Gator Quant Hacks 2026

Inspiration

We wanted to turn a trading strategy into working hardware and see how small we could make the circuit while keeping every decision correct.

What it does

Silicon Trade Core receives prices over UART, tracks separate 16-price histories for two items, and returns BUY, SELL, or NONE actions directly from a Tang Nano 20K FPGA.

How we built it

We wrote the design in Verilog and built it with Gowin. Python reference models and simulations checked correctness before board testing. Sharing arithmetic and using block RAM reduced the logic count.

Challenges we ran into

Shrinking the circuit without breaking full-range price calculations was the main challenge. We also had to get UART timing, session resets, and rounding boundaries right.

Accomplishments that we're proud of

The final design uses 186 logic, 92 registers, and four B-SRAM blocks. It passed 13,777 robust packets on the board with zero mismatches or timeouts.

What we learned

Memory timing and arithmetic widths can change hardware cost substantially. We learned to measure each optimization and verify it in simulation and on the physical board.

What's next for Gator Quant Hacks 2026

Our next step is official judging, including the hidden full-range test.

Final Commit SHA

cdaf494679b556682cee785167e04c3a773bcc0c

Built With

• gowin
• verilog