League of Legends Wiki talk:Discussions/Cost Efficiency Pt 2 Electric Boogaloo/@comment-1330314-20150120082714


 * Road Blocks:
 * 1) Increasing or decreasing gold efficiency on higher-tier items is something you'd notice regardless of the baseline you selected. If your base item is 100% gold efficient on a certain stat, and item X is 200% gold efficient on that same stat, then switching the baseline to item X would make it 100% gold efficient, and the previous base item only 50% gold efficient.
 * 2) I think this would be better elaborated further down, but I think the most basic possible items are the best baselines for gold efficiency because no matter what more expensive item you'll build, you're always going to build that basic item first, or at least have the means to do it. No matter what stat you're building, you're always having to go "through" the basic item, whether due to its build path or due to the gold you could've spent to have that stat alone. Therefore, the priority order should always go to the lowest-tier item possible, because you could technically include it in every other item. This may not qualify as a fact in the same way as a gold value officially listed by Riot (which would be the prime reference if it were to happen), but I do believe there is justification in setting up the current gold value system as it exists now.
 * 3) This may raise the question as to why the model uses the least gold-efficient basic item (why take over  as the baseline for CDR?), and I think this ties into the above: with items like  and, which technically qualify as basic items, I think the priority should go to the least gold-efficient item for a number of reasons: first, it brings the baseline of components down to the most basic item in terms of both cost and build (you can't buy a B. F. Sword as you enter the game, but you can get a ), and second, it means it means we don't have to make changes when we run into a ,  or upcoming  situation, where turning more expensive basic items into non-basic items would require setting a new baseline. As these components are the least cost-efficient among all equivalent basic items, it makes sense to standardize that to composite items, which is why Kindlegem is picked over Fiendish Codex. Picking the cheapest item, rather than the least cost-efficient item, would simply create a host of completely cost-inefficient items, when the idea is that no matter what you're buying, you're getting your gold's worth in countable stats (which is within the realm of gold value) or uncountable bonuses.
 * Guideline Questions:
 * 1) Starting with the definition: my definition of gold value is the base amount of gold I'd expect to pay for a stat. Gold efficiency is how the gold I expected to pay for an item's stats alone compares to its actual cost. As an average player, my thoughts on how much a stat is worth would immediately go to the first item I'd see on the shop (i.e., , etc.), and as I'd fill out my build I'd expect to have more and more stuff, whether they be stats or unique bonuses (not a matter of gold efficiency, just a matter of expectation), therefore the baseline should, as a general idea, start low and go up.
 * 2) A stat's value should represent the amount of gold you'd expect to shell out for it. It's like a power budget: it means you should always expect your gold's worth of stats on an item, and if you don't, that can be attributed to its uncountable bonuses.
 * 3) Whichever effect that has cannot be reduced to a countable stat should not be counted, but the rest is fair game, provided there is an adequate enough description of their context ("Unique" doesn't need much, if any explanation, whereas auras would probably do with more in-depth presentations). This is not a unanimously shared opinion, and has sparked a major debate before, but I think that, ultimately, all we're looking at is stats, and parceling items out according to countable and uncountable effects. The duration or conditionality of an effect may not be something we can slap a gold value onto, but the stats associated with that effect can.
 * 4) Absolutely, I think the system should be able to take any item, look at it, separate the countable from the uncountable, and plop out a detailed description statistical efficiency. I should probably elaborate below, but I'd like to turn my gold efficiency template into something more automated, with a set of rules programmed into it that would take in any item and output its cost analysis.
 * 5) I agree, and I think this can actually be applied to effects that most people would not even think of assigning a gold value to. Conditional movement speed bonuses, such as the one on, can be assigned a gold value, and while that may seem ridiculous I think is proof that the unique effect has a tangible impact on certain champions' power and even other stats. In fact, I think the system should be fully compatible with any champion, regardless of their scaling: while gold value/efficiency is not a measure of a stat's effectiveness on any individual champion, if we had a new champion that, say, converted a new stat into another (health regen to AP, as a random example), then our cost analysis system would be compatible with that and would point out items that could have a significant impact on them ("could" and not "would" here, as it is not meant to direct build choices, which is in itself the complete opposite of theorycrafting).
 * 6) The cost analysis section of any item describes its "budget": how much of its power is placed into stats, how much isn't, and where that power goes. It should directly account for all of an item's stats, which doesn't have to just include the wielder (so auras should also be valued on the stats they provide others with), and explain if necessary how those stats come into play if they're variable. Whatever the analysis cannot assign a gold value to, it mentions as an uncountable part of the budget. Its purpose is to explain how exactly an item's power is distributed, and how that power comes into play.
 * 7) With the framework mentioned above, yes. One should not expect a purely inefficient item, and if one such item existed, it would be something you could replicate better with the same amount of gold (assuming enough inventory space). As cost analyses list both what they do and do not account for, they should indicate exactly what they're not counting.
 * 8) A stat is a value I can expect to have the same properties regardless of how I use it. 10 AD differs from 10 on-hit physical damage because AD can factor into crit or ability scalings, whereas on-hit physical damage, unless specified otherwise, does not. If an effect has the exact same properties and functions as a stat so as to have no differences, then it can be assimilated to that stat.
 * 9) Classification simply means taking in stats as stats. Whatever is uncountable or outside classification is specified as such, and therefore is separate to the stats themselves, which can always be analyzed and evaluated. In the event that a stat is special-cased on an item to be "different" (i.e. the now-defunct 's attack speed boost not counting into ), then that is indicated, but the value is nonetheless accepted as its indicated stat. Classification matters to me in the sense that it helps separate the countable from the uncountable at a very detailed level, and helps establish a clear reasoning for what is and isn't acceptable to cost-analyze.
 * Testing Questions:
 * 1) Wrong on several levels. First off, AP is already accounted for by, so the item's gold value per level wouldn't change. Even if the item were free (which it is), there would be no point in using it as a baseline as it's a champion-exclusive item, and makes therefore as much sense to establish as a baseline as 's old aura stats (which were also free). The per-level component shouldn't matter, as it doesn't change the value of the AP itself (the AP is countable, the per-level restraint isn't).
 * 2) Were I to cost-analyze the item myself, I would list the gold value the item gained with every level, and give a corresponding analysis of its efficiency ("this item is X% efficient at level 1, and gains Y% efficiency per level for a maximum of Z% efficiency). Take the item at any level, and the analysis would give you a perfect valuation of its stats and efficiency, which would be consistent with the system overall. The per-level component doesn't matter in this case, since you are accounting for its context by describing the conditional.
 * 3) I don't. It makes no sense to give the Hex Core a static value, because the bonuses it gives are inherently non-static.
 * 4) Effects with conditionals are, by nature, composite. If you could plot an effect's progression in the same way as you could a stat's, then sure (so the 20 magic damage on-hit could work), but otherwise you would have to refer to the involved stats' baseline components or keep them as uncountable.
 * 5) By the immediate above response, it would make no sense to assign that effect a gold value in the first place, since it having multiple variable conditionals means that you could never plot its progression in the same way as a stat's (what if you increased the shield's duration, or the health requirement? How would that factor in?). Therefore, it would not make sense either to assign comparative gold efficiency to similar effects with tweaked values.
 * 6) Yes, as I think there is a protocol and logical reasoning to these things (see above). Whatever is up to debate is not a matter of opinion, but of reasoning, and one that should be ultimately resolvable using proper application of protocol (which would have to be agreed upon first, but that's why you made this blog).
 * 7) s are a red herring. What you should be looking for are the item's effects, which is 3 minutes of sight in a 1100-unit radius. This is a composite effect with multiple variable conditions (sight duration and radius), which therefore cannot be plotted like a stat and thus does not factor in as a baseline value for anything.
 * 8) Yes, I think temporary stat bonuses can be valued. While the conditionals are uncountable, the stats are themselves countable, so it is possible to describe the value and efficiency of these stats while mentioning the context established by these uncountable effects, as should always be the case. This might lead to surprising results (health pots are massively gold efficient when active), I think the only obstacle here is disbelief rather than any flaw in reasoning. If there is anything that can be done to alleviate the often ridiculously high conditional value or efficiency of those items on a conceptual level, it's that those actives or consumables are meant to be really good while they're in use, hence the high (but non-permanent) value/efficiency.
 * 9) There is no relation whatsoever between runes and items. IP isn't something you unlock over the course of a game, it's unlocked outside of games, and the permanence of IP and runes means that IP efficiency has no real meaning. You could technically do a stat-to-IP ratio (which would be completely separate to gold value or efficiency) and describe runes in terms of IP efficiency, but it would have nothing in common with gold efficiency. You could technically correlate any stat to any other stat (base attack speed to base movement speed on champions, for example), and that wouldn't have anything to do with gold efficiency.
 * 10) See above, I don't think such a system has any real meaning or value. You could, however, make a thought experiment out of it, and establish rules that would allow you to quantify the IP efficiency of literally every rune. I think any such rules would be arbitrary and fairly devoid of meaning, so I wouldn't do it and sound convincing at the same time, but if anyone were to somehow provide a good basis for rules of IP value and IP efficiency, I'd follow those rules and do the math on those runes.
 * 11) For the same reason we don't calculate the RP efficiency of skins, or the IP efficiency of champions. It makes no sense to quantify the cost efficiency of runes, because despite the possibility of establishing stat-to-IP ratios and even a pretty complete system around runes and their IP cost, that system would have no meaning.
 * 12) Cost efficiency tends to be an overall indicator of an item's power budget, which has factored several times into the valuation of items and . Both items were way too stat efficient at one point and were nerfed accordingly, and that efficiency could easily be seen with a cost analysis. It also means an item's power can also be described in terms of its efficiency (e.g. "This item's too cost efficient, and so needs a nerf" or "This item's really inefficient and needs a buff"), and so can the changes made to it.
 * 13) The gold value of AD would increase, and consequently the efficiency of every other AD item would increase. Effectiveness isn't, and has never been a factor in gold value, and would remain unchanged. This may sound weird, but if you think about it: suppose that happened, and consequently had its stat efficiency increased. By both logical reasoning and the cost analysis system (which should be the same), if you were to try to buy the same stats on that item, you'd have to pay a lot more than before, and so your item's more efficient. Avoiding Long Sword and instead using components like B. F. Sword and  doesn't matter, since you'd apply that same reasoning to them, therefore you'd still end up with greater efficiency.
 * 14) It's always possible to change an item's cost efficiency without changing its effectiveness in-game, since efficiency and effectiveness are unrelated. However, you'd likely change the impact of that item, since its build would be affected: you could build it sooner, or would have to wait longer to get it, or could purchase items with similar stats sooner or later in the same manner. This is why Riot often adjusts the price and stats of basic components like Long Sword or or the like, so as to make them easier or harder to purchase.
 * 15) Absolutely. Whatever Riot says is canon, and whichever gold values they list would replace the ones we have now. This doesn't necessarily mean Riot's own valuation is necessarily bulletproof, but whatever baseline they'd state is official would work, as the system would already be able to account for that, bar the event of some gross contradiction (AD not being worth the same on two different items, for example, which would break the system if held as true). If they happen to list the value of a stat or effect that has been traditionally held as uncountable, then it would be made countable, under the proviso that Riot explains how non-plottable effects scale in value/efficiency, if there are any.

I've feel like I've posted a big enough wall of text here, so I'll follow up on the next few sections separately.