Heroes In Horntail - Where do they stand? (Project Pippy)

Heroes have a pretty contentious place in the meta on ML. To some, they are a good starter attacker that a player can switch off of later in their career when they have more NX. To others, they're an important tool of daily bossing for content such as CWK, Core Blaze, and Scarga. To many, they are sadly irrelevant, as Heroes have struggled to find a place on squads when there are so many Night Lords available to be taken instead.

Despite the narratives that exist, however, the DPM Compilation thread tells an interesting story, one that we wanted to explore. Surely if a Hero, even under contrived conditions that are detrimental to the overall run, can put up a 15m DPM 30, there's potential for them on a typical 6 man squad, right?

Enter Project Pippy.

Centered around the ever-eccentric iPippy and his experiments in proving Hero viability, Project Pippy set out to definitively prove or disprove the viability of Hero in Horntail, strictly in terms of actually benefiting the squad. What does this mean?

The goal of the run was not to necessarily maximize Hero DPM in HT. While Huiae's 15m DPM is impressive on paper, the conditions required to actually execute it are overall detrimental to the speed of the run. The goal of this experiment was instead to see if replacing a Night Lord with a Hero on a "standard" 6 man HT squad could result in, on average, a faster run.

Without question, the best attacker class in HT at high levels of play is Corsair. However, it's also no question that the availability of high level, geared and skilled Corsairs is incredibly sparse. As such, we will assume a "standard" and "ideal" 6 man HT squad to look like the following:

• Bishop
• Buccaneer
• Archer (Bowmaster or Marksman)
• Night Lord
• Night Lord
• Night Lord -> Hero

with one of the Night Lords in our case being replaced by a Hero.

We further optimize the set up by equipping the Hero with a 1H sword set up. This allows the Hero to function independently from SI, allowing it to cross to the left side and 2-head cleave when windows open, as well as provide better uptime on right side due to Guardian. While this leads to an overall DPM efficiency loss compared to 2h, this set up allows for more flexibility, which to us felt like an important trait in HT, as the ability to adapt to changing Head HP conditions due to randomness with Cancels is important in preventing scuffed runs.

Thanks to rough DPM estimates provided by akashsky for a 1H Hero between 5-10k range compared to a similarly geared Night Lord, we determined the Hero would be dealing approximately 87.6% of the Night Lord's dummy DPM. Given that incredibly skilled Night Lords can expect an "average" DPM efficiency of 80-85%, averaged to 82.5%, the Hero would need 94.2% DPM efficiency to match the Night Lord in overall body DPM. This number gets lower depending on the skill of the Night Lord being replaced, with a 75% DPM efficiency NL only needing 85.6% DPM efficiency from the Hero in order to match their body performance.

Note that this "average" is technically the highest range of DPM efficiency expressed by NLs on runs with favorable conditions. Since the there's a wide variety of conditions that can be found on runs outside player control (notably Cancel RNG) this "average" should actually be more likely in the range of 75-80%. Since the prehead damage will be lower as well, we just took a nice goal estimate of ideal Hero DPM efficiency at 95% to be pulling its weight, and 100% for the Hero to be an upgrade, as Body DPM is higher weighted than preheads in terms of effort and pot drain, although again, given realistic "average" Night Lord DPM efficiency numbers across ALL runs, this bar is actually artificially high and technically even a lower efficiency could prove "viability". However, we wanted to make sure we were being conservative instead of overly optimistic.

So what was the strategy for reaching these numbers? Instead of creating "idealized" scenarios such as Huiae's, we wanted to minimize detrimental effects on the run while still giving the Hero as many opportunities to cleave as possible. In order to create the conditions of Huiae's run, the Bishop would not have to cast Genesis in order to leave as much Arm and Wing health as possible, and Attackers would not be allowed to hit Wings when Heads cancel. The Hero would additionally contribute little to MH damage, and instead focus purely on Wings/Arms uptime. Both of these have detrimental effects on the overall speed of the run. On top of this, there is a significantly higher risk of early Mass Sed if the cancels don't line up properly. None of these are feasible for regular runs, and only work in the context of a meme run created specifically to make the Hero DPM look as high as possible.

Instead, we opted to keep the normal run strategy as intact as possible. Normal Kill order of Legs > Tail > RH > LH/MH > Wings > LA > RA would still exist, and the Hero would find windows to cleave whenever possible. Head damage was prioritized over Wings and Arms, as Head cancels are the biggest time loss in scuffed runs as well as providing the biggest danger to the run. Since wings and arms are important "filler" targets for attackers to hit when heads aren't available in order to preserve uptime, cleave windows using them were precious and only opted for when the cancels and head health enabled them. The goal was to spend as little time hitting 1 target as possible without "ruining" important filler targets if cancels were bad.

Legs/Tail
The Hero stands as close to the ab line as possible to cleave Legs/Tail together. This was assisted with smoke mule usage with Bishop out of party at the start of the run. After Leg dies, the Hero goes up to 2 cleave LH/MH if they were available, otherwise 3 cleave Arms/Wings/LH.

RH
Hero would swap back and forth between 2 cleave LH/MH, 3 cleave LH/Wings/LA, and 3 cleave RH/Wings/RA, and 2 cleave RH/MH depending on the cancel alignment, current head health, Smoke availability, and Holy Shield availability. It is difficult for me to describe the comprehensive flow chart of how decisions were made in regard to what targets to prioritize during this phase, but essentially:

• If two heads were able to be 2 cleaved, this would be prioritized, especially if Smoke was available (on the right side)
• If Right Head previously cast weapon cancel multiple times and is taking longer than normal to die, the Hero would prioritize staying on the right side to help kill RH. This was done in order to remove RH as quickly as possible (without inconvenience) to remove the spammy 1/1s and reduce the chance of triple darks (since LH and MH were taking more damage than a normal run).
• If there were very few head cancels, the Hero could 3 cleave Wings/Arm in small windows since Wings would be relatively healthy (due to other attackers not hitting them), especially if Shield was available (on the right side)

When Smoke was used, the Hero would try to prioritize 2 cleaving as much as possible as the positioning required to do so is precise and vulnerable to Stance fails. When dropping to 3 cleave, especially if the Bucc was out of ST, Shield would be used to prevent RA status effects. Mid Head would dispel before RH died in over half of our runs.

LH/MH/Wings
A similar flow of decision making exists in this phase. If the head cancels are behaving well and attackers have little need to drop and hit Wings, the Hero would be allowed to drop and 3 cleave. If there were heavy cancels and Wings are low, the Hero would prioritize staying on Heads as much as possible, only dropping for double cancels.

Arms proceeded like standard.

Because of the decision making required to know what targets to prioritize, and the presence of additional Wyverns (due to the additional LH/MH damage earlier on in the fight), and the additional DPs, this strategy requires competent and experienced Hero and Bishop players coordinating throughout the run.

So what was the result?

Spoiler: Findings

https://docs.google.com/spreadsheets/d/1kQGi4xJYLeY-7KwmQ_CjxAaDBnonk9NIVHqNFwcmdzQ/edit?usp=sharing

The results were better than expected, but lower than I'd hoped. With an overall DPM efficiency of 103.23%, the threshold for proving Hero viability was reached. However, this comes with several important caveats:

• The Hero needs to be experienced. They need to be able to make decisions on the fly mid-run as to which targets need to be prioritized. They cannot simply just sit on Wings and Arms the entire run 3 cleaving or else the run becomes vulnerable to scuff with repeated head weapon cancels.
• The Bishop needs to be washed and experienced. There will be more DPs and Wyverns than your typical 6 man run. This is not problematic as long as the Bishop knows what they are doing, otherwise an inexperienced Bishop will become overwhelmed, negatively impacting the attackers' DPM.
• This relation to NL DPM becomes much worse at higher levels of funding. Because of how well NL scales and how poorly Heroes scale, the discrepancy between the two becomes worse at endgame.

That's a lot of info! Because that was a lot of explanation of why and how we did things, I'd like to take a second to summarize what we found.

• A single cleave (Hero or DrK, NOT Paladin or Shad) will, on average, speed up your run time for 6 man HT squads at "low to moderate" levels of funding (rough guess of ~0-15b worth of gear total) over an equally geared and skilled Night Lord.
• At the level of funding tested, a 1H Hero will be doing almost ~10% more DPM in Body than a similarly geared NL with perfect play and ideal conditions.
• At the level of funding tested, a 1H Hero is expected to be doing around ~13% more DPM in body than a similarly geared NL with perfect play and average conditions.
• With generous cancels and good conditions, the run time should be noticeably faster with the Hero DPM efficiency reaching over 110%. Additionally, poor cancels and conditions could be largely negated with good play from the Hero and Bishop and coordination between the squad, and as such poor conditions were often the same or even better than they would be with a standard squad.
• 1H Hero provides a lot of flexibility in HT and is worth considering instead of 2H, due to increased uptime on Right Side and SI independence. This has the added bonus of making the run more dc resistant as the Bucc or Bishop DCing has little effect on the Hero.
• At higher levels of funding (~15b+), Night Lords become difficult to replace with Heroes. This fact is largely made irrelevant by the inexistence of Heroes running HT at this funding range.
• Early LH/MH damage makes for a more involved experience for the Bishop. Additional RA damage also makes judging the HP bar more difficult for the Bishop. While this requires a more experienced Bishop to successfully navigate, it's more enjoyable than the typical 6 man Bish experience.

So, what does this mean for you?

Is it worth taking a Hero to HT?
Assuming they know what they're doing, absolutely, yes! The highs are higher and the lows are about the same or even higher.

Is it worth making a Hero to HT?
Debatable. While this proves Heroes are a powerful tool at low to mid levels of play, if you truly want to make a HT excelling character, Corsair is the way to go, and there is absolutely no replacing a skilled, geared, high level Corsair.

What does this mean for Heroes' place in the meta?
Overall, very little. Because Heroes still will get out scaled even with these findings, and there are so few Heroes at high levels of funding anyways, this changes little in terms of Hero viability. Heroes are still in need of additional assistance from BT, especially with the recent findings from iPippy (see here).

For now, Project Pippy is unable to continue due to lack of a Sed mule and as such, this is the end of this chapter of Project Pippy. In the future, I hope to revisit this content and look into the viability of 2h Hero instead with heavy smoke mule usage, in the hopes that Hero can find a place on actual endgame squads.

 

I will defer to your expertise in terms of the funding windows, I just threw out a rough estimate.

I agree, I would also like to see what 2h is capable of but lack the infrastructure to support one currently. As such, we will have to revisit this in the future.

 

Edited original post to reflect a data discrepancy in the Project Pippy spreadsheet. I forgot to update the run time for the last run, bringing the overall DPM efficiency to 103.23% (up from 101.43%). This outperforms a "typical" NL assuming 80% efficiency on an average run by 13%, significantly better than originally estimated.