I’ve been watching the ‘charts’ (ranks) of our current hit game ‘Fab Tattoo Artist’ on the iOs, Kindle & Google Play mobile stores very carefully.
These charts of the game on iTunes (free app), Kindle (paid app) and on Google Play (free app) for the period of the past 2 weeks are produced below:
My questions (to those who understand charts) or anybody else:
– Is there a ‘trend’ and ‘direction’ that can be spotted (with laser precision) about future game ranks using these charts of historic ‘game ranks’ ?
How can the immediate past predict the immediate and long future?
– What do stocks and game ranks have in common?
Stocks are measured in their prices and games are measured in their ranks (and downloads).
Given the mammoth business building up in mobile games, isn’t there a business model emerging in ‘game charting’?
– Could ‘rank spotting’ become a science that enables companies like games2win to spot outliers and potential mobile game hits in the making and bid to buy them?
– What ‘currency’ can ‘trade’ using these charts, like stock index trading?
Can game ranks become games in themselves? Games of ‘rank wagers’? 🙂
– Can the download velocity of games be scientifically predicted using rank charts?
Example of a dashboard that could say – “Fab Tattoo Artist = Game Rank 7 on Google Play Tablet = 1185 downloads a day”.
This can be very useful for all the players in the mobile games business who keep ‘guesstimating’ the game downloads of their competitors?!
I would love to hear ideas, suggestions and even open up our game data to analysts who have concepts to model mobile game rank and downloads using raw data!!
****
Puneet Singh
1) Yes, immediate past ranking data can predict the future ranks.
How this works: You basically shortlist or use the available parameters for ranking the games. You provide future estimated rankings and trade off with the actual past data and come to an average. You can catch points of expected ‘over-performance’ or ‘under-performance’.
Plus, past data also provides extended data like was there a spike in the ranking due to a new upgrade provided? Was it seen on a weekend? etc.
2) Apart from helping us in prediction, the one thing common between stocks and game ranks is the probability of ‘not-happening’ or ‘one time unexpected happening’ which can result in skewed predictions, both optimistically and pessimistically. Its a major catch point as it can help companies plan their launch/ marketing activities better.
3) Download velocity can be predicted from the game ranking charts but it is further done with precision when both the data are used simultaneously. It is a very strong datahouse that can be used.
As game ranking is a bit more under-estimated area to work on with predictive analytics, I believe it hasn’t got that much attention but given more and more companies are now engaging with customers via mobile applications, this is going to be an interesting area.
Hope this answers your questions to an extent.
Ravi T
Here is my 2 cents from my Stats courses, and personal working experiences.
1) Stock Market technical analysis is typically a Linear Regression (depends on time and past trend). Most of the technical analysis is to uncover the resistance levels, or understand the demand and supply. Stocks could be modeled in multi-variate manner, like including fundamental factors. Example – Say if you analyze a Movie Stock (Marvel), the opening box office weekend of SpiderMan movie can be used as other independent variable to get to opening price on Monday.
2) Game Ranking seems like a classic case for Multi-variate regression. Identify the independent variables this depends on – Game downloads yesterday, Game Uninstalls yesterday, Review Score in Marketplace, # of Competitors, and then you can run a multi-variate regression to identify how the ranking would be a day or two from now. It will give you a confidence level on how accurate your prediction is depending on how tight the correlations are, as well as how independent the variables are.
The question is – Can you identify Breakthrough games, using this method? I believe unlikely. The breakthrough patterns follow different models – Diffussion Models of adoption, but the error levels are high in the model. In the past, I have used such model to say predict Apple’s iPod sales in a quarter.
Summary – It can be done, but not sure if you spot winners accurately.
Prashant Chauhan
I would like to make a point here , Ranking system of game should not be based on number of (downloads Only) but based on more complex data as user reviews,votes,number of downloads,time spent by each user gaming, popularity survey, life cycle analysis from market entry to end phase,expert advise,media coverage ,online and offline analysis etc.
Also we dont need ranks from 1 to 10000 it should be a rating system based on scale of 1 to 10 as IMDB ratings derived from very complex analysis.
the aim should not be to flaunt number of dowands and predict ranks etc insted give users a Rating system which can be trusted by them (Similar to IMDB RATING ) and helps them clearly diffrenciate a medicore game from the best games.
Sridhar V
Interesting comparison. I think there is a lot of similarity between stock trading and gaming in terms of activity and swiftness.
But investing in stocks will not be comparable to this in my view because in the case of investing its just a longer-term activity where one focuses on the Value more than the Price and wishes to stay invested to make longer term gains.
Stock Index Trading: This was mentioned above. My suggestions is that one can create an index which is shown in teh chart showing “US Games”, “Games Casual US”, etc. Similarly we can have an index for different segments like Arcade, Board games, Puzzles, Racing, etc.
In future there could be some way for people to invest or trade in game indices. The invested amount might go towards building the game infra (websites, tournaments, etc) and there could be a winning prize amount for the investors (if they win the match/contests). The fall out of this could also lead to something like IPL franchise (game franchise) based on location, community, groups, etc.and might become like a regular yearly Mega Event.
Most of what I said are just thoughts…..but implementing it will be a Challenge. However, for people who are already in to this industry, it should be relatively easier to see what works and take it forward.