House Rule Review #1: Character Creation

This is the first in a series of posts analyzing some house rules for d20 fantasy gaming that I wrote and later abandoned. I debated how to approach this topic and what information to include, so it took longer to finish than I initially promised. In the end, this post became more complicated and the series will now be only three posts. This is because the material from the fourth chapter, "GM Tricks" works better when integrated into the other posts than when presented in its own post.

When I first read the D&D 3.0 core books, I liked the fact that the designers included a point-buy system for ability scores. D&D 2e was designed on the assumption that everybody would roll their characters randomly, and it was hard (even for the designers of Player's Option: Skills and Powers) to come up with a point-buy system that didn't make life miserable for classes that had ridiculous entry requirements, like the ranger and paladin.

However, my excitement was tempered by the fact that the number of points given to create a character in this system seemed too low. It just seemed like 25 points was too few to make a really heroic character. The characters I made with 25 points in that system also seemed to pale in comparison to most characters I randomly rolled using 4d6-drop-the-lowest.

The internet agreed with me. In a thread on the old WotC boards dedicated to crunching the character creation numbers, one user ran a computer program that found that, when the 3.x hopeless character rule was taken into account, the average rolled PC's stats were worth 29-30 points on WotC's table.

When WotC moved on to 4e and another company took up the mantle of "traditional" d20 fantasy gaming, I was eager to see what happened to point buy. It turns out that the PRD "purchase" system uses a table that scales more harshly than the old WotC table. Furthermore, there seems to be no relation between the default random method and the Standard Fantasy point value. In other words, 15 points on the PRD table is as bad as 25 on the D&D 3.x table.

My Old Solutions

In my abandoned house rules, I wanted my random and planned character creation methods to be roughly equal in power level. The goal was to have different power levels, each with a random method, a point-buy number, and a default array, with the point-buy value and the array roughly equal in power to the average character created by the die roll method. It was an ambitious undertaking, and I did a lot of math and a lot of online research into sites like this one to make things match up.

In the end, I got frustrated trying to come up with default arrays, especially after I decided to try to give two per power level: one with no weaknesses and another with a weakness and a higher top ability score. I decided it would be easier to accomplish this design goal in a system that didn't have the 3-18 range of scores, and at that point, I was basically making another game, and not just houseruling an existing one. Of course, I was almost going down that road anyway, since I made my own point-buy table for those rules.

I abandoned the idea of making my own point-buy table for my potential campaigns, since I didn't think I'd ever find a group of players who would try it. I also abandoned my first system for figuring out how many points to give for each power level. However, I've since come up with a new system for that.

My New Solution

The key to my new system for finding point buy numbers is calculating the cost of a spread of ability scores centered on the average for a given random method. The weakness of even the most thought-out point-buy systems was that they calculated the cost of buying the average ability score six times, and rolling a character rarely gives you a set of scores that close together. Thus, those systems create radically underpowered characters.

The system I finally settled on is described here. As an example of the system in action, I will find the point value of the standard 4d6-drop-the-lowest method for the PRD Purchase Table (spoiler alert: it's not 15 points). However, this method works with the SRD table or any other point-buy table for a d20 game.

1. Find the average die roll of your random method. You can use any number of online tools to do this. A simple tool can be found here. If you're comfortable writing code, you can also use AnyDice. The average die roll for the standard 4d6 method is 12.24.
2. Add the point values for the whole number part of the average and the three scores above and below that number. For 12.24, we drop the .24, keep the 12, and add the point values of 9, 10, 11, 12, 13, 14, and 15. These values are -1, 0, 1, 2, 3, 5, and 7. The total is 17.
3. Multiply the result of Step 2 by 6/7 and round the result to the nearest whole number. In our example, 17 x 6/7 = about 14.57, which rounds up to 15.
4. Repeat Steps 2 and 3 for the number above the one you used in Step 2. In this case, the next higher number is 13, so we add the point values for 10, 11, 12, 13, 14, 15, and 16. Those values are 0, 1, 2, 3, 5, 7, and 10, for a total of 28. Twenty-eight x 6/7 is 24.
5. Multiply the fractional part of your average die roll by the difference between the numbers from Step 4 and Step 3. The difference between 24 and 15 is 9. Nine x 0.24 = 2.16.
6. Add the results of Step 3 and Step 5 and round the result to the nearest whole number. In our example, 15 + 2.16 = 17.16, which rounds down to 17. Our point value for the standard character creation method (4d6, drop the lowest) is 17.

This method has a couple of potential weaknesses. While it generally gives more accurate point values than official sources for any d20 game, it still somewhat underrates point values when used on a steeply scaling table (like the PRD table). However, depending on how highly you rate the power of picking your scores, you may see this quirk as a feature, not a bug.