Recently, while backtesting trading strategies, I’ve started to struggle again with the issue of MACD parameters. Honestly, many people pursue the optimal MACD parameters, but I find that idea itself is somewhat flawed.

Let’s start with the default 12-26-9. This set of parameters is indeed the standard on major trading platforms, and the reason is simple—it's sufficiently stable. The fast EMA (12) can reflect market changes over the past two weeks, while the slow EMA (26) shows momentum over the past month. The difference between the two helps you judge the medium-term trend, and the signal line (9) filters out short-term noise. Most importantly, because many traders use it, the market naturally produces a “resonance effect,” which often attracts a lot of attention at key signals, thereby increasing the reference value of those signals.

But this doesn’t mean 12-26-9 is the best MACD setting. For highly volatile markets like cryptocurrencies or traders who prefer short-term trading, this set of parameters might be too smooth and not responsive enough.

I’ve tried several different combinations. 5-35-5 reacts the fastest, capturing short-term trends precisely, but it also produces the most noise, with frequent signals that often fail immediately. 8-17-9 falls in between, suitable for 1-hour charts or markets with larger fluctuations. Going higher, 19-39-9 leans toward medium to long cycles, effectively filtering out most noise. 24-52-18 reacts the slowest, showing clear trends but with fewer signals, suitable for long-term investors.

The key is understanding this trade-off: the higher the sensitivity, the quicker you can catch turning points, but false signals also increase; the lower the sensitivity, the more reliable the signals, but you might miss some opportunities. There’s no absolute “best” MACD parameter—only the one that best fits your trading style.

I’ve seen too many fall into a trap—overfitting. They tweak MACD parameters during backtesting to make it look more accurate, forcing the settings to fit historical data, like writing answers to a test. The backtest results look great, but real trading turns out to be a mess. Overfitting like this is the worst.

Last year, I compared using Bitcoin daily charts from the first half of 2025. The 12-26-9 showed 7 clear signals during that period, with 2 of them resulting in actual upward moves after a golden cross, but 5 failed. Using 5-35-5, the number of signals doubled to 13, with 5 of them leading to noticeable rises or falls, though small gains and losses were more frequent. Notably, during the surge on April 10, both sets of parameters caught the move, but the 5-35-5’s death cross came earlier, reducing the profit. That’s the cost of higher sensitivity.

My advice is: beginners should start with 12-26-9 to observe and get familiar, then adjust based on their trading cycle. For short-term trading, try 5-35-5 or 8-17-9, but be sure to backtest first to see if they align with your entry and exit logic. Once you pick a set of parameters, observe it long-term and avoid changing it frequently—doing so will only turn MACD into a stumbling block in your technical analysis.

Someone asked if it’s okay to use multiple MACD sets simultaneously. Yes, but that means more signals and increased judgment difficulty, testing your decision-making skills.

Ultimately, the idea of a “best” MACD parameter is flawed. Different markets and timeframes vary greatly, and a single setting can’t be perfect in all situations. Instead of obsessing over finding the optimal solution, it’s better to adjust flexibly based on market characteristics and your habits, then verify through backtesting and review. When your chosen parameters underperform recently, adjusting them appropriately might lead to unexpected benefits.

Finally, a reminder: indicators are always just auxiliary tools. No matter how perfect your MACD settings are, they can’t replace proper risk management and trading discipline.