Do we go to time?

There has been a lot of talk lately about “pace of play” – and as a consequence of that, stats have been circulated showing that more games go to time in X-wing 2nd edition vs 1st edition.

In this post, I will explain how we arrived at those numbers, as well as try to dig a bit deeper to see if there are any good indicators as to what causes more games to go to time. I will try not to speculate and simply state what the data shows. The post is written “in real time”, meaning that I essentially analyze as I go. So you can see my thought-process shine through to some degree.

Thanks to Chris Allen for working with me on this, and to Sozin for providing the 1st edition data needed.

Feedback, in the civilized form, is very welcome – please leave a comment below.

Conclusions

As a sort of TLDR, the summary of the conclusions are:

• About twice as many games go to time in 2nd edition compared to the Post-wave 11 1st edition.
• Less games go to time in larger events. This is a smaller effect, and we see it in both 1st and 2nd edition.
• Games involving the “launch factions” (Rebel, Empire, Scum) are less likely to go to time, which in turn means that games in Extended events historically have been much less likely to go to time, as these events have seen a high representation of these three factions. We don’t know if this is due to increased familarity with the pilots, cards and effects used by those factions, or just because they are inheriently faster to play. Note, however, that the Separatist faction specifically is very close to the “launch factions” in this regard, well within statistical uncertaincy. Also note that it seems unlikely that the “increased familarity” hypothesis will hold, as Resistance and First Order have the lowest game completion ratio of any faction.
• The total number of ships on the table have a significant effect on the completion-rate of the game, and particulary with more that 8 total ships on the table we can see the completion-rate of games fall quite a bit.
• Despite looking at the data in many different ways, the general tendency that 2nd edition games are significantly less like to finish in time compared to 1st edition is stronger than any other effect.

If you are interested in the details, please read on 🙂

The data

For 1st edition I am using a complete database dump of Star Wars X-Wing List Juggler provided to me by Sozin.

For 2nd edition the data is extracted from Listfortress.

Note that as this requires more detailed “matchup data” as opposed to only “tournament data”, not every tournament is included – specifically for 2nd edition, we only started getting matchup data in March 2019.

I’ve defined “Finished at time” as “One player got 200 points” (100 in 1st edition). Technically some of these might have gone “to time”, and completed in the current round, but I still count these games as “Finished at time”.

In order to ensure the data is valid, some filters have been applied:

• Only swiss games are considered. Games in the cut are often recorded as 200-0 in the records, rather than the actual score.
• Bye-games are discarded
• “Final Salvo” and “Draw” games are excluded as it is non-consistent how scoring has been applied from them. These were only explicitely tracked in Listjuggler (1st edition) – and the effect is minimal in either case

Baseline – 1st edition

Only basic analysis is done on 1st edition data – the point is mostly to serve as a point of comparison to the numbers we see in 2nd edition.

1st Edition Swiss games No byes, no salvos	#games	#games finished at time	%games finished at time	%games not finished at time
All	123247	95706	77.7%	22.3%
After August 2017	47200	36772	77.9%	22.1%
After August 2017, 60+ player tournament	19615	15873	80.9%	19.1%

So in conclusion 77.7% af all recorded 1st edition games finished with one player having lost all his ships, while 22.3% of the games was decided on who scored the most points. If we look only at the latter part of 1st edition, the completion-rate increases slightly. If we furthermore narrow down to larger tournaments, where we can reasonably expect more familarity from the players, and also increased focus on the game, the completion rate increases significantly to 80.9%.

Baseline – 2nd edition

In order to compare directly to the 1st edition numbers, here is the same data for 2nd edition. As previously mentioned note that “all” means “after march 2019” as data on individual games were not being tracked before that. Also note that several major tournaments does provide data on individual games.

2nd Edition Swiss games No byes	#games	#games finished at time	%games finished at time	%games not finished at time
All	10846	5911	54.5%	45.5%
All, 60+ player tournament	4261	2437	57.2%	42.8%

So in conclusion 54.5% of all recordded 2nd edition games finished with one player having lost all his ships, while 45.5% did not. If we only look at larger tournaments, we see some of the same effect that we saw in the first edition data. (Increased % of games finish at time).

These numbers support the statement that about twice as many games “go to time” in 2nd edition compared to 1st edition.

2nd edition, split by wave

While the 2nd edition data is limited, I’ve split the data in two:

• Wave 2b games; i.e. games recorded after the points change in january 2019 and before the official launch of Wave 3
• Wave 3 games; i.e. games recorded after the official launch of Wave 3 and before the official launch of Wave 4.

2nd Edition Swiss games No byes	#games	#games finished at time	%games finished at time	%games not finished at time
Wave 2b	1359	800	58.9%	41.1%
Wave 2b, 60+ player tournament	592	387	65.4%	34.6%
Wave 3	8416	4511	53.6%	46.4%
Wave 3, 60+ player tournament	3669	2050	55.9%	44.1%

Here we see an obvious outlier – Wave 2b tournaments with 60+ players have a much higher completion rate. The data is, however, very limited in number – 592 games are not a lot. They come from just three early hyperspace trials. Digging down into those we find:

2nd Edition Swiss games Wave 2b, 60+ players No byes	#games	#games finished at time	%games finished at time	%games not finished at time
Element Games (UK) Hyperspace Trial	236	172	72.9%	27.1%
Game On Bathurst (AU) Hyperspace Trial	168	108	64.3%	35.7%
Games U (US) Hyperspace Trial	168	107	63.7%	36.3%

Hard to conclude anything from this directly. All three tournaments are well above the average in completion ratio with the Element Games Trial being way above. A hypothesis could be that the Hyperspace format itself lends to more games completed. Another could be that these three trials all took place in the back half of Wave 2 – with most of the used content having been out for a long time, and hence players were more familiar with it. A third perspective could be that the completion ratio has dropped as a direct result of content introduced in Wave 3 that slows down play significantly. These hypotheses will be explored below.

2nd edition, split by format

Hypothesis: Hyperspace games are inheriently more likely to finish at time.

Note: In order to counter any specific issues with the small dataset for Wave 2b in general, we will be using wave 3 data only here. If there is a “hyperspace effect” if should be visible in wave 3 data as well. If we were to use Wave 2b data the three tournaments above would be such a large part of the data, that the hypothesis would be proven by default.

2nd Edition Swiss games Wave 3 No byes	#games	#games finished at time	%games finished at time	%games not finished at time
Wave 3 baseline	8416	4511	53.6%	46.4%
Wave 3 baseline, 60+ players	3669	2050	55.9%	44.1%
Hyperspace	5018	2604	51.9%	48.1%
Hyperspace, 32+ players	3874	2066	53.3%	46.7%
Hyperspace, 60+ players	1374	722	52.5%	47.5%
Extended	3398	1907	56.1%	43.9%
Extended, 60+ players	2295	1328	57.9%	42.1%

Looking at the data above, it quite clearly shows that there is no evidence for suggesting that Hyperspace has a higher completion ratio than Extended, quite to the contrary. It does, however, potentially strengthen the other two hypotheses; as Extended in wave 3 was dominated by the three original factions, i.e. most players played one of the three original factions, while Resistance, First Order, Republic and CIS were played a lot more in Hyperspace in Wave 3. So perhaps something available in these four factions are slowing down the games?

2nd edition, split by faction

Hypothesis: Some factions are more likely to complete their games than others

2nd Edition Swiss games Wave 3, 60+ No byes	#games	#games finished at time	%games finished at time	%games not finished at time
Wave 3, 60+ Baseline	8416	4511	53.6%	46.4%
Rebel Alliance	1908	1103	57.8%	42.2%
Galactic Empire	1676	938	56.0%	44.4%
Scum and Villainy	852	520	61.0%	39.0%
Resistance	705	347	49.2%	50.8%
First Order	436	208	47.7%	52.3%
Galactic Republic	542	279	51.5%	48.5%
Separatist Alliance	362	200	55.2%	44.8%

These numbers seem to support that the “original three factions” overall finish more games on time. While the direction of this causal relationsship is not proveable – it is reasonable to conclude that Extended tournaments have more completed games than Hyperspace because most of the participants play those factions, than to argue the reverse. Of special note, however, is that Seperatists are very close in their completion ratio to the three “original factions”, and they are the only faction of the four “new” factions that are above the Wave 3 average.

We can also, with some certaincy, conclude that the “familarity” hypothesis does not hold, at least not for Resistance and First Order. They have been out since December 2018, and despite that they have the lowest completion ratio of any factions.

Let’s explore some other facets of the data – starting with the number of ships on the table

2nd edition, by total shipcount

Hypothesis: More ships on the table decrease the chances the games complete on time

2nd Edition Swiss games Wave 3, 60+ No byes	#games	#games finished at time	%games finished at time	%games not finished at time
Wave 3, 60+ Baseline	8416	4511	53.6%	46.4%
4-5 ships total	162	112	69.1%	38.9%
6 ships total	652	430	66.0%	34.0%
7 ships total	1106	657	59.4%	40.6%
8 ships total	936	478	51.1%	48.9%
9 ships total	398	183	46.0%	54.0%
10 ships total	249	112	45.0%	55.0%
11 ships total	84	40	47.6%	52.4%
12-14 ships total	47	21	44.7%	55.3%

From this we can see a clear tendency that number of ships on the table matter a lot. There is a large effect here. An obvious remark could be whether this effect explains the faction-difference shown before. To examine this, let us look at the average shipcount of lists split by faction:

Faction	Avg. Shipcount for Wave 3	%games finished at time
Wave 3, 60+ Baseline	3.8	53.6%
Rebel Alliance	3.7	57.8%
Galactic Empire	4.1	56.0%
Scum & Villainy	3.3	61.0%
Resistance	3.8	49.2%
First Order	3.3	47.7%
Galactic Republic	4.1	51.5%
Separatist Alliance	5.0	55.2%

There seems to be no strong relation here. First Order and Scum on average have the lowest shipcount (3.3) – but while Scum is also the faction most likely to complete games in time, First Order is the least like to do so. Separatists have the highest shipcount, but actually complete more games at time than the average.

This does, however, lead to two new hypothesis:

• The list size of the winning player matters more than the total number of ships, as the winning player can choose to play slower
• The initiative of the winning player matters quite a lot as evidenced by the low First Order completion rate (“Aces drag the game out”)

2nd Edition, by shipcount of winning player

2nd Edition Swiss games Wave 3, 60+ No byes	#games	#games finished at time	%games finished at time	%games not finished at time
Wave 3, 60+ Baseline	8416	4511	53.6%	46.4%
2 ships	198	103	52.0%	48.0%
3 ships	1213	704	58.0%	42.0%
4 ships	1600	929	58.1%	41.9%
5 ships	357	177	49.6%	50.4%
6 ships	202	91	45.0%	55.0%
7-8 ships	64	29	45.3%	54.7%

From this we can conclude that 3 and 4 ship lists are significantly more likely to finish at time if they win. 2 ship lists are slightly more likely to “win at time”, which might confirm the hypothesis that there is a certain element of aces playing it safe close ti time. In order to explore this further, I had originally intended to look at the average (and/or maximum) initiative of lists and compare their completion ratio – but let us instead look at specific pilots and upgrades that are likely candidates for going to time.

2nd edition, by content of winning list (selected)

2nd Edition Swiss games Wave 3, 60+ No byes	#games	#games finished at time	%games finished at time	%games not finished at time
Wave 3, 60+ Baseline	8416	4511	53.6%	46.4%
Upgrade: R2 Astromech or R2-D2	391	214	54.7%	45.3%
Upgrade: R2 Astromech or R2-D2 – Rebel	229	134	58.5%	41.5%
Upgrade: R2 Astromech – Republic	160	80	50.0%	50.0%
Pilot: Soontir Fel	294	163	55.4%	44.6%
Pilot: Kylo Ren	123	58	47.2%	52.8%
Pilot: Lt. Tavson	126	48	38.1%	61.9%
Pilot: Anakin Skywalker (Aethersprite)	164	76	46.3%	53.7%
Pilot: Poe Dameron	112	52	46.4%	53.6%
Pilot: L’ulo L’ampar	279	135	48.4%	51.6%
List: Quad Phantom	174	110	63.2%	36.8%
List: Rebel Beef	611	387	63.3%	36.7%
List: Iden Howlrunner swarm (6 ships)	104	48	46.1%	53.9%

From the table above it is clear that some pilots (Kylo, Tavson, Anakin, Poe, L’ulo) have an effect in not having the game go to time. Surprisingly Rebel Regen is higher than average in completion ratio, but they are still above the average rate for rebel lists (57.8%)

2nd Edition, by type of ship

Hypothesis: Repositioning is a bigger part of the game in 2nd edition, so moving ships and deciding on actions take longer

Lets look at ships. Note that the numbers no longer mean “number of games” – but rather “number of ship-games” – meaning that for instance one game with 6 TIE-fighters vs 3 Aethersprites will count as 6 “TIE fighter games” and 3 “Aethersprite games”. This also means that we need to calculate a new baseline-number, as included in the first row. Since this table is substantially larger, I will reduce the detail-level in the data, showing only the ship-name and the percentage of games that ended with one player losing all ships. I have also limited the data to ships with at least 200 “ship-games”. This limits the list from 59 ships to 32. If anyone wants the full table of data to check my calculations, please reach out to me.

Ship	%games finished at time
Baseline	53,9%
A/SF-01 B-wing	55,4%
ARC-170 Starfighter	55,2%
Alpha-class Star Wing	53,0%
BTL-A4 Y-wing	54,2%
BTL-S8 K-wing	53,6%
Delta-7 Aethersprite	51,3%
Fang Fighter	65,6%
Firespray-class Patrol Craft	56,7%
Lambda-class T-4a Shuttle	51,9%
Modified YT-1300 Light Freighter	60,2%
RZ-1 A-wing	51,4%
RZ-2 A-wing	43,2%
Scurrg H-6 Bomber	60,5%
Sith Infiltrator	60,1%
StarViper-class Attack Platform	53,6%
T-65 X-Wing	59,9%
T-70 X-wing	51,0%
TIE Advanced x1	57,9%
TIE Interceptor	58,9%
TIE Reaper	50,5%
TIE/fo Fighter	48,0%
TIE/ln Fighter	46,2%
TIE/ph Phantom	57,1%
TIE/sa Bomber	61,8%
TIE/sf Fighter	50,2%
TIE/sk Striker	57,6%
TIE/vn Silencer	48,9%
UT-60D U-wing	54,9%
Upsilon-class command shuttle	40,9%
V-19 Torrent Starfighter	42,1%
Vulture-Class Droid Fighter	48,6%
Z-95-AF4 Headhunter	42,8%

While the numbers above are not directly comparable to the other data, it can be compared to the baseline, and to the other ships in the table.

It is quite clear that the primary driver for the differences within 2nd edition is the factions. If look at some of the ships with linked repositioning actions, there is no strong evidence that this, by itself, drives game-completion-rates down. The Fang Fighter, for instance, has a quite flexible dial, and the ability to both boost and barrel roll – but it is also the ship with the highest game completion-rate of any. The TIE Interceptor likewise is above average. The Aethersprite and and the TIE Silencer also have double repositioning and quite low game-completion-rates – but that is consistent with the general completion-rates seen in the faction-numbers for Republic and First Order.

Conclusion

So what can we conclude from all this?

• About twice as many games go to time in 2nd edition compared to the Post-wave 11 1st edition.
• Less games go to time in larger events. This is a smaller effect, and we see it in both 1st and 2nd edition.
• Games involving the “launch factions” (Rebel, Empire, Scum) are less likely to go to time, which in turn means that games in Extended events historically have been much less likely to go to time, as these events have seen a high representation of these three factions. We don’t know if this is due to increased familarity with the pilots, cards and effects used by those factions, or just because they are inheriently faster to play. Note, however, that the Separatist faction specifically is very close to the “launch factions” in this regard, well within statistical uncertaincy.
• However, it seems unlikely that the “increased familarity” hypothesis will hold, as Resistance and First Order have the lowest game completion ratio of any faction.
• The total number of ships on the table have a significant effect on the completion-rate of the game, and particulary with more that 8 total ships on the table we can see the completion-rate of games fall quite a bit.
• Despite looking at the data in many different ways, the general tendency that 2nd edition games are significantly less like to finish in time compared to 1st edition is stronger than any other effect.

Let me know in the comments if you have a theory on this that would like to explore – I might just do that 🙂