Talk:Test Scores

Just a note, the notion of "scores" that you've developed are called "Expected Values" in the field of statistics. You can perform operations like +5% on the expected values directly. An expected value of 20 with +5% on the base rolls behaves like an expected value of 21. Rounding makes things a little more nuanced but not significantly so.

Read up on https://en.wikipedia.org/wiki/Expected_value#Finite_case if you have time and interest. Thanks for contributing! Thebv0 (talk) 18:23, 27 February 2019 (UTC)

--

You are correct that the scores are based upon expected values (since that allows comparisons of apples to apples and makes the meanings of scores more intuitive and experientially referential), however, because I don't know for sure how bonuses like "+5%" are actually applied in game...

Do they
 * 1) Simply increase the overall number of successes as you indicated - or the number of dice rolled, which is effectively the same for expected value / "score" purposes, in which case, yes, the base score can simply be multiplied by the bonus (as I indicated). However, if they...
 * 2) Increase the success rate of each die (ie, a +10% means a "Strong Die" succeeds 5/10 times and a "Standard Die" succeeds 3/10 times), then the score proportions need to be adjusted accordingly, and bonuses would increase scores (as they're currently calculated) with more Standard Dice more than comparable scores with more Strong Dice...

My "Scores" can't reliably take into account these bonuses (and potentially other mechanics, like caps, I may not be aware of), so they aren't actually the Expected Value (ie, they do not reflect the actual number of success you'd see in all cases), so I didn't want to be misleading. By calling them scores, I make it clear that these are approximations that are useful for making comparisons, but don't necessarily reflect the actual game experience.

Btw: I have a degree in Mathematics and was on my way to becoming a math teacher before health problems prevented that, so I am quite aware of and comfortable using the material in the link you provided. :)

And I actually considered figuring out the deviation for several sample dice pool counts so I could provide accurate samples of actual "win/loss" rates of a few different score comparisons, but just didn't want to spend that much time on it. :) (If you'd like to create such a table, I'd welcome the additional detail and accuracy - I'm not 100% sure that a 1 point difference in scores is as significant as I assume it is!)

Thank you so much for your interest in my article/guide and your feedback!

--

Oh nice! I hope I didn't come off as offensive. I didn't know you had a serious math background (certainly more than I have!) The 5%+ bonus applies to the number of dice rolled, not the success threshold per die, if that helps you. I did some deviation stuff a while back, I'll post it up here if I have time later and I'll have you check my work if you'd like. Thebv0 (talk) 18:23, 27 February 2019 (UTC)

--

No offense, genuinely appreciated your interest! Just wanted you to understand why I was using scores instead of a more technical explanation, and what my background was in case you wanted to get even more detailed in the discussion. :)

So percentage bonuses just straight up increases the number of dice rolled (and therefore the Expected Value and Score). Do you happen to know how it handles fractional dice (does it floor the last die, ceiling it, or apply a fractional chance on the last die - like the last half standard die would have a 1/10 chance of success). Also, do you happen to know if it rounds each contribution to the pool separately, rounds the aggregate, etc. (like 24.4 from Agility + 13.3 from piloting ?= 24 + 13, or ?= 37.7 -> 38, or... Also, are multiple bonuses applied additively or manipulatively? Ie, do two 100% bonuses end up as a 200% increase (x3), or a 300% increase (x4)

One interesting thing about it increasing the actual dice rolled is that it'll not only increases the expected value, but should also decrease the deviation, giving even more stable expected values - that's actually pretty interesting.

Also, I'm curious if you happen to know how ties are resolved - if both parties get say 7 successes, is a coin flipped, does it favor the player, the defender, etc.?

BTW: I also have degrees in Computer Programming and Physics if you want to geek out about either of those at any point. :)

Thanks for the really awesome feedback, I'll update the page with this info. Thanks. If you (or anyone) also happen to know how the "+X% based on the Engine Score comparison (Speed/Agility)" is figured, I'll include that as well. And if a Targetting Assist Module with its +6% Accuracy increases the entire Attack Pool or just the weapon's Accuracy contribution (or something else). Thanks!