Fantasy Grounds Game Stats for 2017: D&D 5E Up By 6%, Pathfinder Holds Steady!

It's time for the statistics which tell us what games are being played on Fantasy Grounds, the virtual tabletop. This always gives a great sampling of which games are popular at the moment. As usual, D&D 5E leads by a large margin, followed by Pathfinder, D&D 3.5, then Savage Worlds. "Core RPG" in the data below refers to Fantasy Grounds' 'default' RPG module, so can reasonably be viewed as the "other" option. These stats represent the whole of 2017.

Fantasy Grounds' Doug Davison notes that there is a bit of a dip around December, but that same thing happened in the previous two years.

See 2016's figures here. Over the year, D&D 5E increased again, this time from 58% in 2016 to 64% in 2017. Pathfinder held steady at 12%, and Savage Worlds at 6%, while D&D 3.5 dropped from 6% to 4%.​

Ruleset	Total Games
5E	395,748
PFRPG	74,551
SavageWorlds	37,693
3.5E	21,992
CoreRPG	12,704
Call Of Cthulhu	10,928
4E	9,948
MoreCore	8,975
Star Wars EotE	7,642
GURPS	6,176
AD&D	4,028
DSA	3,657
Castles and Crusades	3,638
WFRP	3,261
RolemasterClassic	3,208
WH40K	2,593
Numenera	2,507
Fate Core	2,300
VtM	1,852
Basic Roleplaying	1,482
Shadowrun	1,044
Mutants and Masterminds	1,006
World of Darkness	997
DCCRPG	783
d20 Modern	768
Grand Total	619,481

 

[damned]

That’s an interesting question. When we see 33K 5E games in sequential months, is that 33K new campaigns started each month, or are the same campaigns being counted anew each month? The totals add them together, suggesting it’s the former.

They are games running each month. Not new games starting.

 

[smiteworks]

A game that runs weekly would get counted as 4 games a month. Our available statistics don't allow for us to count distinct campaigns. If a campaign skips a month, it wouldn't get counted for that month. That is why you see drops around Christmas, for instance.

 

[Nylanfs]

Hmmm, I wonder if you could get the name of the campaign reported also, then you could run it through a check for unique names in x month compared to y month.

 

[JesterOC]

Interesting that From 2016 to 2017 about 172k more games being played in total. But at the same time 5e itself had 140k more plays. Which means from a certain point of view 81% of the growth comes from 5e. Pathfinder did great to hold its share with the influx.

 

[Morrus]

They are games running each month. Not new games starting.

So its about 33K 5E games, not 395K as the lower table says?

 

[Achan hiArusa]

Math errors

I ran the Zipf distribution on these data. The Zipf distribution is easiest to visualize: One plots the log-counts (taken from the data) on the log-ranks (here 1 to 25). A Zipf distribution says that the frequency of occurrence is proportional to the inverse of rank. A Zipf distribution fits if this plot is linear and negatively sloping. It's not though it is approximately so.

The Yule-Simon distribution would probably fit better but I don't have a program to estimate it (looking around there's one in R... I'll have to check). The usual explanation for something having a Yule-Simon distribution is "preferential attachment." Anyone who knows about making a game knows all about that!

View attachment 92626

I don't know what those numbers are supposed to be. On the X-axis Antilog 2 = 10^2 =100 and there are not quite 100 game systems or let alone Antilog 3 = 10^3 = 1000 game systems there. On the Y-axis Anti-log 14 = 10^14 = 10,000,000,000,000 or 10 trillion games played and since it tops at around 395,748 games played which the log of is 5.6 who ever made this program needs to correct those. The math is bad.

Given the large numbers a simple Arrhenius plot of log (games) vs. Rank would suffice.

I am curious what this type of plot is called I called it a Lalanne plot after Leon Lalanne the mathematician who first used it in my Astronomy thesis, but I never found out the actual name.

 

[Jay Verkuilen]

I don't know what those numbers are supposed to be. <...> The math is bad.

The math is fine.

Those are natural logs, not log10. Given that switching from log bases is just a scale change it does not affect the linearity of the plot.

The reason I plotted log-log is because that's what is connected to the Zipf distribution and its generalizations. The Zipf is linear in log-log space, and that plot is close to it, though not perfectly so. I'm not familiar with the Ahrrenius plot (I assume it's just a semi-log plot) but it won't be linear and thus isn't useful in this case.

As to why I used natural log, just habit I guess. In statistics, natural log is the convention but maybe it isn't in other fields.

 

[damned]

So its about 33K 5E games, not 395K as the lower table says?

Hi Morrus some will be one shots and Adventurers League. Some will be 2-4 sessions and some will be campaigns lasting many months. There is no way to extract the number of campaigns from the number of games with the reporting stats that are used.

 

[Morrus]

Hi Morrus some will be one shots and Adventurers League. Some will be 2-4 sessions and some will be campaigns lasting many months. There is no way to extract the number of campaigns from the number of games with the reporting stats that are used.

They're all numbers too big to really imagine, whichever way you cut it! Our hobby is definitely still growing!

 

[Fandabidozi]

I'd totally be up for playing 3.5, specifically Eberron, but building a character is way too daunting.