Except we don't live in a perfect world, and for most every color combination except for maybe Mardu, Abzan, and Jund, we hit a point after a few removal spells where the options for flexible, unconditional removal that's 3 CMC or less starts to dry up. We're faced with the dilemma - do we pay extra for Murder and Vraska's Contempt? Or settle for Doom Blade and Terror? How often are those going to whiff, anyways?
I looked at the 15 most common "qualifications" on removal spells, mostly in terms of quantity but also in terms of the recency and power of the few cards with the text. For each, I looked at EDHRecs data for the most played Commanders and Creatures, and then came up with a few "scores" for each card - a "Hit Rate" of how often it should be a live card, as well as a "value rate" for how often it should be capable of removing the most threatening card on the other side of the table - the Commander.
I'll post a bunch more about that methodology in spoilers below, alongside a link to my raw data. But, for those who aren't interested and want to get to the Fireworks factory, here's a breakdown.
Acceptable Relevance (Hit Rate of 66%+): Your mileage may vary, but to me I'd want my removal to have a hit rate of at least this. In a 4-player game this level of rate should mean it is unlikely for my cards to be 100% dead in the match and rare to be dead in hand for more than a few turns.
- First up is "Toughness 5 or less", which really is damage and -X/-X effects, with a rate of 77%. That's a bit of an overestimate as auras and equipment can shove targets out of range, but it does mean that Dismember and Magmatic Sinkhole are relevant includes to their respective colors, possibly more so than other removal spells.
- Next, surprisingly, is "CMC 4 or greater" (as seen on Despark and Disdainful Stroke), with a hit rate of 74%. Considering these hit multiple permanent types and I only looked at creatures, as well as the fact that these get around "Indestructible" and "Regenerate", these seem like borderline auto-includes in decks of their colors.
- After that, in what is probably my most shocking finding, is "Green or White", as seen on Noxious Grasp and Deathmark. That's right - on average, in a blind metagame, these are more likely to find targets than Doom Blade. I think especially given that Grasp hits planeswalkers as well, it merits a lot more testing - these cards have a rep as sideboard only cards in Standard and Modern, but the "meta" of EDH today is such that they should rarely be dead draws.
- Lastly, we have "Power 3 or greater", as seen on Skywhaler's Shot and Abzan Charm (also Dusk // Dawn and Meekstone). At a 69% hit rate these are nice inclusions, although it should be noted that in my data they have a high hit rate but lower overall value - they are active and live on the most dangerous creatures to come out of a 99, but they miss on a LOT of commanders. As a result, they aren't quite auto-includes especially in Abzan where you have tons of options. It does mean that Shot is one of the better instant-speed answers in mono-white, Azorius, Selesnya, or Bant after the usual suspects though.
Shaky Relevance (Hit Rate of 50%+): To me this is starting to get in the danger zone - there will be games where these are dead especially in the endgame after a few opponents are down, and they may sit, useless, in hand for several rounds before a target presents itself. They can still be worth running in some color combinations, but may deserve a second look in other decks or if your meta is one where speed is less important and you can get away with Murder.
- The highest rated restriction here, coming sooooo close to being in the next tier, is "Black or Red" as seen on Celestial Purge, Surge of Righteousness, Inquisitor's Snare, or Devout Decree. They had an impressive hit rate of 64%. Again, these effects are seen as sideboard-only but I think now that I've done the math they deserve additional glances for EDH, especially as they are high "Value" (hitting many commanders) and because two of the four of these are capable of hitting multiple permanent types.
- Next at sort of the bare minimum I'd feel comfortable with is "Nonblack". This is the removal restriction du jour nowadays, seen most prominently on Doom Blade but also on many, many, many other removal spells, and had a hit rate of 62%. Honestly? I think you can do better, but sometimes in Monoblack or Dimir running these might be necessary. For my money, I'm pretty sure I'm just never running blade or Hideous End again - if I run a removal spell with this restriction it will be because it generates extra value like Malicious Affliction or because it can be cast for free like Snuff Out or Slaughter Pact, and even then I may run those a little less.
- Unsurprisingly next is "Nonblack, Nonartifact", which is 5% worse than the prior restriction with a hit rate of 57%. Only 1000 decks run Terror on EDHRecs, but I hope that number vanishes soon - there's simply no reason for the risk. Honestly, this makes me take a long, cold, and hard look at running Nekrataal and Bone Shredder outside of the most durdly of Meren and Alesha decks, too.
- Next is "Toughness 4 or less" which is also at 57% hit rate but a much higher value rating, as you hit significantly more commanders. This is the minimum level needed for a -X-X or X damage effect to make it in Commander, I think - in black you should be able to do better than Grasp of Darkness or Sudden Death but in many Red, Blue, Simic, Gruul, Izzet, or Temur decks it makes running Char, Lava Coil, Psionic Blast, and similar cards at least worthy of argument.
- Rounding out this category, and just barely, is "Power 4 or Greater" as seen most relevantly on Reprisal, Eternal Isolation, The Wanderer, and Collective Effort as well as on the card that started my little quest - Big Game Hunter. With a hit rate of 51% these should rarely be dead, but I'd have to stress that that hit rate is being buoyed a lot by how "effective" these cards are against threatening finishers out of the '99 - these effects have a profoundly lower value level than all other removal in this tier and are dead against most all of the most popular modern Commanders. Run them at your peril, or if (in the case of Hunter or Defiance) they bring some additional value to your deck via synergy.
Bad Idea: Below 50% hit rate: This is mostly the stuff you'd expect; I did the math on "Blue" and "Red" (as seen on stuff like Pyroblast or Hydroblast ) and found that those just aren't going to hit often enough even with the modals. For what it's worth, "Blue" won the category with a hit rate of 35% so it isn't totally insane at least. "Toughness 3 or less" had a sad hit rate of 31% - I had mentioned offhand a few weeks ago that Lightning Bolt seemed underplayed, but it's actually played just right as it is going to whiff a lot in this format. The only shocker here is "CMC 3 or less", with our worst hit rate of 23%. Of course almost no one is playing Smother or Baffling End in EDH, but Abrupt Decay is in over 10000 decks and Tyrant's Scorn is also in a surprising amount of decks. I won't say that's wrong as those spells are modal and should never be truly dead draws, but one should probably look closely at how often decay is just a Natural State and Scorn is just an Unsummon and consider adjusting accordingly.
EDIT: As a bonus to call to mind just how questionable these are, I ran "nonlegendary" real quick (as seen on Cast Down). Cast Down has a hit rate of 41% and a value of 65, outstripping all of these, despite being completely unable to handle any commander. I don't advise running Cast Down, but hopefully that also is extra illustration on just how "dead" these effects are often going to be.
- I used EDHREC data as of the last few days.
- I did not weight any particular creature for "likelihood of targeting it" anywhere except the "Relevant" Column. For instance, I did not add any consideration that removal on Yuriko, The Tiger's Shadow is not very effective, or that Edgar Markov is unlikely to ever leave the command zone.
- I also did not weight by presence - it made no difference that Sun Titan is the most omnipresent creature in the 50 relevant creature list and Etali is the least.
- Korvold, Fae-Cursed King was considered a 4/4 with a +1/+1 counter for the purposes of targeting, as the window in which he has no counter is very small and easily interrupted by a sacrifice outlet. This consideration was not extended to Ramos, Dragon Engine, Edgar Markov, or Deadeye Navigator, Mother of Runes, and similar creatures, because the windows to respond to those cards is wider. It does mean a bit of caution is required in evaluating Heartless Act and Dismember, however.
- Jarad, Golgari Lich Lord and Prime Speaker Zegana have variable power; I considered them as likely targetable by an effect that destroys cards of "3 or more" power (since it's rare to have someone cast them without them being at least a 3/3), but as unlikely targets for any "5 toughness or less" effects (because they can't, reliably, be counted on to be destroyable by them).
- I did take a creature's text into consideration somewhat - all of these are "Destroy" effects for the most part, so I did consider Xenagos, God of Revels untouchable by them. A few exceptions - because not all effects are black, I did not factor in Karmic Guide's protection. Because all effects that target cards of CMC 4+ either exile or counter at this time, I considered them successful against the Gods. I did not take into consideration the ability to sacrifice Sakura-Tribe Elder or Burnished Hart in response - removal is considered capable of "hitting" them even though it will often fizzle.
- "Toughness X or less" in general was tough. I generally evaluated it all as damage, even though -X/-X also kills Indestructible critters and regenerators. Meanwhile, it is totally possible that these cards quickly fail due to equipment, auras, or counters - and also possible to extend their reach by combining them or tossing them out after damage. I simplified it to purely look at printed Toughness (minus the aforementioned exceptions) with the understanding that those results should be taken with a grain of salt.
- In part because of Elder and Hart, the "Relevant Target" category was created as a curated list by me of the 50 most common creatures in EDH one might actually want to kill. The complete list is below; in it I effectively excluded creatures only played for their EtB or Death triggers and creatures that almost all removal fails against (the gods, Underrealm Lich) to get a wider range of targets. I did this because a full half of the top 10 most played creatures are things I'd never ever burn removal on - Solemn Simulacrum and Eternal Witness and Sakura-Tribe Elder. Another decent chunk are cards that, in competitive play one may want to kill (e.g. always bolt the Birds of Paradise) but that I'm unlikely to want to specifically target in a multiplayer game for a variety of reasons both strategic and political. I wanted to see the efficacy of removal against specifically cards one might actually want to surgically remove instead of just waiting for them to be mopped up as collateral damage in the next boardwipe.
- I do recognize some of that is debatable - most people play Prime Speaker Zegana for her EtB trigger as she's just a vanilla beater otherwise, and sometimes killing Acidic Slime or Baleful Strix really is relevant in determining the outcome of a game. My final consideration in these cases is "how likely is this creature going to be capable of swinging for lethal unassisted?" Really anything over a 5/5 is still plenty threatening. On the flip side, I had to make a call on cards like Goblin Electromancer, Deathrite Shaman, Beast Whisperer, and Kiora's Follower - these look a lot like "little birds" as small enablers, but I decided they enable enough nonsense to be priority targets in a way Elvish Mystic rarely is.
- The final scores also deserve some explanation. Hit Percent is, simply, the total number of hits divided by 140 (the total number of cards looked at). Because there's a ton of overlap in the four data columns, there is also "Strict Hit", which looks only at "Commanders of the last 2 years" and "Top 50 Relevant Targets", divided by 70. These are the percentages actually reported above.
- Finally we have "Value"; I wanted to have a column that properly considered that killing Commanders is typically more important than just creatures. In an extreme example, if you can kill Meren or Grave Titan, you MUST kill Meren - as a sufficiently experienced Meren will simply reanimate the Grave Titan. Not all Commanders are this extreme, but most are - killing commanders can shut down turns or entire decks. Value thus gives a small numerical boost to the two commander columns to more accurately reflect how life-saving that removal spell will be in practice.