Uploaded January 2018 | Updated September 2026, 3 weeks ago
An in-depth analysis of the factors impacting today’s retro video game prices. Combines economic theory with statistics to explore the impact of game supply, YouTube exposure, console exclusivity, and a variety of other topics.
Uses the Sega Dreamcast library as primary data source (see pinned comment for more on this).
Methods - 2:19
Overview of Game Prices - 3:36
Economics 101 w/Shenmue - 6:51
Supply Impact - 9:52
Exclusivity Impact - 13:38
YouTube Impact - 14:41
Genre Impact - 17:00
Perceived Quality Impact - 18:37
Licensing Impact - 20:16
Publisher Impact - 21:29
Conclusion - 23:12
Important notes / disclaimers:
Accurate video game sales numbers (for both consoles and individual games) are notoriously difficult to come across. As stated in the video, the original USA Dreamcast software sales for this analysis come from tinyurl.com/y9qw5v9n and are clearly unofficial. However, in comparing some of the numbers with other data sources, I do believe the figures are within an acceptable range of actuals. While some numbers vary by source, I also believe that the ratios between games hold steady.
All used Dreamcast eBay prices for this analysis were gathered from pricecharting.com in late October, 2017.
Regression Model Details:
R-squared for the OLS regression model is 0.64. This almost guarantees that I am missing independent variables that are influencing retro prices. However, nearly all regression models are missing some variables (but of course I would say that!).
Due to the nature of the dependent variable distribution, I used the log form of used game prices to more closely approximate a normal distribution. I also used the log of original Dreamcast game sales (a key independent variable) for the same reason.
All of the noted findings in the video are significant at the 0.05 level, with one exception. Licensing was slightly above the 0.05 level.
Regarding “overfitting” the regression model — I worried greatly that I had included too many variables for such a small sample (247) and therefore experimented with collapsing certain variables together (e.g. only including publishers that had more than 10 games and lumping others into the ‘other’ category). Through various versions of the model, I am satisfied that the findings are fairly consistent across models.
Any other questions about the regression model? Let me know!
A note on Capcom and the Role of the Publisher:
As discussed in the video, I am very confused over the findings related to Capcom. I ran the analysis a variety of ways, but the results always seemed to indicate that Capcom was a significant variable. Thinking that maybe the results were tainted by some publishers only having a small number of games, I ran the results comparing Capcom against every other publisher, instead of just Sega. The results were still similar — Capcom published games could be predicted to see much more than other publishers.
References and Thanks:
thedreamcastjunkyard.co.uk (background information)
The Ultimate Dreamcast Collectors Guide by Mike Phelan (Dreamcast Junkyard)
Special Audio Note:
I suffer from tinnitus which prevents me from dialing in the audio as well as I would like (thanks to my wife for helping with audio issues throughout the video). Please excuse any rough edges. I’m learning and adapting as best I can.
An in-depth analysis of the factors impacting today’s retro video game prices. Combines economic theory with statistics to explore the impact of game supply, YouTube exposure, console exclusivity, and a variety of other topics.
Uses the Sega Dreamcast library as primary data source (see pinned comment for more on this).
Methods - 2:19
Overview of Game Prices - 3:36
Economics 101 w/Shenmue - 6:51
Supply Impact - 9:52
Exclusivity Impact - 13:38
YouTube Impact - 14:41
Genre Impact - 17:00
Perceived Quality Impact - 18:37
Licensing Impact - 20:16
Publisher Impact - 21:29
Conclusion - 23:12
Important notes / disclaimers:
Accurate video game sales numbers (for both consoles and individual games) are notoriously difficult to come across. As stated in the video, the original USA Dreamcast software sales for this analysis come from tinyurl.com/y9qw5v9n and are clearly unofficial. However, in comparing some of the numbers with other data sources, I do believe the figures are within an acceptable range of actuals. While some numbers vary by source, I also believe that the ratios between games hold steady.
All used Dreamcast eBay prices for this analysis were gathered from pricecharting.com in late October, 2017.
Regression Model Details:
R-squared for the OLS regression model is 0.64. This almost guarantees that I am missing independent variables that are influencing retro prices. However, nearly all regression models are missing some variables (but of course I would say that!).
Due to the nature of the dependent variable distribution, I used the log form of used game prices to more closely approximate a normal distribution. I also used the log of original Dreamcast game sales (a key independent variable) for the same reason.
All of the noted findings in the video are significant at the 0.05 level, with one exception. Licensing was slightly above the 0.05 level.
Regarding “overfitting” the regression model — I worried greatly that I had included too many variables for such a small sample (247) and therefore experimented with collapsing certain variables together (e.g. only including publishers that had more than 10 games and lumping others into the ‘other’ category). Through various versions of the model, I am satisfied that the findings are fairly consistent across models.
Any other questions about the regression model? Let me know!
A note on Capcom and the Role of the Publisher:
As discussed in the video, I am very confused over the findings related to Capcom. I ran the analysis a variety of ways, but the results always seemed to indicate that Capcom was a significant variable. Thinking that maybe the results were tainted by some publishers only having a small number of games, I ran the results comparing Capcom against every other publisher, instead of just Sega. The results were still similar — Capcom published games could be predicted to see much more than other publishers.
References and Thanks:
thedreamcastjunkyard.co.uk (background information)
The Ultimate Dreamcast Collectors Guide by Mike Phelan (Dreamcast Junkyard)
Special Audio Note:
I suffer from tinnitus which prevents me from dialing in the audio as well as I would like (thanks to my wife for helping with audio issues throughout the video). Please excuse any rough edges. I’m learning and adapting as best I can.
