D.S.S.

Inspiration

We were inspired by the Basic Quoter and the workshop.

What it does

The Basic Quoter has been extended by a semantic analysis, hedging and capturing Dual Listing profits.

How we built it

Specifically, in the beginning of each iteration, we will check for Dual Listing profits. Afterwards, we will run a semantic analysis based on the news feed. This happens in two steps: First, we analyze, which company is affected by the news statement. Afterwards, we evaluate, whether the statement has a positive or negative influence. The weighted impact of the news statement (encoded into 1, 0, -1 for positive, neutral or negative and multiplied by their respective probability) is multiplied by the score of 'how much a company is impacted by the news itself'. Then, since now some time has passed, we check for Dual Listing profits again. Next, we traverse all primal instruments, delete all existing orders and retrieve their theoretical mid_prices and calculate our bid and ask prices. The volume is calculated based on the calculated weighted company score, which is decayed by the amount of time that has passed, since the news occured. All ordering decisions are immediately "mirrored" in the dual market, by swapping the bid and ask volume as well as calculating a new bid and ask price for the new market.

Challenges we ran into

Tuning the weighting of the news statement impact was difficult, as the current solution tends to order conservatively, but other values did not show improvements. Further, designs to group the impact of news statements towards several companies did not significantly improve performance.

Accomplishments that we're proud of

Getting to know the model and at least some mechanics behind it. Having some results in the top ten, even if only briefly :)

What we learned

You need to work an many different problems at the same time to manage the time pressure efficiently. The trading models are very sensitive, even in this somewhat abstract simulation.

What's next for Team D.S.S.

tbd

Built With

• tbd