There are other factors to consider. How much rare candy do you have to invest in bulky legendaries? What secondary types, if any, do the specific attackers have? Are you using them for tier 3 solos, small-group tier 4/5 raids, gym offence, PvP or a combination of the above? For raids, how hard does the boss hit and what moves does it have?
At what rate would you trade DPS for TDO?
Seems like a lot of non-legends that usurp current top of the line attackers (legendary or not) do so by beating them marginally in DPS, but losing out slightly more in TDO.
Electivire gains a 3% DPS advantage but loses 25% TDO compared to Raikou
Breloom gains 3% DPS over Machamp, loses 30% TDO.
Rampardos gains 20%+ DPS over Ttar, loses 39% TDO.
Weavile gains 6% over Ttar, loses 33% TDO.
FP Sceptile gains 6% DPS over FP Venusaur, loses 14% TDO.
Obvious exceptions exist, Mamoswine for example should beat out both Articuno in TDO and Weavile in DPS.
That poses the question- at what point do you think the sacrifice in TDO compared to the incumbent is enough? Personally I feel that anything that takes more than a 20% hit is tough to swallow. Some notable exceptions for me include:
1. >20% TDO loss is for a type with an immunity
So if you take a Weavile to a psychic mewtwo battle, it's not exactly going to die that quickly.
2. If >20% loss, but Greater than 1% DPS per 2% TDO drop. Rampardos is clearly the one that I tried to fit this around, but I feel like if you gain a 20% DPS for a 40% TDO, not exactly a bad deal. Breloom and electivire sound like rip offs to me, but the Sceptile trade sounds very solid.
(The sources for all my numbers are in this reddit post):
Going to talk about Raiding only in this post.
TDO matters to me when I'm in a short-man or Solo situation where time is of the essence, and I can't waste 10-15 seconds bringing in a new Party. (example: duoing Togetic) So I will take into account things like Resistances and Super Effective Damage and Movesets, to maximize how much damage I can do and minimize the amount of time I have to spend bringing in a new party.
TDS matters to me when I'm the minority Team in a larger raid AND I know that the battle is going to be over quickly, so I need to do as much damage as I can in the shortest amount of time. (example: 10 high level players vs Palkia) I know that my Level 38 Stamina 15 Rayquaza is going to be One-Hit by Palkia's Draco Meteor, but if I can dodge 1 DM, I can get off 3 Outrages and a dozen Dragon Tails off and do more damage than bringing in my Gardavoir and soaking a DM while hitting for 4-5 Dazzling Gleams... and neutral damage for my Charge Move. So I'm going to lose my Rayquaza and 3 Dragonites (my phone SUCKS for Dodging), but I'm doing more damage and getting more raid rewards.
It also matters to me when I'm going in on a raid with a bunch of low-level players who have poor counters - which happened a LOT last summer against the Regi's, but more than a few times against Groudon last month. I've got 350+ Max Revives, I seem to get about 1 per every 2 Gifts opened, so I can use a bunch in a Raid. So when I'm the Level 40 and everyone else is Level 30 or lower, I go with the DPS crew, no dodging, do as much damage as possible, heal up and go back in with that group with the goal of doing a stupidly large amount of damage to make up for the fact that the others in the raid are going in on Regice, for example, with a Level 20 Machamp with Bullet Punch and Close Combat because that's the only Machamp they have, and they don't have any TM's.
"if I can dodge 1 DM, I can get off 3 Outrages and a dozen Dragon Tails off"
I'm genuinely curious about this. I always bring Rayquaza in my Palkia team, but in my experience, dodging a Draco Meteor just results in the dodge bug kicking in since the move one-shots Ray. How have you been able to get around it?
I almost never suffer from that particular Dodge Bug.
The problems I have with dodging are device based (ie: my phone just doesn't react properly to the screen swipe left or right) and not the known bug. So if my phone is reading the swipes properly and accurately, and I do it at the proper time, I almost always dodge normally. Otherwise I take the full DM and get one-shot as per a normal attack.
When I do encounter the bug, I'm not sure it's the same one either; it rarely happens on my First Attacker (in this case Rayquaza), but starts on the 2nd or 3rd attacker instead. And as I'm not that great at dodging (phone issues make it hard to practise), I'm not sure what I'm running into is the same as what others are running into, and might be more device-based than server based.
Your simplified assessment of +1DPS/-2TDO fits with my feelings.
I think Rampardos might be a worthwhile investment of stardust if I ever catch enough to evolve one, but I don't really want to spend rare candy on it.
Similarly, I can't see dumping that much rare candy into a second Rayquaza when I already have three max Dragonite, a max Latios, and another Dragonite around level 35. If Salamence gets Dragon Claw or Outrage for its eventual CD move, then I will want to max one of those too. Also, aside from the 96% Rayquaza I powered up, the rest of mine are under 90%, and Rayquaza has a high potential to return to raids with a shiny and/or exclusive move.
I sincerely hope Salamence gets Fly instead of one of the existing dragon moves. It fits better given its "mythos" for lack of a better word and differentiates it from the rest of the premier dragon types.
It's also worth mentioning that despite its name, Rayquaza's main series signature attack Dragon Ascent is actually a flying type attack.
I'm trying to max one of each species with different moves. So it's mostly composed of both. If other options are considerably inferior I go add weather boosted lv30-35 repeats.
For example, I will definitely keep Rampardos at the first slot or maybe keep 2 of it and fill the rest with Tyranitar or Rhyperior depending on matchups.
For grass I simply love Venusaur and will probably just go with it again with Roserade in front line.
Again, electivire, Raikou, Zapdos , magnezone kind of line up
Not really have a clear cut formula for it, whatever I have in hand, whatever I feel like investing and using within the winning boundaries of time limit for raid.
Well one thing to mention is Grass knot Roserade, by far best grass attacker
But theres no catch all solution, its somewhat match up dependent. So for match ups like mewtwo and Cresselia. theres times you will want Tyranitar or weavile for mewtwo over gengar because gengar dies so often while the other two last much much longer. But against Cresselia, because its attack is so low, gengar faints out less, making the extra tankiness of the dark types less impactive.
Another one is Suicune and kyogre. against the harder hitting opponent, you will want some more TDO, but against tanky, low damage opponents the DPS is better
Disclaimer: I'm not trying to do tough solos.
I would consider myself a casual player at this point, so I agree with the first answer. I'm not drowning in stardust and rare candies are very hard to acquire. Because of that I will almost always go with the non-legendary option regardless of any stat. I did have a few exceptions, I maxed my team bird, Zapdos, and powered up Kyrogre and Mewtwo. I have level 40 Zapdos, Jolteon, Flareon, and Arcanine. I wouldn't invest in any of the legendaries of those types.
When it comes to 2 non-legendaries, I will invest in others just for a difference factor. For example I maxed out a shiny Hariyama and powered up a Breloom just for the sake of not running 6 Machamps.
I think my logic is pretty sound.
I follow these rules:
1) if practical DPS is higher, I'll always take it as a first slot. I specify practical because I need to actually be able to use it (i.e. deoxys attack is worthless because it won't get a charge move off). However yes, I'll give up 50% TDO for 2% DPS on my leading pick - I did in fact use breloom on the start of my Tyranitar team before the HP changes.
2) I'll take more DPS over TDO so long as the pick can reliably survive a charge move.
3) if it can't survive a charge move (e.g. against DM palkia) I'll still take more DPS if I can reliably get off my own (outage requires less energy than DM so I can use it) AND the DPS difference is significant (dragonite significantly out damages gardevoir).
4) otherwise I'll take the best DPS option that has sufficient TDO to survive a charge move.
Note that I sort of skipped a rule that Comes before 2, let's say 1b: if I can finish a raid without fainting out, I'll take fully best DPS with no considerations for TDO. So in 20 man raids I'll always use my best DPS regardless of their TDO since they won't all faint before we win.
Overall, these rules basically mean I'm almost always picking the highest DPS counter, unless it's a glass counter: in which case I take one of that and the rest are my highest DPS non-glass counters.
I'd also like to specify, however, that I'm Stardust poor - so I'll continue using many things like Tyranitar that I've invested in already rather than entirely replacing them. I'll definitely power up 1 rampardos ASAP / once there's a raid for it, but I doubt I'll ever fully replace my Tyranitars.