23 minutes ago, stonestokes said:

Edit: though this might be misleading, because some of those data points are actually stacks of data points, instead of just a single entry. I'll have to figure out how to represent that.

https://www.myexcelonline.com/blog/bubble-chart-3-variables-on-a-chart/

Thanks. I replaced it with vertical stacks instead. I was worried that the bubbles would start to interfere with each other. That, plus I am biased against bubble charts for a variety of reasons.

Here is the updated graph again:

Note that the 0-ship fleets represent data points where the placement was known, but the number of activations was unknown.

Without more access to the data/more stats tests run on it, I see fours and fives dominated the lower end of the chart, but also dominating in terms of numbers present, which will require at least a small amount of work to dissociate.

17 minutes ago, GiledPallaeon said:

Without more access to the data/more stats tests run on it, I see fours and fives dominated the lower end of the chart, but also dominating in terms of numbers present, which will require at least a small amount of work to dissociate.

Here is another look at it.

The TOTAL column is the number of fleets with a given number of ships. TOP 4 is the number of those fleets which finished in the top 4. TOP 4 % is that same number expressed as a percentage of the total number of fleets of that size. Likewise for the top 8 columns.

Even though 4-ship fleets were most common, only 9% of those who brought 4 ships to regionals last year finished in the top 4. Yet almost half (48%) of those who brought 4-ship fleets finished in the top 8.

On the other hand, 6-ship fleets were very successful last year. 36% of 6-ship fleets finished top 4, and just as many finished in 5th to 8th place.

All of this is based off of this Wave 5 data set:

30 minutes ago, stonestokes said:

Here is another look at it.

The TOTAL column is the number of fleets with a given number of ships. TOP 4 is the number of those fleets which finished in the top 4. TOP 4 % is that same number expressed as a percentage of the total number of fleets of that size. Likewise for the top 8 columns.

Even though 4-ship fleets were most common, only 9% of those who brought 4 ships to regionals last year finished in the top 4. Yet almost half (48%) of those who brought 4-ship fleets finished in the top 8.

On the other hand, 6-ship fleets were very successful last year. 36% of 6-ship fleets finished top 4, and just as many finished in 5th to 8th place.

I think this provides for some clear analysis. It suggests that while you don't have to take a high activation fleet to do well, you do well if you take a high activation fleet (for the most part). Hence, we can see that activations, and thereby first/last, are a high (if not highest) priority.

14 minutes ago, ryanabt said:

I think this provides for some clear analysis. It suggests that while you don't have to take a high activation fleet to do well, you do well if you take a high activation fleet (for the most part). Hence, we can see that activations, and thereby first/last, are a high (if not highest) priority.

I've taken some care to not state things quite that emphatically. Here is my opinion summarized:

1. Activation count is somewhat important and is at least a mild predictor of performance with more activations having at least a mild advantage;
2. The law of diminishing returns is at play here as well — there is such a thing as too many activations;
3. Six seemed to be the sweet spot for the number of activations, but that is probably also in part due to the fact that most people had 4- and 5-ship fleets;
4. Therefore, my recommendation would be to guess what the modal number of activations will be at a tournament, then have +1 or +2 on top of that.

Also, this analysis is obviously limited, and must ignore player skill — perhaps all of the best players brought 6-ship fleets. That is much more difficult to discern.

1 minute ago, stonestokes said:

I've taken some care to not state things quite that emphatically. Here is my opinion summarized:

1. Activation count is somewhat important and is at least a mild predictor of performance with more activations having at least a mild advantage;
2. The law of diminishing returns is at play here as well — there is such a thing as too many activations;
3. Six seemed to be the sweet spot for the number of activations, but that is probably also in part due to the fact that most people had 4- and 5-ship fleets;
4. Therefore, my recommendation would be to guess what the modal number of activations will be at a tournament, then have +1 or +2 on top of that.

Haven’t Rebel lists had a slightly higher winrate lately? If 6+ activation lists are doing well, that might be a factor.

19 minutes ago, stonestokes said:

I've taken some care to not state things quite that emphatically. Here is my opinion summarized:

1. Activation count is somewhat important and is at least a mild predictor of performance with more activations having at least a mild advantage;
2. The law of diminishing returns is at play here as well — there is such a thing as too many activations;
3. Six seemed to be the sweet spot for the number of activations, but that is probably also in part due to the fact that most people had 4- and 5-ship fleets;
4. Therefore, my recommendation would be to guess what the modal number of activations will be at a tournament, then have +1 or +2 on top of that.

True, I stated my conclusion in a provocative way and one which also lacked nuance. I believe, however, that it remains true. If by high activation we mean "more than the majority of fleets," then it absolutely fits.

Of course, there are diminishing returns since the number of activations can lose some of the punch you have in your big hitters. However, it is absolutely clear that when planning a fleet, you should have a very clear and compelling reason NOT to have more activations than your opponent. That, to me, suggests that first/last is a significant and unwelcome game mechanic.

I would be reluctant to make any blanket statements on any of this data. We have a lot of correlations, but we have no good way to pick out causality. For myself, I know that the players in my area that bring six ship fleets tend to be those I consider more skilled, not in the least because they need to be able to manipulate a six ship fleet effectively. Which direction is that trend actually going? Thus we have a chicken and egg problem, one I don't believe math can solve. (Someone better at stats than me is welcome to contradict that.)

10 minutes ago, GiledPallaeon said:

I would be reluctant to make any blanket statements on any of this data. We have a lot of correlations, but we have no good way to pick out causality. For myself, I know that the players in my area that bring six ship fleets tend to be those I consider more skilled, not in the least because they need to be able to manipulate a six ship fleet effectively. Which direction is that trend actually going? Thus we have a chicken and egg problem, one I don't believe math can solve. (Someone better at stats than me is welcome to contradict that.)

This has some definite truth to it; we can’t do a perfect estimation of MSU’s skill ceiling due to its skill floor. I’d really like to see a “for the stats” tournament (Vassal would work fine,) where evenly sized groups of list archetypes and activations are assigned to a large group of players across all skill levels. It would be a fun twist, and it would provide some useful, balanced data.

3 hours ago, The Jabbawookie said:

This has some definite truth to it; we can’t do a perfect estimation of MSU’s skill ceiling due to its skill floor. I’d really like to see a “for the stats” tournament (Vassal would work fine,) where evenly sized groups of list archetypes and activations are assigned to a large group of players across all skill levels. It would be a fun twist, and it would provide some useful, balanced data.

This may help

Its a 225pt tournament, but the key is that no one played with their own fleet, so skill and activation count are seperated a fair bit.

Here it is for the Salon des Perdants tournament.

This graph is the score of a fleet as a function of the number of activations. For this, the score is mostly divorced from player skill because this is the scores of the fleets, not the players, and each fleet was played by three different players. A higher score is better.

The modal number of activations was 2, with nine 2-ship fleets. It appears that 3 was optimal number of activations, with 40% of 3-ship fleets finishing in the top 4.

This seems to corroborate the earlier findings. It would certainly be nice to have similar data for 400-point fleets instead of 225-point fleets, though.

8 hours ago, stonestokes said:

It would certainly be nice to have similar data for 400-point fleets instead of 225-point fleets, though.

@CaribbeanNinja

Did someone say vassal tourney?