mouthful. What that really means is that the attribute listed on the player card may or may not be as good on the field as the card implies.
The first problem here is that the player card implies a linear relationship between attributes from player 1 and player 2. Obviously a player with 75 SP is slower than a player with 90 SP, but how much slower is he?
If the speed calculation was strictly linear [x = y], the 75 SP player would be 21 calculated SP points slower using the distance formula than the 90 SP player [(15^2 + 15^2)^0.5]. In a parabolic system [y = -(x^2 – x + 100)], his calculated SP is much closer to 15 points lower than the 90 SP player. (This is also why in the 0.3.x – 0.4.2 days, 90 SP players were at so much an advantage over < 90 SP players, and the Go route with it's SP vs SP calculation ruled all the leagues.)
NOTE: I'm only using the above formulas as examples. I do not know what the actual SP calculation is.
The second problem is that weights are percentage based. So even if you understand the calculated value is different from the card value, you have to set your weights based on a linear scale … unless you take the time to determine ratios.
Prior to 0.4.5, I tried a number of different ratios (all of which are on this thread), and the corresponding percentage point that went with it. In and after 0.4.5, I will be using a parabolic equation to set ratios.
Rather than give a full explanation for this, I've included a chart in the weights that I have been using. The basics go something like this:
1) Determine the player's primary attribute
2) Determine how much more important (factor) that attribute is from all other attributes
3) Determine the number of secondary, tertiary, and auxiliary attributes you want to weight
4) Find the corresponding percentage value based on the number of attributes and the importance factor (so a WR in my system has Speed factored at 8, which means all of the players secondary attributes have a much lower percentage weight than if Speed was a factor of 2 or 4)
Wrapping up this threadI'm taking the time to post this because I'm closing down this thread. I don't care if JDB permanently locks it or not. I will just not be responding to it anymore.
This will be my last post in any of the public forums. I plan on playing in a few leagues, but the only forum I will be actively participating in will be MFN-1. In the past, I have been publicly criticized on the MFN podcast and in multiple forums for trying to help all players figure out how to play the game, and have been actively accused of making the game worse for volunteering my time in MFN-1. I have no time for it anymore.
If you are interested in game mechanics or player evaluation based, you can PM me.
/thread/