MACD Trading Strategy Tested: Does the Crossover Work?

What MACD measures

Gerald Appel developed the Moving Average Convergence Divergence indicator in the late 1970s and later laid out how he uses it in Technical Analysis: Power Tools for Active Investors (2005). With its default settings:

• the MACD line is the 12-period exponential moving average of price minus the 26-period one;
• the signal line is a 9-period exponential average of the MACD line;
• a bullish crossover is when the MACD line crosses above the signal line, a bearish crossover when it crosses below.

The most common MACD strategy buys on a bullish crossover and sells on a bearish one. Traders often add a filter: bullish crossovers above zero, when the longer trend is already up, are seen as stronger.

How we tested it

• Data: weekly closes, adjusted for splits and dividends, for SPY and for Apple, Microsoft, Nvidia, Amazon, Alphabet, Meta and Tesla.
• Settings: the defaults, 12, 26 and 9, on weekly bars. We didn't try any other settings.
• Measure: returns over the next 4 and 13 weeks after each crossover, compared with all weeks.
• Random comparison: 10,000 random sets of weeks of the same size.
• Timing rule (SPY only): hold SPY when MACD is above its signal line, hold cash otherwise, decided at each weekly close.

Results on the S&P 500

• Bullish crossover59
• Bullish crossover above zero40
• Bullish crossover below zero19
• Bearish crossover60
• Any week1,370

Exact values from table 2 in this article. Bar lengths share a zero baseline; no new calculation or simulated price path. Read the table and its stated sample and limits in this article.

Bullish crossovers were just average. The next 13 weeks averaged +2.44%, against +2.35% for any week. Random sets of 59 weeks did as well almost half the time.

Bearish crossovers were followed by more gains, not fewer. After the 60 sell signals, the next 13 weeks averaged +3.19%. On an index that went up most of the time, a sell signal was more often a pause than a turn.

The above-zero filter looked better, on a small sample. Bullish crossovers while MACD was above zero were followed by gains in 82% of cases. With 40 signals, some overlapping, that's suggestive, not proof.

Crossovers came about twice a year, so 26 years only gave about 60 of each kind.

The MACD strategy, tested as a timing rule

The rule cut the worst decline, but not by much, and gave up most of the return. It was out of the market 45% of the time, often during the bounces that come after sell signals. For comparison, the simpler 10-month moving average rule on the Nasdaq-100 cut the worst decline by about two-thirds while keeping more of the return.

The cash periods here earn nothing. Adding Treasury-bill interest would raise the rule's return by well under one point a year.

Results on single stocks

Three stocks did better than their average after bullish crossovers, four did worse. Apple and Amazon came closest to a meaningful result, with about 1 chance in 10 of matching it at random, but when you test seven stocks, one or two results like that are expected from luck alone. Tesla's bullish crossovers were followed by way worse returns than its average.

These seven stocks were picked with hindsight, which inflates their average returns, but that doesn't favor crossover weeks over other weeks.

Why MACD disappoints as a stand-alone signal

• It's built from moving averages, so it reacts after the move. By the time weekly averages cross, a lot of the move has often already happened.
• In rising markets, sell signals are expensive. Every exit is a bet against the long-term drift.
• Crossovers pile up in sideways markets, creating whipsaws.

A lot of published MACD strategies look better because they search for the best settings on one market and one period. Cheol-Ho Park and Scott Irwin, reviewing the technical analysis literature in the Journal of Economic Surveys (2007), pointed to that kind of data snooping as a central problem.

What this means for traders

• Use MACD to describe momentum, not as an entry trigger. Whether MACD is above or below zero says something about the trend. A crossover alone said very little here.
• Be careful with sell signals in uptrends. On the S&P 500, they were followed by above-average returns.
• If you use MACD intraday, test it there. Weekly results don't automatically carry over to 5-minute charts.

Same as with RSI, the indicator describes what already happened better than it predicts what's next.

How to check it yourself

1. Add MACD(12,26,9) to a weekly chart of the market you trade.
2. Mark every bullish and bearish crossover over at least ten years.
3. For each, note the close 13 weeks later, and compare with all 13-week periods.
4. If you add a filter, like "above zero", decide on it before looking, and count every signal it gives.
5. If you use MACD divergences, test them with the same discipline. We didn't test them here.

Limits

• Weekly bars only. Daily and intraday MACD are different signals.
• Overlapping windows. Thirteen-week returns from nearby signals share data.
• Survivor stocks for the single-stock test.
• No costs or taxes in the timing rule; cash earns zero.

How we did it

• Weekly adjusted prices from Alpha Vantage (TIME_SERIES_WEEKLY_ADJUSTED), through October 9, 2026. The first 35 weeks of each series were only used to initialize the averages.
• MACD(12,26,9) on adjusted weekly closes; signals at the weekly close; returns measured from that close.
• Random comparison: 10,000 draws per market, with a fixed seed.

Sources

• Appel, G. (2005). Technical Analysis: Power Tools for Active Investors. FT Prentice Hall.
• Park, C.-H., & Irwin, S. H. (2007). What do we know about the profitability of technical analysis? Journal of Economic Surveys, 21(4), 786–826. IDEAS/RePEc
• Alpha Vantage. TIME_SERIES_WEEKLY_ADJUSTED (SPY, AAPL, MSFT, NVDA, AMZN, GOOGL, META, TSLA). Data retrieved October 2026.

FAQ

Does the MACD trading strategy work? On weekly S&P 500 data since 2000, bullish crossovers were followed by average returns, and bearish crossovers by above-average returns.

What are the best MACD settings? We used the defaults (12, 26, 9) and didn't optimize. Picking the best settings from past data usually overstates future results.

Is a MACD crossover above zero better? In our sample, bullish crossovers above zero were followed by gains 82% of the time over 13 weeks, on 40 signals. That's suggestive but not reliable.

Does MACD work on individual stocks? On seven mega-caps, results were mixed and none clearly beat luck.

Should I use MACD for day trading NQ or ES? This test only covers weekly bars. Test intraday MACD separately, with fixed settings and costs included.

---

Research and education only. Not financial advice. Past results do not predict future returns. A backtest is not real performance.

How this article was made: drafted with AI assistance from public data and the sources listed above; facts and figures checked by Gabriel Décarie and Karel Panneton.