Quoting my Beggar's Bazooka guide - "MvM is a game about damage. In fact, compared to PvP, which is objective-oriented, MvM stresses far more on damage output: if you deal enough damage to kill all the robots, you win."

After the (more or less) success of my last guide on beggar’s bazooka, I’ve decided to make a more detailed analysis of the most important aspect in MvM, damage. Or to be precise, the maximum potential dps that common damage weapons can deal to a single target consistently without inflicting self damage.
They represent the limits to the damage capability of weapons, and are not necessarily comparable to actual damage numbers in-game (although they remain mostly true).

This guide is intended towards experienced MvM players with decent amounts of knowledge about weapons and upgrades.

Warning :(again) expect even more numbers. If you have a natural hatred towards math, logic, or any form of reasoning, please close this guide and remind yourself never to open it again.

The calculation of single target dps is rather straightforward. It’s (amount of damage)/ (time required to deal damage).
In the case of the tests that we’ll be conducting throughout this guide, it will be (damage per shot)*(firing rate).

Inheriting the method from my previous guide, I’ll compare similar weapons/ same weapon with different upgrades to hand out a more direct idea of the difference in firing rates.

Note that the timers use frames instead of centiseconds, so 30:30 would be 30.5 seconds.

Combining with the table from my last guide, here're the results of the tests, ammo consumed, and the calculated firing rate.

A lot of these numbers will need further explanation so I’ll do a brief analysis on individual classes.

• For clip based weapons, I started off with no ammo in clip, in order to represent average dps.
  
• Tap fire weapons (beggar's, stickies, huntsman with 0 draw) tests are done with a rapid tap script.
  
• The panic attack doesn't benefit from tap firing as the beggar's does.
  
• The tests were done by spawning a puppet bot on a local server, then giving it different cheats depending on the tests.

• As pointed out by user 'itsgalf', not firing off an entire clip gives a lot of scout primaries a disadvantage in terms of dps. Hence I used the time required to fire off 36 shots (instead of 50 in the video), since 36 is a multiple of 18 and 6. The difference is not too significant, and the ultimate conclusions are the same.

• The Air Strike shows a 37%-48% increase in firing rate, but at a cost of higher vulnerability (and health from rocket jumping).
  
• The base jumper barely increases average firing rate. As does clip size upgrades. However the latter does benefit burst damage capability, but we won't cover that in this guide. Reload & firing speed are still the essentials for this weapon.

• The firing speed varies slightly from the stats given by the wiki. This could be caused by errors from my timer or the source engine itself, but the difference is negligible. However, since its damage isn’t based on ammo consumed, the result from the firing speed test is useless.
  
• After watching the video for pyro damage in slow-motion, the flamethrower seems to deal damage 22 times per second (disregarding afterburn). This also matches closely to the stats on the wiki, so I will use this number later on.

• The 2nd firing speed upgrade does nothing, and the amount of firing speed boost is nowhere close to what the upgrades claim to be.

Ok, this class is a mess
• Without the wrangler, the 1st upgrade in firing boosts it to ~120% of original, while the 2nd brings it to 150%. The 3rd does nothing.
  
• With the wrangler, the 1st firing speed upgrade and the wrangler boosts it to 300%. The 2nd/ 3rd firing speed does nothing. Nada. Nichts.
  
• Due to errors in the timers, some of the results were close but not identical. For the sake of simplification, I took the mean average of those numbers.
  
• Crit canteen seems to have the same effect as maxing out firing speed. The test for an upgraded sentry with crit canteens wasn't done, but the numbers should be easy to come up with based on other tests.

• The time in the video is incorrect, so I counted the time manually.
  
• The shield's firing rate seems to match the server's tick rate (66/s)

This is the second element in dps calculation. As previously mentioned, for normal damage, due to damage ramp-up, we’ll test the maximum possible damage without inflicting self damage. In addition, damage spread and random crits are turned off.
We’ll also test crit damage, but not mini-crits, since mini-crits damage is simply normal damage multiplied by 1.35, even with damage ramp-up, while full crits neglects damage ramp-up altogether.

Further notes:

• Crit damage is the same on bots and tanks, with the exception of heavy, who has a 75% damage penalty.
  
• You may have noticed that for most weapons, tank damage is slightly lower than bot damage. This is due to weapons reaching a lower ramp-up on tank (bigger model size).
  
• The Source engine adds damage after rounding. For instance, the flamethrower actually deals 13.64 damage per hit, which rounds up to 14, and 10 hits with the flamethrower deals an accumulated 140 damage instead of 136.

• It's impossible to stay in close range with the Air Strike while rocket jumping. Hence I'm using its base damage.
  
• The bison's damage for each shot is affected by the number of times the projectile hits the target. The tank has a bigger hitbox than bots, hence more damage. Still, the bison's tank damage is fairly inconsistent, but 120 regular & 370 crit damage are the numbers that showed up the most.

• Got lazy and used the wrong clip for bot damage in the video. It's 226 at close range.
  
• Detonating stickies that are planted for longer periods of time seems to neglect damage falloff.
  
• Stickies beneath tanks only do ~70% full damage, while stickies on the sides do full damage.

• The sentry gun, despite stated otherwise on the wiki, is still affected by damage ramp-up even within its scan range. A level-3 reaches a maximum of 19 at point blank and 16 at the edge of its scan range.
  
• Sentry rocket damage is not included.

• With minicrits, the projectile shield does 3 damage per hit on bots and 2 on tanks.

• Since charge time and the rifle's own damage per shot doesn't affect explosive headshot damage, and the majority of sniper's damage come from EH, it should be clear that rapid quickscopes do far more damage over time on groups compared to fully charged shots.
  
• Explosive headshots do flat 190 damage and isn't affected by splash modifier, charge amount, damage per shot, crits, or Fists of Steel's resistance. They can still be minicrit boosted and deal 256.
  
• Huntsman's damage with 0 draw seems to be inconsistent. I wasn't using a rapid tap script there so I'll use the lowest number that showed up.

• Backstab damage isn't affected by the battalion's backup.

Now that we have the two factors needed for the calculation, firing rate and damage per shot, we can proceed into the number that matters: the multiplication of the two.

Keep in mind that the following numbers maybe slightly differ from actual stats in-game due to rounding and testing errors, and only represent the max potential of a weapon, while in-game results can be influenced by a lot of other factors (bullet spread, distance, projectile deviation, running for ammo, “indirect” hit (stickies), not hitting anything, etc.).

Also, we’ll only use stats from maxed out weapons.

• These are based on the assumption that there are no robots nearby. IF there are, the phlog's and demoknight's tank damage skyrockets, and without crit canteens or a medic, can bury every other weapon under 9 feet of dirt and tank rubble, except for one texan.
  
• Engi vs tanks will be discussed later.

As I've previously mention twice, the chart (and numbers) only represent the max potential of a weapon. While for weapons of the same class (scattergun & sodapopper), or weapons for similar classes (Beggar's vs pipes), the chart can roughly quantify the difference between them, different weapon/ class can lead to varying playstyles, and hence very different usage of the weapon's potential.

Heavy is a tank class, and deals the most damage when the enemy is right in his face. He doesn't mind drawing aggro at all, and in fact, should do so. He can make a lot of use out of resistance upgrades, which can also boost his damage besides his minigun upgrade, often times far more than the latter.

Demo is not a tank class. It makes little to no difference if the enemy is in his face or not, so having resistance doesn't benefit him in the slightest apart from survival and he should never try to draw aggro.

Heavy is a class that relies on close range damage. Anything farther than say, mid-range, can heavily (heh) gimp his damage. Also, the minigun deals continuous damage over time, so streams of robots that flow steadily can make his damage much higher.

Demo, on the other hand, doesn't need to be at the crime scene to kill. He is also great at burst damage, which means a demo would much prefer a tight pack of robots coming in short bursts instead of a never ending stream.

The dps chart (and numbers) gives out a good representation of average damage, which illustrates weapons' abilities against continuous streams much better than packs.
This should explain why heavy (and beggar's soldier), with the higher dps, can do so much more on mannslaughter, where both the map and the waves forces the robots into streams, while demo can carry so hard on bigrock and mannhattan, where the team just spawncamps for half an hour.

The two subclasses inherit very similar playstyles, but the beggar's soldier also have incredible support utilities. And this is where the dps chart comes in and say soldier is the wiser choice.

http://steamproxy.net/sharedfiles/filedetails/?id=462408278

As the game progresses, the engi's role as a damage class becomes far less significant than support. Additionally, his essential upgrades for his buildings and wrench nearly top off at roughly 3k credits. Despite that, engi is still one of the highest tank-damaging classes even in final waves, and the player can be very creative when it comes to dealing with them, instead of crouching behind your sentry and asking your oblivious gibusvision pyro who maxed out his fireaxe. Since tanks do not fight back and are relatively slow, they become much easier targets for engies with other weapons in their arsenal.

The costs in the following section represents the extra cost after getting the essential upgrades.

DPS: ~350
Cost: 0
This seems pretty obvious. A wrangled sentry outputs a very impressive 396 dps and can keep up with other high dps weapons. However, when wrangled, a maxed out sentry will chew through its ammo in 9 seconds, while it takes up to 3 seconds to refill. A tip is to quickly pick up then put down the sentry after its ammo is depleted to let it keep firing after switching off the wrangler. The sentry's average dps with this method is ~350.
Downside: Very metal consuming. Also, since the sentry is wrangled, you won't be able to guard the bomb at all.

DPS: ~316-492
Cost: 0
Since your most likely be upgrading your wrench's swing speed, why not use it for damage as well? Given melee weapons' high random crit chance in PvP (up to 60%), the stock wrench + sentry combo can do up to 492 dps on tanks. Then again, I haven't paid any attention to random crits in MvM, and I'm not sure if it uses the same mechanics. The combo can do at least 316 dps with the jag and no random crits.
Downside: None

DPS: ~430-549, 1116(crit canteen)
Cost: 1150~2700
The panic attack is an often overlooked shotgun in MvM, and TF2 in general. After all, who wants to stay at low health? Unless... you're fighting a tank.
At 1 hp, the panic attack can drop a ridiculous 350 tank dps by just holding down mouse1. And since the tank never fights back, this strategy is highly viable. Minicrits boost the combo's dps to 741, higher than every other weapon w/o crits, and you can even pop a crit canteen (oh the meta), and reach an absolute unlawful 1116 tank dps with the combo. Keep in mind that a crit beggar's do 1118 dps, and due to possible calculation errors, who know if this combo has the actual highest tank damage in all of MvM?
Quite conveniently, you can use the wrangler's to damage yourself to very low hp.
Even without clip size upgrade, its dps is still a staggering 329, and at this point it costs a mere 1150 credits. With crit canteens, you can reach 1070 tank dps with theoretically just 700+1150=1850 credits! No other class in MvM can even come near this number.
At full hp and with no clip size upgrade, its dps is still a very impressive 231. Also, despite its high firing rate, the player can still have full ammo with an medium ammo pack if he has max ammo cap.
As ridiculous as using a crit canteen as an engi sounds, I hope you see where I'm getting with this. The panic attack deals a phenomenon amount of point blank damage, can fully utilise crits, and costs literally negligible amount of credits to upgrade.
The panic attack can even be a choice for fat scouts, should you even wish to do a 5-heavy rush, but that's for a totally different topic.
Downside: Giving up the rescue ranger can cripple your versatility as an engi.

DPS: ~479-500
Cost: 1150~2700
The frontier justice doesn't require the player at low hp to be fully effective, but it is a lot more cumbersome to use due to the crit mechanism. Still, it can theoretically store up to 70 (35*2) crits without rebuilding, and you can stay near your dispenser for ammo, since you can have full hp, and crits have no drop-off. It deals 300 dps with full size, and 280 w/o.
Downside: Same as above; crit canteens are mostly useless.

DPS: ~439
Cost: 400
Despite being very economically friendly, the widowmaker cannot get metal back from tanks, and since even dispensers and full ammo packs wont replenish your metal fast enough, it is mostly unviable for this case. It's similar to the bison in the sense that its upgrades are cheap, but it doesn't achieve much with them. Still you can use it for a quick burst of damage.
Downside: Very metal consuming; giving up the rescue ranger.