Meta Analysis: Mid Wave 3

Wave 3 dropped on the 20th of March. Since then, the results of 15 Hyperspace Trials primarily from the US were entered in listfortress: Krakow, Santa Clara, Wyoming, New South Wales, Lima, Nürnberg, Texas, Alaska, Flint, Milwaukee Maryland, Paris, California, Florida and Minnesota. Unfortunately, players or TOs in Ontario, Dublin, Cardiff and the first of the Florida Trials did not enter their data. With 7 factions I need considerably more lists to draw somewhat reliable conclusions. If you are reading this, you are probably interested in data about the meta. So please, tell your TOs to add the information to listfortress – or even better, collect it yourself!

The 15 trials resulted in 655 lists in swiss and 140 making it to the cut, giving a 21.4% conversion rate baseline. As last time, I’ll come back to the conversion rate later.

For the 655 lists, 14% would be perfectly balanced. But what we see (Fig. 1) is 210 Rebels (32%), followed by 133 Imperial (20%). The rest is very similar at 77 Republic (12%), 73 Resistance (11%), 56 Scum (9%) 54 First Order (8%) and 52 Separatists (8%).

Fig. 1

The squad size is also always interesting to know, though it didn’t change much since wave 2. Rebels increased a bit to 3.79 (from 3.7), Resistance is now 3.93 (3.68), Empire decreased to 4.59 (from 4.78), FO has a small increase to 3.52 (3.46), Scum went down to 3.09 (3.36), and the new factions start at 3.74 for Republic and staggering 5.04 for Separatists!

Going to cut and looking at the same metrics in figure 2, we see percentage increases for Rebels (to 36%, +6%) and imperials (to 28%, +8%), demonstrating their current strength. Republic stays at the same 12%. Resistance decreases by a third to 7% while FO increases a bit to 9% (+1%). Separatists drop (to 4%, -4%), and Scum is hit the hardest (to 3%, -6%). This further cements the widespread opinion that scum is in trouble at the moment. I do not necessarily agree, but these number certainly point in that direction. The conversion rate is this time not much more telling: Rebels (29%) and Imperials (25%) are above, Republic (22%) and First Order (22%) pretty much on it. The others are below: Resistance (14%), Separatists (10%) and Scum (7%) all seem to have trouble.

Fig. 2

The comparison to wave 2 for Resistance shows that even back then they were not overperforming as they brought pretty much as many to cut as to swiss. One possible explanation is that more players bring Resistance after hearing about the success, only to find that it is not quite as easy as rumor had it to make cut with four i5 Resistance pilots. A better explanation though for that particular list is probably that the meta changed to more high initiative ships, and now these i5 are not dominating anymore. You will later see that they underperformed a lot. As last basic metric for the cut, the squad size remained everywhere pretty much the same – with the notable exception of Separatists, where the size dropped from 5.04 to 3.2! The second notable change to wave 2 is that imperials don’t increase their size drastically going from swiss to cut, and instead even slightly reduce from 4.59 to 4.54. Resistance is still the four-ship high-initiative faction, and that might be their downfall for the moment.

Moving on once more to list composition. Generics vs uniques continues to amaze me as the amount of generics further increase to a mind-blowing 33% (Fig 3.)! This is something I have not seen in the two years I’m looking at data like this. The high amount of squads using at least one unique is unchanged high (93%). The purely generic lists use again mostly Y-wings (74), T65 X-wings (24), TIE Advanced (19) (?!), TIE SFs (18) or ARCs (Republic, 14).

Fig. 3

One of my favorite plots shows which ship or pilot is used in how many lists. Comparing swiss (Fig. 4) and cut (Fig. 5) allows us to see at a glance which types of ships made cut more often. So far in wave 3, the best increases can be seen for Fangs (+43%), Sith infiltrators (+35%) and Starvipers (+21,4%), but small sample size makes this less interesting. Rebels see an increase for the X- and Y-wing (+5 and +6%) while B- and A-wing and the YT1300 have a decrease. The empire gains a bit with the Advanced (+4%) or the reaper (+6%), while interceptor (-4%) and surprisingly the TIE fighter (-4%) seem to be a trap to some degree. The next three also have too few squads to make the numbers meaningful, so don’t treat their single digit percentage with the same relevance as those for rebels or imperials. Republic apparently should less focus on Aethersprites (-4%) or torrents (-13%) and more on ARCs (+14%). First Order should bring the /SF, but the rest remains similar. Resistance should bring even more often their A-wings (why don’t you have one?!) and ignore the YT1300 or the MG-100 entirely. Separatists have the mentioned huge infiltrator increase, while vultures get the largest drop with -45%! More infiltrators, fewer vultures!

Fig. 4
Fig. 5

These next two graphs (Fig. 6 and 7) are a small variation of the same, but here the percentages on the bars are now telling us how many squads across factions had one of those. For example, every fourth squad had at least one T-65 X-wing, or 10% of all squads brought an Aethersprite – which is already more popular than the RZ2-Awing.

Fig. 6 has different percentages on the bars – here they tell you how many squads independent of the faction used a certain ship.

Fig. 7

But that was just ships when we’re often more interested in pilots. Of course we can do the same there. There are more pilots, so I limited the list to the 30 most popular, meaning most frequently used, pilots of hyperspace. The “who’s who”, kind of. And just like any other popularity contest, it does not say too much about the quality. Sure, more lists containing a pilot in cut means that many did well with it. But we also have to consider the conversion rate. Looking at which pilot gets an increase from swiss to cut tells us that this pilot seems to be necessary or at least beneficial for that faction to make it to cut. But it is the conversion rate on top that tells us whether a ship is overperforming compared to the rest of the field, not faction! And that is in the end something we’re much more interested in. Here’s the table if you want to see more numbers.

Fig. 8: same type of graph, but now looking at Pilots!
Fig. 9

Anyway: The king is dead, long lives king Lulo! He increased from his 70% in wave 2 to whopping 81%! Four out of five lists for Resistance bring Lulo. Next is Tallie at 67%, followed by Quickdraw (67%) and the most frequently used overall (one in five lists!), Wedge at 63%. Vader manages 47%, while his younger self is more popular (57%). Again the changes swiss to cut are more interesting. Nien Nunb sees the largest increase (+20%), followed by the 104th ARC (+17%), Quickdraw (+17%) and Del Meeko (+10%). Lulo still makes the list of successful pilots with +9%, as does Wedge (+9%). Here, the Dark Lord wins with +5% vs Anakin with -4%! Also, you really shouldn’t bring V19s (Goldsquadron Troopers, -10%). Important to note that while Nien and Lulo help Resistance making it into cut, they also have a conversion rate at or below random (21% and 15%, respectively). The 104thARC has also a great conversion rate at 34%, but I will get back to all that at towards the end.

I skip the hit points this time. It is not as interesting, and I haven’t integrated the changes for the Delta7 or the Bellbulab. So directly to initiative!

Rebels and First Order are all over the place except i1. The Empire occupies i4, Resistance has taken i5. Separatists sit at i3 and i1, while the Republic has limited itself largely to i6, i4 and i2. We’ll watch their career with great interest. Scum… there’s so little, but their i5 and i6 make at least cut, so that’s a plus. If you want, you can look at the initiatives for cut here. Also, 531 out of the 655 lists (81%) had at least one ship with i5 or i6. 269 lists had just one, 163 had two such ships, and 99 had 3 or more. Overall, around 55% of all ships were i4, i5 or i6. That means the game has a majority of higher initiative ships at the moment. But this is quite a reduction from wave 2 when i5+i6 alone made 43% of all ships, and i4-i6 made 63%!

Fig. 10

The next figure shows only high initiative pilots, i5 and i6. The y-axis shows their number. The blue bars are all lists, while pink shows lists with 0, 1 or 2 points bid. That is interesting because it tells us that only a faction of Wedge bids deep, while most lists with Anakin have somewhat deep bids. So ultimately this graph tells you whether you should bid deeper to fight a certain pilot. If you’re afraid of Vader (at 47% bids above 2 points), then you don’t have to bid more than 3 to underbid half of the lists. In principle. But of course that is not true for the one game that you have in that one round where it really matters. I can easily check other bids if you want, just leave me a message. For example, a 5pt bid underbids 75% of all Lulos (+25%, from 50% at the previous 3pt bid), but it only underbids 9 more Vaders out of 62 (+14%). And looking at all lists with entered points, only 22.5% had a bid of 195 or deeper. Accordingly, you should be able to get your choice quite often with 195 points.

Fig. 11 shows i5 and i6 pilots (blue) and how often they had a bid of more than 2 points (pink).

Further on to archetypes, or rather a certain combination of ships (table 1). Because we all know that XYYU and XXYU are the same archetype, but they are two different combinations. I limited the following table to squads (=combinations) that occurred four times or more often. That adds up to 64% of all combinations. That is the next interesting point: there were now 206 different combinations of ships! Of course that increase from the already massive 138 in wave 2 is in part due to the two new factions. But it is fascinating just how many lists people create and compete with in the “limited” hyperspace format! I still stand by saying that the meta diversified even further since wave 2, because those 64% are made up of 51 different lists, where wave 2 had only 28! I can’t even show that many in one picture, so this next figure (table 1) has 38 lists that add up to just 56% (368/655) of all lists in swiss.

Further on to archetypes, or rather a certain combination of ships (table 1). Because we all know that XYYU and XXYU are the same archetype, but they are two different combinations. I limited the following table to squads (=combinations) that occurred four times or more often. That adds up to 64% of all combinations. That is the next interesting point: there were now 206 different combinations of ships! Of course that increase from the already massive 138 in wave 2 is in part due to the two new factions. But it is fascinating just how many lists people create and compete with in the “limited” hyperspace format! I still stand by saying that the meta diversified even further since wave 2, because those 64% are made up of 51 different lists, where wave 2 had only 28! I can’t even show that many in one picture, so this next figure (table 1) has 38 lists that add up to just 56% (368/655) of all lists in swiss.

Table 1

Combining the squads reduces the amount of course drastically though. I didn’t count all that for swiss. But the rebel beef… I’d argue that within the lists that made it to cut somewhere (table 2), at least 7 combinations are rebel beef with Leia, for example. You can make your own judgment with the list here.

Table 2

If looking at the conversion of these ship combinations (still table 2), we see numbers all over the place. You can compare them to the wave 2 results. Interesting to me is how the 5 Awings went down from 33% to just 14% and below the threshold. Or how well rebel beef does, particularly less widespread versions like XXXU, XYBB or YYYU. Another interesting case is the 50% conversion for Resistance’s XXA list. And we also get a high, 56% conversion for a YT1300+X+Y (but only 20% for YT+X+X!). The high conversion imperial lists are all with low numbers (2-3 showings), so that’s unreliable. The Republic has great success with a Delta and 3 ARCs at 46% (6 from 13 showings), which is much better than the 28% for two Deltas and one ARC, or quite bad 12% for 1 Delta 2 ARCs or the slightly better triple Delta (14%). I won’t go through all 74 lists that made at some point cut (45 just once!). They include lists as fun as two T65 and a RZ1 Awing, or 5 TIE /SF. But just once means we really can’t draw any conclusion from it.

The last two figures are more colorful. Figure 12 compares the win percentage with the amount of games that were played with that pilot. I had to limit the pilots to those played at least 10 times for “readability”, leading to around 60 games (if all tournaments had 6 rounds). Also important to mention that these win percentages do NOT include cut rounds. The reason is first that I’d have to compare the final ranking with the cut size and add wins/losses accordingly. And second, that cuts are notoriously tricky and player dependent. You might have noticed that I never look at how well something did within the cut, and that is why. Also, keep in mind that a large number leads to a value closer to 50% if the pilot is very well balanced in this meta. The outlier is – hilariously enough – Horton Salm, who was barely played but did very well. You can also appreciate the cluster of TIE fighters who did very well. Next follows a wall of red, from Dutch to Han. I know that the readability is horrible, but that comes with a diverse meta. I can make more limited versions if you want.

Fig. 12 shows the win percentage (wins /wins+losses) vs total games. The dot scales with the number of squads using that pilot.

And finally the conversion rate for pilots against their frequency in the cut (Fig. 13). Now combine and crossreference this with the “% of faction” in swiss vs cut from the beginning in figures 4 and 5 or the table. That told us which pilots help a faction to make cut, and an increase hinted that the pilot might be necessary. Here, we have two different conclusions based on the position left or right of the random conversion. Left of it, below 21%, the pilot failed a lot. If it still has an increase in the comparison swiss vs cut, then that pilot has probably a low performance floor and a high performance ceiling (something which I call a large dynamic range). An example is Lulo, who had an increase from swiss to cut (+9%!) but only a 15% conversion rate. One interpretation is that he helps better players make cut, and helps lesser players lose more often. Pilots on the right of the vertical line can be suspected of carrying a list, or at least of overperforming. Del Meeko, Norra, Partisan Wampa or Garven are such candidates. They all require their own spotlight to answer why they did so well. But if every other list with a certain pilot makes it to cut, then that should be a hint that this pilot is useful.

Fig. 13

That’s it for this time. Wave 3 is still young, and wave 4 should still be 9 weeks or so away. That is plenty of time to explore more options for all factions. For now, Rebels with their beef dominate. But Separatists with their double Infiltrators or TIE swarms can take them on, and maybe the Rebels will be pushed down a notch or two again.

Until next time!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s