A couple of graphs I put together. Both sheets are quick and dirty.
Broken out by ship performance:
https://drive.google.com/open?id=17jMXbt-CTLS8v1Wdl_n7qwFUIb_fVJSR
Broken out by variant occurrence:
https://drive.google.com/open?id=1Ff_Uhu_94slDyw622l2S1CLKSKnR9pWS
Inb4 thread full of QQ that W1-3 are dramatically overrepresented.
Really the most interesting thing to me is how flat most of these lines are, relatively speaking. Certainly, there are some clear trends, but for the most part there aren't a lot of really sharp slopes on the Occurrence vs Placement lines, which would indicate performance disparity.
My hypothesis is that--at least as far as ships are concerned--placement is largely determined by factors other than listbuilding (like skill, dice luck, matchups, etc). That's a very good thing IMO.
2 minutes ago, Ardaedhel said:Really the most interesting thing to me is how flat most of these lines are, relatively speaking. Certainly, there are some clear trends, but for the most part there aren't a lot of really sharp slopes on the Occurrence vs Placement lines, which would indicate performance disparity.
My hypothesis is that--at least as far as ships are concerned--placement is largely determined by factors other than listbuilding (like skill, dice luck, matchups, etc). That's a very good thing IMO.
I really love where this game is currently.
Can you walk me through what the occurrence vs placement graph actually means? I'm having trouble wrapping my head around it.
In other news I'm mildly surprised that there have been more Gozanti s than GR-75s so far. Had you asked me to guess, I would have flipped it and halved the margin.
5 minutes ago, GiledPallaeon said:Can you walk me through what the occurrence vs placement graph actually means? I'm having trouble wrapping my head around it.
In other news I'm mildly surprised that there have been more Gozanti s than GR-75s so far. Had you asked me to guess, I would have flipped it and halved the margin.
I think you have to consider that the hammer head and CR90 eat into them more then a raider would into a gonzati.
1 minute ago, GiledPallaeon said:Can you walk me through what the occurrence vs placement graph actually means? I'm having trouble wrapping my head around it.
In other news I'm mildly surprised that there have been more Gozanti s than GR-75s so far. Had you asked me to guess, I would have flipped it and halved the margin.
Yeah, sorry--easy to assume your assumptions are obvious when you're the one steeping in the data.
Occurrence vs Placement shows how many of each ship has shown up in each successive bracket, with analysis isolated between Rebels and Imperials (no particular reason, just easier to see the information I was specifically looking for that way). So, for example, the CR90 line shows that the average overall Rebel list contained .75 CR90's, while the average Top 4 Rebel list contained .88 of them.
Worth noting that plotting it this way gives extra weight to occurrence rates of less expensive ships like the flotillas and corvettes, but is normalized to number of activations. If you were interested in how much the average player invests into CR90's, for example, it would be pretty straightforward to normalize the charts for cost of the hull (though that graph isn't broken out by variant, which would have some impact on the quality of your data).
9 minutes ago, GiledPallaeon said:In other news I'm mildly surprised that there have been more Gozanti s than GR-75s so far. Had you asked me to guess, I would have flipped it and halved the margin.
And yes, that surprised me as well, and made me go dig in a little bit to check for mistakes. I think this is explained by the overrepresentation of Imperial lists overall (52 R - 80 I).
2 minutes ago, Ardaedhel said:Yeah, sorry--easy to assume your assumptions are obvious when you're the one steeping in the data.
Occurrence vs Placement shows how many of each ship has shown up in each successive bracket, with analysis isolated between Rebels and Imperials (no particular reason, just easier to see the information I was specifically looking for that way). So, for example, the CR90 line shows that the average overall Rebel list contained .75 CR90's, while the average Top 4 Rebel list contained .88 of them.
Worth noting that plotting it this way gives extra weight to occurrence rates of less expensive ships like the flotillas and corvettes, but is normalized to number of activations. If you were interested in how much the average player invests into CR90's, for example, it would be pretty straightforward to normalize the charts for cost of the hull (though that graph isn't broken out by variant, which would have some impact on the quality of your data).
Ok that makes more sense. The Y-axis is just missing a legend, so I wasn't sure what exactly was going on (might be a flub of shifting over to Google Sheets).
Just now, Ardaedhel said:And yes, that surprised me as well, and made me go dig in a little bit to check for mistakes. I think this is explained by the overrepresentation of Imperial lists overall (52 R - 80 I).
That miscue on representation certainly sounds large enough to account for our suspicions.
4 minutes ago, ripper998 said:I think you have to consider that the hammer head and CR90 eat into them more then a raider would into a gonzati.
also take into consideration how effective they are at moving squads via relay. The carrier is great for superior squadron activation(flight controllers ftw) but if you're just wanting to move squads it's hard to get more efficient than Gonzati's
I don't have the slightest idea what these mean or how to read them. Math sucks, and was invented by the devil. If someone could translate for us math challenged idiots, it would be highly appreciated.
Also I think you can see if you sort the list to small base sub 50 point ships only, A bit more variety on what a rebel list has then Imperial.
3 minutes ago, Darth Lupine said:I don't have the slightest idea what these mean or how to read them. Math sucks, and was invented by the devil. If someone could translate for us math challenged idiots, it would be highly appreciated.
Basically they are graphs to show the total amount of ships being used across all lists. The second graph removes the flotillas because they skew the graph with their high numbers so you can see if there are more patterns
6 minutes ago, Darth Lupine said:I don't have the slightest idea what these mean or how to read them. Math sucks, and was invented by the devil. If someone could translate for us math challenged idiots, it would be highly appreciated.
If one line is higher than another, that line's ship is taken more often than the other one. If it rises on the right, that ship is more common on top tables. If it sinks on the right, it's less common at top tables.
Thanks, Ard! Back in wave 5, I did something similar, but normalized across ship cost. Rebels were relatively ship-agnostic while Imps loved their ISDs. Top tables spent more points on flotillas, raiders, mc30s, and ISDs. It was also interesting which ships preferentially ended up at the bottom. (AFM, lib, arq, INT)
30 minutes ago, GiledPallaeon said:Ok that makes more sense. The Y-axis is just missing a legend, so I wasn't sure what exactly was going on (might be a flub of shifting over to Google Sheets).
Yeah, quick and dirty, like I said.
I may or may not get around to fixing it.
19 minutes ago, Baltanok said:If one line is higher than another, that line's ship is taken more often than the other one. If it rises on the right, that ship is more common on top tables. If it sinks on the right, it's less common at top tables.
This, exactly.
2 minutes ago, Baltanok said:Back in wave 5, I did something similar, but normalized across ship cost.
Feel free to yoink my logic for the main spreadsheet if you wish, but it's nothing revolutionary. I literally just threw these together in a few minutes to answer a couple of questions, and figured I'd post them for anybody interested.
4 minutes ago, Ardaedhel said:This, exactly.
Feel free to yoink my logic for the main spreadsheet if you wish, but it's nothing revolutionary. I literally just threw these together in a few minutes to answer a couple of questions, and figured I'd post them for anybody interested.
I'll probably put something like this in the main sheets eventually, but I wasn't sure exactly how to present it. Your graphs are a good idea.
Great work @Ardaedhel ! The occurrence v. placement graph lends more credence to the notion that the number of activations plays a role in performance. If all of the lines are flat except for the flotillas, that means that the number of flotillas is the biggest indicator of number of activations (something I'm confident we could have already guessed beforehand). Then, the fact that the curves for number of flotillas v. placement has positive slope indicates that these two variables are positively correlated — more flotillas, better performance. Ergo, more activations, better performance.
Note that I have been cautious to not say 'causation'. It could be that the best players simply bring more activations. But my hunch is that the number of activations is having a measurable impact on the performance of a fleet. I just cannot state that for certain.
Assault Frigates seem to be more popular than I'd realised. I'd expected the GR-75s, CR-90s and MC-30s to be most numerous in Rebel lists, but I thought the MC80H would see more play than the Assault Frigate. Interesting.
Thanks for creating the graphs!
good data. with the exception of two ships, almost everything is viable to win or get in the top 4. That and almost everything is getting play.