Learn

Guides on exchange and broker MCP servers, exchange API setup and error codes, and the risk arithmetic behind position sizing — written for traders who run their own keys and their own model.

Topics

• Exchange & Broker MCP 20 articles

  How MCP servers expose your exchange to an LLM, what they can and cannot do, and why order placement needs an approval layer in front of it.

• Exchange APIs 35 articles

  Key permissions, testnet setup, IP whitelists, rate limits and the error codes you will actually hit, per exchange.

• Risk & Position Sizing 30 articles

  The arithmetic behind position size, liquidation price, drawdown recovery and the limits worth enforcing in code.

• AI in Trading 14 articles

  What language models are and are not good for in a trading workflow, and where the failure modes are.

• Glossary 1 article

  Short, precise definitions of the terms used across these guides.

Latest

• Session times and why 24/7 is not uniform

  Crypto never closes but it does go quiet. The same strategy run at the wrong hours pays more and fills worse, for reasons unrelated to the signal.

• Surviving exchange API changes

  Endpoints get deprecated, fields get added, and libraries rewrite themselves. What breaks silently, and how to find out before your positions do.

• AI and trading discipline

  A model can make discipline easier or much harder, and which one depends on where you put it. The failure mode is subtle and worth naming.

• Scaling in vs entering all at once

  A planned ladder and a rescue look identical in the arithmetic. The difference is whether the full size was decided before the first entry.

• Why position sizes drift upward

  Nobody decides to oversize. It happens through six mechanisms that each look reasonable, and only one of them is about greed.

• Choosing an exchange for automation

  The API properties that decide how much work an integration is — and they are not the ones exchanges advertise.

• Reading an equity curve honestly

  The shape tells you about variance and sequence, which the headline return hides. What to look at, and the three ways a curve flatters a strategy.

• Backfilling trade and price history

  Pagination, archive endpoints and retention limits. The three things that make historical data harder to collect correctly than it looks.

• Multi-agent setups and why they compound risk

  Several specialised models sounds like separation of concerns. It multiplies non-determinism, adds untrusted channels between components, and diffuses accountability.

• What to record in a trading journal

  Most journals record outcomes, which is the least useful column. The fields that make a review answer questions, including the ones usually missing.