A moving average crossover is a rule that acts when a faster moving average crosses a slower one. The fast line follows recent prices closely, the slow line follows them loosely, so a cross says that recent prices have moved away from the longer average.
The rule, written in full
- Choose two lengths, for example 10 and 30, and a type for each (simple average or exponential average).
- Buy signal on the close where fast is above slow after being at or below it the bar before.
- Sell or exit signal on the close where fast is below slow after being at or above it the bar before.
- Nothing else is added. A test with extra filters is testing the filters.
- The 50 and 200 day pair is a famous special case: see the golden cross.
Worked example (arithmetic, not a chart)
Say the 10 bar average was 49.6 and the 30 bar average 49.8 yesterday, and today they are 50.1 and 49.9. Yesterday fast was below slow, today it is above: a buy signal at today's close. The numbers are made up to show the check.
Why every pair is its own test
10 and 30, 20 and 50, 5 and 20, simple or exponential: each choice is a different rule. Try enough pairs on one market and one of them will look good by chance alone, because some pair always fits the past. That is why a result from "the best pair" tells you little. When we test crossovers we count every pair we run, not only the winner.
What it cannot tell you
- Both lines look backward, so the signal comes after the move has begun.
- In a sideways market the lines cross again and again and each cross costs spread or commission.
- Shorter pairs trade more often, which raises costs; longer pairs react later.
How we will test it
We will fix a list of pairs in advance, run every one of them on many markets and years with realistic costs, keep part of the data unseen until the list is fixed, and report the whole spread of results, not just the best, with sample size and the markets where they failed. Last tested: pending re-run. No result is stated here until it is run and dated.
Questions people type
Does a moving average crossover work? It depends on the pair, the market and the costs. The test above measures the whole range; we do not state an answer before it is run.
What are the best moving average settings for a crossover? There are no best settings in general. The best pair on past data is the one most likely to be a coincidence.
Simple or exponential? An exponential average reacts a little sooner. Each is a different rule and each needs its own test.
Related
Golden cross, 200 day moving average, MACD indicator, how we test, backtesting, method and limits.
Education only. Nothing here is financial advice or a signal.