Meta Analysis: Wave 2

I limited my data to 13 hyperspace trials (regionals) since the Upsilon nerf with mostly full lists entered. That includes Blacksun Trials (AU), Redmond (US), Malmö (SE), Germany, Bromley (UK), Ohio (US), Bathurst (AU), Stockport (UK), Kentucky (US), Minnesota (US), Arizona (US), Austin (US), Quebec (CA).

That means I have 624 lists for swiss and 110 lists for the cuts, with an average of top8 per tournament. That is a 17.6% conversion. Every ship or pilot with larger conversion rate performs better than expected by pure chance.

Out of the 624 lists, every third (210, 34%) was Rebel, every fourth (158, 25%) Resistance, over a sixth Empire (111, 18%), and just few were First Order (76, 12%) or Scum (69, 11%).

Fig. 1

In figure 1 you see the barplot for these numbers, with lists on the y-axis and the amount of ships per faction labelled on the bars. On average, the different factions brought 3.7 (rebels), 3.68 (resistance), 4.78 (empire), 3.46 (FO) or 3.36 (scum) ships per list.

Going to the cut and figure 2: Rebels (39, 35%, 4.05 ships/list) and Resistance (31, 28%, 3.81 ships/list) are very similar. Imperials increase a lot (27, 25%, 5.26 ships/list), while the other two are decreased for FO (8, 7%, 3.25 ships/list) and scum (5, 5%, 3.4 ships/list). More important might be the conversion rates: Scum and FO are very low, with just 7% or 10.5%, respectively. Rebels are close to the pure chance-based value (18.6%) and Resistance are a bit above (19.6%). But imperials get 24.3%, which is a plus of 7% and quite important. This tells us that rebels and resistance in part just advance because they are played a lot, while imperials advance because they are doing very well!

https://i.imgur.com/JnOBkLu.png
Fig. 2

Moving on to the list composition. First off, generics vs uniques with a 73.5% to 26.5% split if looking at all ships. Maybe more interesting is that 578/624 lists used at least one unique pilot. That leaves just 7.4% of lists or 46 lists that used only generics. Keep in mind that lists like BBBB Zeb count as non-generic list because of Zeb. I do not have a way to count all lists with mostly generic ships, as it is an all-or-nothing. Most frequent generic ships in these purely generic lists were the rebel Y-wing (119), followed by T65 X-wing (26), TIE /sf (20), RZ2 A-wing (18) and 5th place is shared by Firespray and T70s at 9 each. Strikers, Upsilon, U-wing, MG100, YT1300 and silencer are all very low.

Fig. 3: 73.5% of all pilots were unique

More list composition: figure 4 and 5 show, for swiss and cut respectively, how frequently a certain ship type was used in that faction. The y-axis is the amount of lists total, but the bars themselves are labelled with the percentage per faction. To give an example: a good ¾ of rebel lists used at least one (or more!) X-wing. The hyperspace format means that only relatively few ships are available, and that is clearly visible in the percentages. The interesting part is the comparison of swiss and cut. To give another example, U-wings or Y-wings strongly increase, as did RZ2-Awings. Strikers, Reapers and the advanced all decrease, telling us that players using them did not make cut as frequently as those using the workhorse that is the TIE fighter. Scum and FO suffer from low numbers, but at least we can tell that a scum list really should include a Fang fighter.

sPK96LM.png
Fig. 4
6DrBKeI.png
Fig. 5
Fig. 6

We can do the same with pilots instead of ships, but I had to limit the amount of bars for practical reasons. Just looking at the 30 most frequent pilots (Fig. 6), we see a lot of imperial, rebel and resistance. This represents of course that these three factions made up 77% of all lists. Outstanding are Wedge and Lulo. Wedge is the MVP for rebels and in 61% of lists. But Lulo is even more popular and 70% of resistance lists use him! But both Wedge and Lulo are surpassed by Quickdraw, who is in 75% of all FO lists. Granted that there are not too many, but it is important to consider. Resistance has a second group of popular pilots, with Poe, Tallie and Nien. This serves as first hint for what combination is most popular in the resistance faction. The popular rebel pilots after Wedge are Dutch Vander, partisan renegade, gray squadron bomber, and Luke Skywalker. Imperials are led by Darth Vader, followed by TIE fighters.

Fig. 7

Going to cut, again frequency of pilots (Fig. 7). Scum and FO are gone from the 30 most frequent pilots, and Lulo is now dominating not just his faction with 84% but everything else, too. And we get another hint that resistance is mostly Lulo, Tallie, Poe, and Nien. Rebels are similar as in swiss. Wedge gets taken down a noth, but the partisan renegade is an example of a pilot who really increases your chance of making cut. Dutch barely increases with 2%, as does the gray squadron bomber with 3%. Biggs sneaks into the top5 of pilots. Luke gets declassed and is nowhere to be seen. The message is clear: don’t use Luke, use Biggs instead. Imperials have massively increased frequencies for the TIE fighter pilots: whatever the exact build of your swarm, you should take Howl, Iden, an Academy and Gideon. The rest is up for taste.

Next I will take a look at hit points and initiative (you might see PS for pilot skill, which means the same). Figures 8 to 11 are not shown here but can be found in the album on imgur. Hit points analysis (Fig. 8) shows that each faction has their typical spot: Rebels have six or eight HP (X-wing and U/Ywing). Resistance has four or seven (RZ2 and T70), and imperials have three or five (TIE fighter and Advanced). This information is not immediately useful, but you might want to consider it when thinking about particularly difficult matchups. The cut shows almost exactly the same pattern, with the exception of 3 HP. This highlights yet again how successful imperials were, and how their swarms helped making cut. Maybe more interesting is the total HP of a list (Fig 10 and 11). Imperial list have a low amount of health points, resistance occupies the intermediate range, and rebels are mainly upwards of 26HP. Going to the cut, it looks like higher HP lists are much more successful for rebels, while lower HP squads are more successful for resistance and imperials.

The initiative has cooled off a bit since December and compared to extended. The i5+i6, largely due to Resi5tance, still make up 43% of all ships! Largely resistance means, 285 out of 623 i5 ships. And we’ve seen before how massive the contribution of Lulo is: 111 out of those 285 resistance, or 111/635 i5 ships are Lulo. That is massive! Tallie (68), Poe (64) and Nien (58) are close together. High initiative rebels are mainly Wedge with 129 out of 274 i5+i6 ships. Luke (40), Norra (32), Lando (32), Han (21) and Thane (20) are the others. For comparison, Vader had 60 showings. You can see the bars in Fig 12 for swiss and Fig 13 for the cut.

Related to initiative are bids. Here I have one interesting graph that is a tad more complicated and in nice baby blue/pink. Fig. 14 shows pilots vs count for high initiative pilots, decreasing left to right from frequent to rare. The dataset is filtered for lists where the total point cost is entered, so I have slightly fewer lists/ships than in the rest of the post. Now the pink bars are the same lists, but filtered for small/medium bids of 0-2 points. This means that pilots with a large fraction of blue bars visible had deeper bids than 2 points. An example is Poe, Nien or Ello. Apparently, players like to bid deeper with those pilots. Wedge on the other hand has more often than not a smaller bid of 0-2 points, as does for example Howl.

B0E9zFA.png
Fig. 14
https://i.imgur.com/X3lzETt.png

I took a look at archetypes, too. The 624 lists had 138 different combinations of ships (not pilots!). Of course some are very similar, but I have to rely on automated processing and that means that is as resolved as I can make it. The most frequent lists are shown in table 1. These 28 archetypes amount to 398/624 lists, so the majority (64%).

But the most important part is the conversion rate. Keep in mind, everything above 17.8% is good, below is bad. On top, a higher frequency means that it will regress towards that cutoff. So a high frequency AND a high conversion rate is very impressive. Here are all the lists as image because tables are messy to include.

xNcaZAO.png

There are plenty of points that can be made here:

  • Imperials have the most amazing conversion rate:
    • TIE swarms, 43% made cut!
    • TIE salad performs rather poorly at 15%
    • But Vader +Miniswarm is again very good at 27%
  • Rebels are great and it looks like many combinations of X, Y and U work:
    • XUUU at 67% (but just 3 lists!)
    • XXYU at 41%
    • YYYU at 33%
    • XYYU at 22% is just a bit better than random
    • XXXYY at 40%
    • YYYYY at 29%
    • YYYY at 33%
  • The resistance is not as great as commonly “known”, but they have some good lists. As usual, high initiative is good, indicated by the XXAA list. Also high efficiency plus two high initiative as shown by the 5As which commonly have Lulo and Tallie:
    • XXXA at 21% is just a bit better than random
    • XXAA at 30%
    • XAAA at 27%
    • AAAAA at 33%
    • XXA at 14% is not that good

Last figure is the winrate. We see again how frequently used Wedge and Lulo are. But now the x-axis is the amount of won games for each pilot. For example, Zari won 2/3 of all her games (56 wins, 28 losses). There were not enough cuts and not large enough cuts to get a meaningful distribution. But that in itself indicates that there are many different lists that advance.

IPLRgnY.png

I can always take a closer look at individual requests. Wave 2 meta might not be interesting anymore, but more timely posts certainly are.

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