Semtum

The letters

Every two weeks I email what I tested: what failed and why, with the number that killed it, and the rare one that survived. These are the ones that went out, in full. Each carries the figures of its own day.

Get the next one

One email every two weeks with what failed and why, and the rare one that survived. No tips, no signals, and one click to stop.

Your address is used for this and nothing else. Every email has a one-click unsubscribe.


1 November 2026

Seven strategies tested in two weeks. None survived.

Hi,

Every day I take one trading strategy that people teach or sell, and test it on real price history. The last two weeks' verdicts:

Bollinger reversion: rejected. Buy when the price drops unusually far below its recent average, betting it bounces back to the middle. It lost money overall.

RSI mean reversion: rejected. Buy when a market has fallen hard and looks "oversold", betting it bounces back. On the broad test it lost 61%.

MACD: rejected. Buy or sell when two lines that track how fast the price is moving cross each other. Across 10 markets it lost money overall.

VWAP as a signal: rejected. Trade around the day's average price, where busier moments count more. A line many traders watch. Tested on two different markets with two different exits, it did worse than the same rule using a line with no meaning at all.

Session-range breakout: rejected. Mark the high and low of a quiet stretch of trading, then trade the move when the price breaks out as the next market opens. On the German stock index from 2010 to 2025, every version lost money: 0 of 24 start times made a profit.

SuperTrend: rejected. Follow the trend, and get out when the price falls back through a line that trails behind it. It lost money overall, and lost again when it was tested a second time.

Opening-range breakout: rejected. Trade the breakout from the high and low of the first minutes after the market opens. There is something real here, but the famous opening timing is not special: the 09:30 range ranked 13th of 21 start times. Set aside, not traded.

The lesson: test a rule against a fake version of itself

VWAP is a line many traders watch, so a rule built on it feels meaningful. I ran the same rule twice: once on VWAP, once on a line with no meaning at all. On two different markets, with two different exits, the meaningless line did better.

That is the cheapest test there is. Before you trust a signal, swap it for something random and run the rule again. If the random version does as well, the signal was never doing the work.

The running total

800 ideas tested so far, one survivor. Every one is in the public ledger, pass or fail.

The method behind all of it is the course, and the first lesson is free: https://semtum.com/free/?utm_source=newsletter&utm_campaign=2026-11-01

Been sold a trading strategy, or seen one in a Reel? Send me its name and I'll put it on the list to test. Just reply to this email.

Education, not financial advice.