Last week, Citadel came to my school to host a fun game.

The game

Setup
There were six zones and six teams. Each team had 10 food trucks, which they could distribute amongst the six zones however they chose. They could put all 10 food trucks in one zone, or do 2-2-2 and so on, as long as it was 10 food trucks spread across the six zones.
Goal
Maximize the customers you get per food truck.
Public forecast
Before each round, the predicted customer forecast for each zone was revealed, plus or minus 300.
Private slip
At the start of each round, each team got a private slip with a forecast for two random zones that gave a customer estimate with better accuracy: plus or minus 100 instead of plus or minus 300. So each team technically had some private information.
Moves
After all teams distributed their trucks, the number of trucks in each zone was revealed, and each team had the option to move two of their trucks to different zones. Then the final number was revealed.
Rounds
There were three rounds of this.

Ultimately, my team won, which I think can definitely be attributed partially to luck. But I also think there were a few things we did right.

1. We barely moved

I think our biggest alpha, was that out of the six possible times we could have moved trucks (two after each of the three rounds), we only moved one truck total after seeing the numbers.

Our reasoning was that while optionality is good in an uncertain situation, having only limited optionality meant that after teams saw the trucks distributed within the zones, most teams would likely make similar moves: moving trucks from an area with lots of trucks to an area with fewer. We were banking on the fact that those moves would be mimicked, that teams would essentially be copying each other and all congregate together. Thus, not doing anything would be the advantage, because then you're not congregating or going with the crowd.

2. People gravitate towards the extremes

Another takeaway, and something we observed, was that people gravitate towards the extremes. If they saw a zone had a low predicted customer forecast, they would put proportionally more trucks in that zone than they should have. I think this was because they wanted to capitalize on doing something different from the market. They might assume fewer people would go there, so they would be the ones to take advantage. On the other extreme, people would almost avoid the places with a high predicted customer forecast, assuming lots of people would aggregate there.

But after the trucks were revealed each round, the most common moves were from the low forecast zones with a lot of trucks to the high end, and vice versa. The zones in the medium range of predicted customers would be almost neglected, and that's exactly where a lot of our advantage came from. The middle range would be around 1,000 to 1,500, while the zones with the most predicted would show around 1,800 and the least would be 400 or 500. So I noticed that people typically gravitate towards the extremes, and make those moves when they have limited optionality to change.

3. People are optimistic about the forecast

Lastly, I think psychologically people are optimistic about the predicted forecast. Often our biggest advantage came when the private information we received was for a lower zone. If a zone publicly displayed 500 customers and our slip showed 300, we had a better estimate that was lowering it. That was a greater advantage not just because it's a bigger percentage of the customer prediction, but also because people are optimistic about the estimate. When they see the predicted forecast, they often treat it as the floor, or assume the upper end of the predicted range. So knowing it's under, we could essentially bet on and exploit the fact that people are going to overestimate, since we know it's lower.

Jason Fang, 2026