This week will focus on the initial development of the 2nd nightmare scenario. For this dev diary, we will begin work on a different market-condition with a fundamentally different approach. 

As outlined when setting up the 1st nightmare scenario (Dev Diary 5), the first step for designing a scenario is determining the expected change in stock prices. In this situation, a slightly different representation will be considered:

current stock price * (x - (stock dividend))

x is approximately the overall expected return. During the “run-in period” prior to the first turn, x will be randomly selected from a uniform distribution between -0.05 and 0.1. After the first turn, x will be determined from a sin function (returning expected overall returns between -25% and 25%). The sin function will be shifted based on the relative characteristics of the stock price prior to turn 1. Specifically, the base function without a shift is below.

This sin function ensures that stocks will experience higher and lower expected returns over the course of the game, requiring an active trading strategy to ensure high returns. In order to create differences in the expected return across stocks, we can add a shift to the equation:

0.25 * sin(pi/5 * (turn - shift))

The definition of shift, as described below, ensures that some stocks will have higher overall returns based on information available to each player:

shift = [(stock price at turn -10) - (stock price at turn 0)] / 5 + [max(stock price between turns -10 and 0) - min(stock price between turns -10 and 0)] / 5

Turns -10 and 0 capture the “run-in period” for each game, which means that the expected returns on each turn during the game are defined on the 1st turn. Overall, this underlying truth aims to encourage relatively active trading based on where the stock may be on the sin curve. If a stock is at the bottom of the sin curve, a player should buy it. Conversely, if a stock is at the top of the sin curve, a player should sell it.

This example game achieves that goal. It feels like trying to buy “at the bottom” and sell “at the top”, as illustrated by the large looping curve on THC. However, players are likely to have difficulty knowing whether they are achieving that goal without understanding the shift definition in the underlying truth.

Thanks for spending time on this relatively brief diary. Next time we will look into developing the market-specific AI for this market condition.