Grid trading bot, backtested properly: a variance ratio test on 12,000 bars shows why this algorithmic trading strategy cannot work. Full Python backtesting code, every number on screen.

I built the grid bot the way the tutorials say – adaptive spacing, indicator
filters, risk management – then ran one test before switching it on. The test
said the strategy cannot work on this market. Here is that test, the number it
produced, and what I did instead.

Data: MOEX ISS public API, no key required

00: 00 The claim
00: 35 What a grid bot actually bets on
01: 22 Building it properly
01: 56 The variance ratio, in ten lines of code
02: 29 Running the test
03: 03 The 30-day sample that nearly fooled me
03: 59 The grid on trending data
04: 25 Taking the other side of the bet
05: 17 The number nobody puts at the end

Verdict: DOES NOT WORK

Topics: grid trading bot, variance ratio, mean reversion, algorithmic trading,
backtesting in Python, walk-forward analysis, quantitative trading, trading bot
tutorial, overfitting, Monte Carlo drawdown.

—
I test trading strategies with real data and publish the verdict, including when
the answer is no. Every episode ends with one of three: WORKS / DOES NOT WORK /
WORKS BUT NOT WORTH IT. All code open source, all numbers on screen.

No signals. No courses. No performance promises.
Nothing here is financial advice.

#algotrading #quantitativefinance #backtesting #python #tradingbot