Part of our research goals include working to deepen our understanding of the dynamics of data markets under various conditions. In particular, it would be very useful to understand how a data market grows under various conditions. In this post, I’ll discuss a simple simulation I put together to model the effect of market growth under differing amounts of patron support. I’ve put together an ipython notebook with this model, but will talk through the basic concept here.
The core idea behind this model is that it’s useful to posit that as the size of a market’s reserve grows, the number of makers interested in joining the market increases. This likely isn’t an entirely accurate assumption (a rational maker would likely look at the average earnings other makers are making in that market, but coding this would require a more sophisticated agent model), but likely isn’t terrible since more funds in a market will result in more publicity which will draw in makers. We model this effect by the following response function
# This is the maximum number makers on the network MAX_MAKERS = 1000 # This is number of listings *per* maker N_LISTINGS_PER_MAKER = 100 def get_number_makers(reserve_size): return min(reserve_size // 10, MAX_MAKERS)
We then posit that the number of buyers who enter the market is proportional to the number of listings available on the market. We model this assumption with the following code:
# This is the maximum number buyers on the network MAX_BUYERS = 100 # Size of an individual data purchase in ETH PURCHASE_SIZE = 1000 * WHOLE # ETH def get_number_buyers(n_listings): return min(n_listings // 1000, MAX_BUYERS)
The purchase model here is pretty crude. We assume a thousand extra listings draws in a new buyer who’s willing to spend a 1000 ETH. This is something like $200,000 at today’s prices. This assumes that there can be up to 100 buyers, which would be a demand ceiling of $20 million for the market.
Under this crude model, what we see is that the reserve grows by the following curve:
This model intuitively suggests that while the market hasn’t hit saturation, adding more patron capital boosts growth. Once the market has hit saturation point, the growth levels off. In fact, the tail end growth in the reserve is probably purely an artifact as extra patron capital is added to the market. This simple simulation suggests to me that the true model of the reserve-patron curve is likely a sigmoidal curve
I’m not entirely confident in the simulation yet, since it uses a very crude model, but the sigmoidal shape seems intuitively reasonable. It might be a good proxy to use in more complex ecosystem simulations.