Market Modeling

Market Modeling is Hushcrasher's research on the economics of the video game market to understand how it works (how players choose between games, how producers set their prices, and how the two interact) and answer business questions with simulations (what a game would sell at another price, what a discount is worth, how an acquisition would change prices and revenues, or what a new business model, such as a subscription, would do to the whole market). It is in development: no results are published yet, but its first building blocks already inform our pricing advisory.

Currently, studios, publishers, investors and platforms take market decisions (what price to set, when to discount, whether to acquire a competitor or launch a subscription) with tools that have known limits:

  • Competitor averages. Pricing advice usually takes the average price of similar games. It describes what others charge, not what players would pay, and says nothing about how demand responds when the price changes.
  • Past outcomes. Benchmarks and ratios, such as wishlists to sales, describe what happened. They cannot say what would happen under a decision not yet taken: another price, a merger, a new business model.
  • Price and quality move together. Better games are often priced higher and also sell more, so comparing prices and sales across games understates how much players react to price.
  • Categories are not competition. Two games under the same tag may not compete for the same players, and games under different tags may.

We model the market as a whole, players and producers together, from market data and from the behavior of individual players: what they wishlist, what they buy and when. The aim is to measure how demand responds to price, separate the effect of price from that of quality, see which games actually compete, and simulate the effect of decisions not yet taken.

Methods

We develop a structural model of demand, that is, an explicit model of how players choose which game to buy and when, given current prices and the discounts they expect. Estimated on store data (prices, discounts, market shares, game characteristics) and on the behavior of individual players (what they buy, when, and what they play), such a model measures how sensitive players are to price and how far ahead they anticipate future prices. Because it describes the decision itself, it can simulate situations never observed, such as a merger or a subscription. This work is carried out with Anatole Cheysson, economist at ZEW, and advised by Felipe del Canto Monge, from Harvard University.

To complement this approach, we use causal machine learning to estimate the effect of a decision, such as a price change or a discount, directly from observed variation, and policy learning to derive from these effects the best decision for a given game. This approach relies on fewer assumptions about player behavior, but is limited to decisions close to those already observed in the data.

Applications

  • Measuring price elasticities and substitution: how sensitive players are to a game's price, and which games draw players from which.
  • Pricing: the revenue-maximizing price of a game, at launch and over its life, including the depth and timing of discounts.
  • Defining a market from substitution between games.
  • Simulating a merger or an acquisition: its effect on prices, on producers' revenue and on players.
  • Simulating platform-wide decisions, such as launching a subscription and setting its price, or changing the revenue share or discount rules: their effect on total revenue, on how it is split between producers and the platform, and on what players play and buy.

We license our datasets and model outputs for non-commercial academic research. Reach out →