OptiMind

Inspiration

Modern markets move fast - especially when news hits. Traditional market making can’t keep up when prices jump suddenly, and simple quoting strategies often get caught on the wrong side of a move. We built a market maker that doesn’t just place quotes, but actually thinks about what the market is reacting to.

The inspiration was simple: What if a market maker could read news, react instantly, and protect itself - or even profit - from sudden shocks?

What it does

OptiMind is a news-aware, multi-market quoting engine that:

📰 Understands news

It processes incoming social media/news messages and detects which stocks are affected and whether the sentiment is positive or negative.

⚖️ Neutralizes risk instantly

Whenever impactful news is detected, OptiMind:

pulls all quotes for the affected stocks,
neutralizes any open positions,
temporarily stops quoting for a short cooldown.

📈 Trades the news directionally

Instead of just avoiding losses, the system also takes a small, controlled directional hedge based on sentiment - BUY on positive news, SELL on negative news.

This hedge is placed on both markets where the stock is listed (e.g. US & EU variants), ensuring full coverage.

♻️ Unwinds hedges safely

After the cooldown, the system unwinds the hedge, and continues quoting normally.

💱 Finds arbitrage opportunities

Outside of news events, OptiMind watches for cases where one side of the market becomes mispriced and executes quick arbitrage trades to capture free edges.

How we built it

We built OptiMind by combining:

a stable market-making base (with spreads, credit, and position limits),
a sentiment-driven news reaction module,
arbitrage detection for crossing-quote opportunities,

Instead of building a complex model, the focus was on robust behavior:

react fast,
take controlled risk,
and never let news catch the system blind.

Challenges we ran into

Balancing speed vs. rate limits: Responding instantly to news while respecting exchange limits required careful pacing.
Dual-listed instruments: Ensuring both markets reacted consistently without double-counting took some work.
Noisy sentiment: Not every message is clear - the system had to be resilient to occasional misclassifications.
Unwinding hedges safely: Correctly pairing hedge entries and exits was trickier than expected.

Accomplishments that we're proud of

Our bot no longer crashes during news shocks - it actively profits from them.
The hedging framework is clean and controlled.
Arbitrage logic adds free alpha without disturbing normal quoting.

What we learned

Market making is as much about avoiding bad trades as it is about placing good ones.
News events drive strong, persistent price moves - reacting within seconds matters.
Combining NLP with trading logic is powerful, even with simple sentiment analysis.