Goals of the gameUltimately, the goal of the game is to beat the strategy of 'buy and hold'. The primary goal is to maximize one's returns, without tying up a lot of capital while doing it. So the goal of the game is to identify patterns that lead to increases in value of the stock, and be invested at those times.
You are evaluated on the average gain per day invested, and the corresponding annualized gain from that. This is calculated by taking the total gains made across all stocks played, and calculating an equivalent daily gain that, if achieved for every day invested, would produce the same result. This gain is then compounded daily over a year to calculate the equivalent annualized gain. However, if you are only invested for a few days worth of play. The 'annualized' gain will be so far extrapolated as not to be very meaningful.
Equivalent daily gain calculationsThe equivalent daily gain is calculated such that this gain achieved every day would give the same result. This is slightly different from just taking the gain and dividing it by the number of days invested. For example, if you had a stock quadruple in value over two days, this would give you 300% over 3 days. But if you divided the 300% by three, that would give you 100%, or a doubling every day, which would suggest an 8-fold increase after three days. The equivalent daily gain to quadruple over 3 days is actually a 58.75% gain per day.
It's hard to beat buy and holdThe original goal was to have the user beat buy and hold for the whole period played. However, playing the game, it turned out to be nearly impossible to beat buy and hold based on that criteria.
This because stocks, on average, go up a little over time. In real life, if you always have money in the market, chances are you will do better than having money in only occasionally. But if one does actively play the market, one can use the money while it's not invested in one stock to play other stocks. So I now evaluate the player based on gains achieved over the days that the money was actually invested in a stock, which gives the user benefit for actually having the money available to play other stocks with.
With the more favourable method of calculating gains only over days invested, doing completely random trades over the long term should cause you to perform roughly equivalent to buy and hold. Interestingly enough, on playing long enough, most player's performance ends up being very close to buy and hold.