I participated in the UBC Election Stock Market again this year. This time I put $50 into the market. My strategy was to follow the predictions given by DemocraticSPACE for the seats market and the popular vote market.

By election day, I had expected to make about $5.69, but now that the preliminary results are in, it looks like my profit will be about $5.26. This is an improvement over my previous 74.4¢ loss.

I didn’t have the information or I didn’t know how to compute the optimal Kelly number of shares to purchase. At first, I simply bought and sold a fixed number of shares surrounding the target values given by DemocraticSPACE. Eventually I settled on a strategy of maintaining a constant proportion portfolio. This means that I kept the fraction of the market value of my shares for each party in proportion to the target values given by DemocraticSPACE. This fraction is calculated after hypothetically transferring all cash into bundles of shares. This strategy was more conservative that my initial strategy, but I felt it gave similar directions to what I would expect the Kelly criterion to give.

I found it a bit tricky to determine the market value of my shares that is needed for this strategy. The first step was to compute the virtual bid and ask prices, because sometimes is it cheaper to short all other parties rather than buy a given party. I used the virtual bid prices to compute a lower bound on the value of my shares, and I used the the ask prices to compute an upper bound on the value of my shares. To compute the number of shares I want to have at a given price, I compute the lower and upper bounds of the value of my portfolio (with the value of the party in question set to the price in question). This is used to compute lower and upper bounds on the number of shares I desire at that price (then rounded up or down). I then simply searched for a bid price such that I would buy at that price and an ask price such that I would want to sell at that price.

These lower and upper bounds that are computed from virtual bid/ask prices don’t necessarily add up to 1. An improvement could be made to do enforce this. There would be a prices such that it is not possible to add up to 1; these prices correspond to arbitrage opportunities.

Most of my $5.26 profit comes from market making, and most of that was made the first few days of the market when I was taking higher risks because I hadn’t yet settled on the constant proportion strategy.

LISPOP did better than DemocraticSPACE this year. Next time I will use LISPOP, not because I am chasing the last winner, but because I’m keeping a tally with LISPOP being my personal tie-breaker.