Round 2 was extremely time-constrained. For the algorithmic portion, my approach was essentially as described in my previous post. The manual trading component, however, proved more tricky.
The manual trading mechanism worked as follows:
⦁ Each participant submits a price between 9000 and 11000
⦁ If your price is below the median of all submissions, you get to trade
⦁ You sell at your submitted price and buy back at the average price of all qualifying traders (those below median)
⦁ The trade quantity is fixed at 300
Theoretically, the optimal strategy is to submit a price just below the median to maximize profits. The profit distribution follows this pattern.
This resembles the classic game theory problem of “guess 2/3 of the average,” where a Nash equilibrium exists but is nearly impossible to achieve in initial attempts. Research shows that in real-world experiments, results typically converge to about two levels of reasoning:
1. First-level: Uniform distribution → median at 10000, buyback at 9500
Profitable range: 9500-10000
2. Second-level: Assuming others think like first-level → median at 9750, buyback at 9675
Thus, submitting between 9675-9750 seems reasonable
However, Discord discussions revealed most participants planned to submit 10000+, as:
⦁ The Nash equilibrium is everyone submitting 11000
⦁ No individual has incentive to submit lower prices (would lose money)
This makes the actual outcome harder to predict
Round 2 will likely reshuffle rankings significantly because:
1. Algorithmic trading:
⦁ Some may achieve huge profits by simply betting on price movements
⦁ Conservative pair trading might cap at ~50k
2. Manual trading:
⦁ High luck factor – guessing right could yield 250 profit × 300 = 75k
⦁ Typical algorithmic profits per round are only ~30k
