As you might have already guessed, today’s Skill Build Analysis has a bit of a theme. It’s been a month since her release, and Legion Commander is still the 2nd most played hero in Dota2’s public matchmaking. It’s likely that much of this popularity is driven by her ability to jungle, as the general public loves anything with an ounce of carry potential that has a respectable early neutral kill rate. With that in mind, today’s SBA theme is that jungling is not always a good idea, so before we move on to builds, let’s take a moment to look at the supposed efficiency of jungling.
The biggest arguments that get brought up in favor of jungling is that using one nets you a fourth farmer. From an Experience per Minute perspective this is a relatively reasonable conclusion, provided you have a decent selection for the offlane that won’t get zoned out. However, if you are more focused on the gold economy the benefits of a jungler are severely overstated. With rare exceptions, a solo offlane just isn’t going to be making bank against a competent safe lane. They’re not supposed to. By committing to a jungle lineup you are trading away your claim to a risky, but larger farm source in favor of a safer but smaller guaranteed flow.
But when we phrase it like that we miss half of the story. Not only are you downgrading your ability to convert those offlane creeps into items by jungling, you’re also weakening your ability to harass the opposing safe laner farmer. Suppose the enemy team picks Dazzle, and you immediately grab Axe because Culling through Shallow Grave gets you off. Your team picks a Dark Seer, so it’s looking like a pretty standard jungle Axe, but then the enemy team closes with an Antimage pick. Should you still jungle at this point? Dazzle/Antimage is a pretty weak lane, and Axe and Dark Seer can cause them a lot of problems with both Ion Shell and Battle Hunger for attrition followed up with a Surge into Beserker’s Call for the kill. And even if the lane isn’t productive from a killing perspective, if your presence costs Antimage a significant amount of CS you’re successfully delaying the biggest time bomb in the game. This isn’t to say that you should always duo offlane in this scenario. Maybe the enemy team has a jungler of their own available that can make Axe hijinx unacceptably risky. The real point is that we often think of jungling in this scenario exclusively along the lines of +150 GPM for the Axe and forget that it can also mean +100 GPM for the Antimage. In my opinion, that exchange rate would not be in Axe’s favor.
Finally, the last cost involved with jungling, and the one we’ll be focusing on today, is that it can require you to adjust your skill build to effectively jungle. This can be a liability when it sucks points away from other skills that are more productive outside of the jungle.
So with the intro covered, here are the skills we’re looking at for Legion Commander:
At first glance, Moment of Courage looks like your standard early one point wonder. The proc rate increase per level is fairly weak (16/2/2/2), and while the lifestealing scaling is significant, percentage lifesteal is most effective at higher levels of base damage.
Some one point wonder, huh? My belief is that it’s likely the case that Legion Commander’s skill build correlates strongly with whether she’s played in a lane or the jungle. The high prevalence of Moment of Courage(E) builds indicates a strong preference towards jungling, and the distinct minority of Overwhelming Odds(Q) builds are likely the people playing her in lane. Press the Attack(W) builds are more ambiguous, but as it turns out this doesn’t seem to matter much. Incidentally, this is also the largest SBA sample yet, with the Very High bracket including over 18 thousand games.
In all three brackets the presumed laning build of Overwhelming Odds(Q) wins out over the Moment of Courage(E)-first jungling build, but interestingly the gap in success closes significantly in Very High. Maxing Press the Attack(W) first never looks like a good idea, but Q->W->E builds do seem to outperform Q->E->W builds in Very High (though at this point we’re getting into very small sample slices). Legion Commander as a whole appears to fall off in success as you move up in the brackets, which is interesting because preliminary results indicate that she is picked more often in higher rated games. I also ran a quick farm dependency test on these matches and her score would place her in the top 20 most dependent heroes in previous runs.
In any case, this illustrates the point I was making earlier. Overwhelming Odds appears to be Legion Commander’s most effective early skill, but jungling is perceived to require early investment into Moment of Courage. Jungling with Legion Commander is certainly viable, but there’s a tradeoff involved, and that tradeoff won’t always be worth it.
With this in mind, I thought I’d bring in Lifestealer and Ursa for comparison. Both heroes get used as junglers and have a skill strongly associated with their jungle builds that happens to be a passive with relatively poor scaling. Let’s look at Lifestealer first.
Here we can see that Lifestealer’s jungle passive Feast(W) is the dominant build, particularly at lower skill levels. What’s really interesting is that this is the highest incidence of Split builds in Very High that I can recall seeing in any of the heroes that I’ve looked at. Traditionally, split builds as a whole do not do very well, get less common in higher skill brackets, and are generally extremely unpopular in higher skill brackets. Lifestealer violates all three of these expectations, and his Very High Split build rate is almost double the next closest I can recall seeing (Viper at ~22%). Those peculiarities aside, how are his builds doing in terms of win rate?
Like Legion Commander, the jungle-oriented passive first build underperforms the lane-oriented build, but it’s Open Wounds (E) not Rage(Q) that appears to be the top performer. Alternatively, maybe Open Woulds-first is a somewhat viable jungle build, but either case it does appear to be Lifestealer’s most effective build option at the moment. And if we break things down to a point-by-point comparison:
You can basically see the emergence of 2-1-4-1 as the dominant level 8 build provided you ignore High which is probably a sample irregularity. Lifestealer’s sample was smaller than average.
Ursa is possibly the most interesting case of the three because his jungle passive has the most utility. Fury Swipes(E) doesn’t have the best scaling per point, but skilling it up improves his ability to sneak an early Roshan. You might also recall that a year ago I made a case for Earthshock(Q) being a highly undervalued skill. So how do the skill builds shake out.
Unfortunately, Earthshock(Q) builds are still a massive minority. I went out of my way to triple the sample size to alleviate this, but the win % data on them will still be somewhat unreliable. Meanwhile, our expectations about Fury Swipes(E) proved correct. It’s the dominant build, and unlike Feast, it actually becomes more popular in higher skill brackets. But does its win rate hold out?
This is honestly the closest Skill Build Analysis that I’ve done. All three primaries are within one percentage point of each other in Very High, and there’s no consistent trends across all three brackets. What I can say is that Fury Swipes(E) appears to be the strongest ability in Normal games, but this advantage dissipates in higher brackets where players are better at using his active skills. Meanwhile, both Earth Shock(Q) and Overpower(W) do fine in the upper brackets. I think you could honestly make a case for a wide variety of skill builds as long as you pick one and stick to it. You’ll notice that out of the low performing, red arrow builds, 9 out of 13 were split builds, and virtually every exception (both Split builds that did well and non-split builds that did poorly) happens to represent a tiny sliver of the total games. Yes, it’s possible that you’ll start with a build then come to realize it’s inappropriate given the conditions of the game, but you want to both minimize the number of games where you need to make this kind of adjustment, and minimize the impact of such adjustments by making them early and fully committing to them.
As for jungling, we’ve looked at three heroes with significant carry potential and jungle capability today, all three have jungling passives as the centerpiece to their most popular builds, and in all three cases these builds were not their most successful. None of this is to try to prove that these heroes should never be jungled (at least for Legion and Ursa…) but you should really consider laning them if your team composition allows. They’re all pretty effective heroes in lane, and as long as you can last hit and your lane isn’t a disaster, lane gold should drastically outpace jungle gold in the early game.
Since this is still ostensibly about Legion Commander, let’s finish it off by looking at her item trends:
I don’t like looking at item win rates too much, they can be terribly misleading due to a number of factors, but there’s a trend here that I feel warrants it.
These three items are Legion Commander’s most likely opening big purchase. From the previous chart we can see that Armlet is significantly more popular in Normal and Shadowblade and Blink Dagger are far more likely to be built in Very High. My immediate reaction upon seeing the item stats was “Armlet is in her official core item list, isn’t it?” And of course it was. What’s clear is that Very High players are prioritizing initiation tools over stats, and they’re likely correct in that decision. It’s not something we can test for through the API unfortunately, but I strongly suspect that creating successful Duels is one of the best predictors for a successful Legion Commander game, and Armlet isn’t going to do a lot for you there. Meanwhile, if you’re not building an Armlet early, Desolator and Assault Cuirass likely outpace it in total utility for the mid to end game, leaving Armlet a significantly less popular item in the upper brackets. Given its cost, Blink Dagger appears to be the best bang for your buck, provided you’re capable of finding successful initiations. And if you can’t, you should probably get better at that, since it’s a pretty integral part of the Legion Commander experience.