AlgoGrade

Tag

#build-log

6 articles

strategies8 min read

Risk sizing is a linear dial, no free lunch (part 6)

Why 2x risk per trade means 2x return and 2x drawdown, always. How I size a live Binance futures bot for the bad case, not the good one, and why 25-35% drawdowns are normal.

#build-log#trading-bot
strategies8 min read

ADX regime filtering: when the best trade is no trade (part 5)

How an ADX regime filter cut chop losses from a Binance futures trend bot. Trade above 25, sit flat below 20, hysteresis in between. Why doing nothing beats both alternatives.

#build-log#trading-bot
strategies8 min read

Three bugs that a backtest can't catch (part 4)

Three live bugs a backtest never shows: shared-margin sizing, Binance -4120 algo order migration, and stale-candle stop rejection -2021. Real fixes, real cost.

#build-log#trading-bot
strategies8 min read

The backtest that lied: same-bar re-entry artifact (part 3)

A backtest artifact made one coin look like +5,032% in a year. Fixing the same-bar re-entry bug dropped returns 30-50x and turned 5-7 winning strategies into zero.

#build-log#trading-bot
strategies6 min read

Survivorship bias killed my 1H bot (part 2)

Part 2 of the build-in-public bot series. A holdout test on 11 unseen coins revealed the 1H strategy barely beat a coin flip. The 4H version survived.

#build-log#trading-bot
strategies6 min read

I built a futures trading bot. The honest first month.

Part 1 of building a Binance futures bot in public. The strategy, the bugs that cost real money, and a first month that started down 13%. No fake P&L.

#build-log#trading-bot