Contents
The frame of the comparisonThe landscape tablefreqtradehummingbotjesseHow people actually chooseThe frame of the comparison
These tools overlap on the surface — all three are Python bot frameworks that talk to exchanges — and diverge underneath, because each was built around a different primary job. Comparing them on "features" misleads; comparing them on the job they were built to do is honest.
The landscape table
| Tool | Built for | Feels like | Weak spot |
|---|---|---|---|
| freqtrade | Running a strategy of your own design, long-running | A well-documented framework | Not a market-making engine; the UI is optional and thin |
| hummingbot | Making markets and arbitrage across venues | A trading desk in a terminal | Heavier config; its own vocabulary to learn first |
| jesse | Research and strategy design | A lab notebook with a live mode | Smaller ecosystem; fewer ready strategies to borrow |
freqtrade
Its centre of gravity is the strategy file: a small, readable Python class that the framework runs for you. The framework handles the plumbing — exchanges, orders, candles, the web UI, telegram controls — and stays out of the way of the strategy. The ecosystem is the largest of the three, which cuts both ways: more examples to learn from, and more inherited bugs floating around.
hummingbot
Its centre of gravity is the order book: keeping two-sided quotes up, moving with inventory, or crossing venues when prices diverge. Strategies here are more like recipes with dials than like code you write yourself, and the framework's own concepts — connectors, controllers, inventory — need to be learned before the tool opens up. For market-making work there is little else in open source at this level.
jesse
Its centre of gravity is the research loop: define a strategy in a very clean API, backtest it quickly, iterate. The API is the tidiest of the three, and the community is smaller and quieter. For someone whose main activity is designing strategies rather than running them, the smaller ecosystem is a fair price for the cleaner loop.
How people actually choose
- "I have a strategy idea and want it running on my own machine" → freqtrade.
- "I want to quote two sides of a market or bridge two venues" → hummingbot.
- "I want to design and test strategies, not administer a deployment" → jesse.
- "I want to be sure I understand what my bot does" → whichever one you will actually read the source of.
The last bullet is the real one. All three are open and auditable; the differences in features matter far less than the difference between "the tool I read" and "the tool I trust blindly".