For Blizzard's official numbers regarding MP levels, check out their blog post here. Contents Part 1: Mathematical Basis 1-1. Farming Quantified 1-2. No Such Thing as Half a MP Level Part 2: The Goods 2-1. MF/GF Chart 2-2. MF/GF Conclusions 2-3. Keyfarming Chart 2-4. Keyfarming Conclusions 1-1. Farming Quantified Before I get into the results, let's talk about how farming should be properly measured. Farming consists of two essential types of events. One of these is DPSing, the act of actually attacking monsters to make them drop loot. The other, far broader category, we'll label downtime, a blanket term to cover anything that isn't DPSing. Running from a dead monster pack to a living one, kiting (maneuvering around living monsters for survivability purposes), deaths, recharging (waiting for cooldowns to reset or resources to regenerate), selling items to merchants, stash dumps, and even putting found items up for auction are all considered downtime. However, only serious time counts; inefficiency caused by not really playing the game with your full attention does not count towards this definition of downtime. Unfortunately, the line between downtime and DPSing isn't quite that clear-cut. Killing white mobs doesn't have the same magic find potential as killing elites, and the time spent DPSing before achieving 5-stack valor is not as farming-effective as time spent on 5-stack. For the moment, we will assume that such activities count as partial downtime; finding the exact coefficient would be a subject for later inquiry. Let F equal your farming efficiency, m equal your magic find, gold find, or the key-find modifier (depending on what you're looking for), x equal the health of the monsters at MP0, h equal the monster power health modifier, d equal DPS (damage per second), and w equal downtime in seconds. F = m/[(hx/d)+w] The interesting thing about the Monster Power system is that many of these variables remain at least somewhat constant. The premise of all of the following math is: downtime remains constant moving from one monster level to another. In regards to time spent moving from one mob to another and time in town, this is a pretty safe assumption; in terms of kiting, deaths, and recharging, it's a much less safe assumption. However, most skilled farmers take care to avoid exactly these forms of preventable downtime, by over-gearing for content and maximizing resource regeneration/sustainability. 1-2. No Such Thing as Half a MP Level Due to the integral nature of MP levels, finding breakpoints is the practical application; in order to find the breakpoint of an inequality, we need merely find the point where they are equal. A quick glance at the MP chart shows the monster health outpaces increased magic find, even at MP1. Therefore, only downtime can make a higher MP level more attractive. However, as DPS increases, DPSing time naturally goes down, increasing the relative amount of downtime in a given farming run. So what we need to find is a measure of downtime. Remembering that downtime is a constant, eliminating other constants such as DPS*, and setting F[SUB]1 [/SUB]to the lower MP level, we get: F[SUB]1[/SUB]=F[SUB]2[/SUB]=m[SUB]1[/SUB]/[h[SUB]1[/SUB]+w]=m[SUB]2[/SUB]/[h[SUB]2[/SUB]+w] Solving for w, we get: w=(h[SUB]1[/SUB]m[SUB]2[/SUB]-h[SUB]2[/SUB]m[SUB]1[/SUB])/(m[SUB]1[/SUB]-m[SUB]2[/SUB]) Calculating the exact amount in seconds is too obsessive for the layperson, so what we want instead is a ratio of DPSing to downtime. In other words, if I ask you right now how many seconds you have of downtime per run, you probably wouldn't have an answer, but if I ask you whether your downtime is greater or less than 50%, you could probably tell me. Framing this downtime as a percentage of overall farming-run time for the lower MP level gives us: W=w/(h[SUB]1[/SUB]+w)=(h[SUB]1[/SUB]m[SUB]2[/SUB]-h[SUB]2[/SUB]m[SUB]1[/SUB])/((h[SUB]1[/SUB]-h[SUB]2[/SUB])m[SUB]1[/SUB]) Now it's simply a matter of applying this formula to MF and keyfarming scenarios. 2-1. MF/GF Chart The chart below has two different types of breakpoints on it: The 0DT breakpoint assumes the character has 0% MF (or GF) from gear AND no Paragon levels whatsoever; however, it does assume 5 stacks of Nephalem Valor; The 300DT breakpoint assumes the character has 300% MF (or GF) from gear and/or paragon levels, as well as 5 stacks of Valor. As a reminder, breakpoint values are the minimum percent of downtime per farming run at the listed MP level before graduating to the next MP level. Code: MP 0DT 300DT 0 71.4 89.5 1 75.0 90.0 2 75.2 89.4 3 75.1 88.7 4 77.2 89.1 5 79.2 89.6 6 80.8 90.0 7 82.2 90.4 8 83.3 90.7 9 84.4 91.1 2.2 MF/GF conclusions The TL;DR is: MP0 for Act 3, and MP1 for Acts 1 and 2, are awesome. Everything else pretty much sucks. However, for you nitpickers out there, this is where the unclear definition of downtime becomes somewhat important. For gold-finding, I think everyone agrees that killing whites is beneficial, and therefore it's almost impossible to graduate up to higher MPs while maintaining efficiency. For loot-finding, auction house actions and the possible downtime aspect of killing whites (or if you don't kill whites, moving from one elite pack to another) enter the picture. Even then, it can very difficult for a serious farmer to graduate to a higher MP level while maintaining efficiency. For example, if you go on a 19-minute run killing 17 elites, finding one AH-worthy item which takes you 1 minute to list afterwards, your elite battles averaged a mere 20 seconds each, and you don't count whites or the elites you downed to get 5 stacks of NV (not even partially!), you still only have 80% downtime, which means you are staying at MP1 unless you have near-zero MF from gear or Paragon levels. Even if you're the most godly DPSer on your server, I can't imagine anyone who's ideal farming MP level is higher than 4; for 99% of the population, it's 0 or 1... as the TL;DR said. 2-3. Keyfinding chart As a reminder, breakpoint values are the minimum percent of downtime per farming run at the listed MP level before graduating to the next MP level. Code: MP %DT 3 17.0 4 37.2 5 50.1 6 58.3 7 64.3 8 68.8 9 72.2 2.4 Keyfinding conclusions You might notice there are no entries for levels 0, 1 and 2. That's because the value is negative; if your downtime is zero or greater, increasing MP level benefits you! MP3 is the minimum MP level that anyone seriously farming keys should consider; for virtually every keyfarmer, MP5 is a much better goal to aim for. Keyfinding also scales much better with increased MP level than MF and GF, although at the high MP levels it still likely becomes inefficient. However, 60% downtime isn't really that hard to imagine, especially since whites do absolutely nothing in the keyfinding process and should rightly be treated as a waste during downtime computations. Therefore, MP8 keyfarming runs aren't necessarily an undertaking reserved solely for the insane. One thing to keep in mind though: Deaths (and extra kiting) both decrease farming efficiency. Don't try to farm something you can't handle. If you have a very powerful character capable of easily handling the higher MP levels, this is an area where you actually enjoy a competitive advantage over other farmers... as opposed to, say, trying to out-farm Legendaries against a Tactical-Advantage spamming demon hunter with low defenses who runs MP1 Act1 over and over again. Very many characters can farm items efficiently; very few can farm keys well. As such, high-powered characters who aren't keyfarming are pretty much doing it wrong. * Very nitpicky math note: Spoiler Upon eliminating DPS as a constant, w is no longer downtime in seconds, but downtime expressed as health. At first it seems weird, as if you're DPSing your stash when you're dumping into it, or DPSing the ground as you move. However, if you think of it from an economics opportunity-cost sort of angle, that is exactly what you're doing, and I wanted to eliminate the DPS constant as soon as possible to avoid rewriting it numerous times.