tradefloorStart free

Demos

Skill or luck?

3:02 · Recorded on the live product on 10 October 2026. The video has no sound, and a caption at the top narrates each step. Watch on YouTube

About this video

A momentum bot holds the 5 strongest stocks of the last 10 days. Claude runs it with tfrun on tradefloor's tf-quick benchmark, 8 fixed public markets of 20 days, and tries three changes one at a time, running the benchmark after each.

The score is the share of 10,000 random portfolios the bot beat on each market, averaged across the markets. A score of 50 means luck. Each run opens on tradefloor's Benchmarks page, which compares it with the run before on the same markets. None of the changes beats run 1, and Claude puts the original code back.

That code then runs three times on tf-quick sealed, which draws 8 fresh hidden markets of the same kind for every run and reveals them when the run ends.

Chapters

  1. The bot
  2. Tune
  3. Sealed
  4. Result

Figures on screen

Every figure here is on screen in the video and was checked against the product's own data.

  • On tf-quick, run 1 with the original code scored 39, “No sign of skill”. Run 2 with LOOKBACK = 5 scored 28, run 3 with equal weights until 10 days of prices exist scored 37, and run 4 with TOP = 3 scored 30.
  • On tf-quick sealed, the original code unchanged scored 51, 33 and 41, each “No sign of skill”, a spread of 18 points.
  • On tf-quick the bot beat buy-and-hold on 5 of 8 markets, p = 0.36. On each sealed run it beat it on 4 of 8, p = 0.64.

The guide

Skill or luck in a trading strategy runs the same kind of test in Python on 40 generated markets, and shows how a single market can make either strategy look like a genius.

More demos

Run it yourself