Many people are still struggling with how to set up MACD, but in fact, this question itself is wrong.
很多人以為存在某個「最完美」的MACD參數，但實際上沒有。 Many believe there is a "perfect" MACD parameter, but in reality, there isn't.
我自己用MACD這麼多年，最大的體會就是：適合你的參數，才是好參數。 Having used MACD for many years, my biggest insight is: the parameters that suit you are the good ones.
先說最常見的預設值12-26-9。 Let's start with the most common default values 12-26-9.
這組參數確實有它的優勢，穩定性強，快線EMA（12）抓短期動能，慢線EMA（26）看長期趨勢，訊號線EMA（9）用來過濾雜訊。 This set of parameters indeed has advantages: high stability, the fast line EMA (12) captures short-term momentum, the slow line EMA (26) observes long-term trends, and the signal line EMA (9) filters out noise.
因為大家都用這個，市場上形成了某種共識效應，關鍵訊號時會吸引很多人關注，這反而提升了訊號的參考價值。 Since everyone uses this, a consensus effect forms in the market, and key signals attract many attention, which in turn increases the reference value of the signals.
但對加密貨幣這種高波動市場，12-26-9有時候反應太遲鈍了。 But for highly volatile markets like cryptocurrencies, 12-26-9 can sometimes react too slowly.
我見過不少短線交易者就是因為參數不夠靈敏，眼看著機會溜走。 I've seen many short-term traders whose opportunities slip away because the parameters are not sensitive enough.
這時候就要考慮調整。 At this point, it’s time to consider adjustments.
常見的調整方案大概是這樣：想要反應更快可以試試5-35-5或8-17-9，這兩組能更快抓出趨勢但雜訊也多； Common adjustment options are: if you want faster response, try 5-35-5 or 8-17-9; these can catch trends more quickly but also generate more noise;
如果偏好中長期就用19-39-9或24-52-18，訊號少但更可靠。 If you prefer medium to long-term, use 19-39-9 or 24-52-18; these produce fewer signals but are more reliable.
我自己在做MACD參數最佳化時，會根據交易週期來選。 When I optimize MACD parameters myself, I choose based on the trading cycle.
比如日線通常12-26-9就夠，但如果看小時圖就會考慮5-35-5。 For example, daily charts usually suffice with 12-26-9, but for hourly charts, I consider 5-35-5.
去年我試過用5-35-5回測比特幣半年的數據（2025年1月到6月），發現訊號確實多了一倍，從7次增加到13次。 Last year, I tested Bitcoin data from January to June 2025 using 5-35-5, and found the signals indeed doubled, from 7 to 13 times.
但這也意味著失敗訊號也更多。 But this also means more false signals.
有意思的是在4月10日那波起漲，兩組參數都抓到了，只是5-35-5的死亡交叉來得更早，導致利潤反而不如12-26-9。 Interestingly, during the rally on April 10, both sets caught the move, but the death cross with 5-35-5 came earlier, resulting in less profit compared to 12-26-9.
這就引出了MACD參數最佳化最常見的坑：過度擬合。 This highlights the most common pitfall in MACD optimization: overfitting.
很多人在回測時為了讓參數看起來完美，會刻意調整到完全符合過去的盤面。 Many people, during backtesting, deliberately tune parameters to perfectly fit past market conditions.
這根本就是看著答案寫考卷，回測數據再漂亮也沒用，一到實盤立馬失效。 It's like writing an exam with the answer key; no matter how perfect the backtest looks, it fails immediately in live trading.
我的建議是這樣：新手先用12-26-9觀察一段時間，習慣了再考慮調整。 My advice is: beginners should first use 12-26-9 for a period, and only consider adjusting after getting used to it.
如果決定換參數，一定要做充分的回測，但回測時要注意別陷入過度擬合的陷阱。 If you decide to change parameters, be sure to do thorough backtesting, but watch out for overfitting traps.
找到一組適合自己交易風格的參數後，就堅持用一段時間，別頻繁換來換去。 Once you find a set that fits your trading style, stick with it for a while and avoid frequent changes.
有些人會同時看兩組MACD來對比，這也可以，但訊號變多意味著判斷難度也上升。 Some traders compare two MACD setups simultaneously, which is fine, but more signals mean increased difficulty in judgment.
最後想說的是，MACD只是工具，參數調整也只是優化工具的過程。 Finally, remember that MACD is just a tool; parameter adjustment is merely part of optimizing that tool.
真正重要的還是你的交易策略和風險管理。 What truly matters are your trading strategy and risk management.
再完美的MACD參數最佳化也替代不了紀律和耐心。 Even the most perfect MACD optimization cannot replace discipline and patience.
最近在看不少人還在糾結MACD要怎麼設，其實這個問題本身就問錯了。
Many people are still struggling with how to set up MACD, but in fact, this question itself is wrong.
很多人以為存在某個「最完美」的MACD參數，但實際上沒有。
Many believe there is a "perfect" MACD parameter, but in reality, there isn't.
我自己用MACD這麼多年，最大的體會就是：適合你的參數，才是好參數。
Having used MACD for many years, my biggest insight is: the parameters that suit you are the good ones.
先說最常見的預設值12-26-9。
Let's start with the most common default values 12-26-9.
這組參數確實有它的優勢，穩定性強，快線EMA（12）抓短期動能，慢線EMA（26）看長期趨勢，訊號線EMA（9）用來過濾雜訊。
This set of parameters indeed has advantages: high stability, the fast line EMA (12) captures short-term momentum, the slow line EMA (26) observes long-term trends, and the signal line EMA (9) filters out noise.
因為大家都用這個，市場上形成了某種共識效應，關鍵訊號時會吸引很多人關注，這反而提升了訊號的參考價值。
Since everyone uses this, a consensus effect forms in the market, and key signals attract many attention, which in turn increases the reference value of the signals.
但對加密貨幣這種高波動市場，12-26-9有時候反應太遲鈍了。
But for highly volatile markets like cryptocurrencies, 12-26-9 can sometimes react too slowly.
我見過不少短線交易者就是因為參數不夠靈敏，眼看著機會溜走。
I've seen many short-term traders whose opportunities slip away because the parameters are not sensitive enough.
這時候就要考慮調整。
At this point, it’s time to consider adjustments.
常見的調整方案大概是這樣：想要反應更快可以試試5-35-5或8-17-9，這兩組能更快抓出趨勢但雜訊也多；
Common adjustment options are: if you want faster response, try 5-35-5 or 8-17-9; these can catch trends more quickly but also generate more noise;
如果偏好中長期就用19-39-9或24-52-18，訊號少但更可靠。
If you prefer medium to long-term, use 19-39-9 or 24-52-18; these produce fewer signals but are more reliable.
我自己在做MACD參數最佳化時，會根據交易週期來選。
When I optimize MACD parameters myself, I choose based on the trading cycle.
比如日線通常12-26-9就夠，但如果看小時圖就會考慮5-35-5。
For example, daily charts usually suffice with 12-26-9, but for hourly charts, I consider 5-35-5.
去年我試過用5-35-5回測比特幣半年的數據（2025年1月到6月），發現訊號確實多了一倍，從7次增加到13次。
Last year, I tested Bitcoin data from January to June 2025 using 5-35-5, and found the signals indeed doubled, from 7 to 13 times.
但這也意味著失敗訊號也更多。
But this also means more false signals.
有意思的是在4月10日那波起漲，兩組參數都抓到了，只是5-35-5的死亡交叉來得更早，導致利潤反而不如12-26-9。
Interestingly, during the rally on April 10, both sets caught the move, but the death cross with 5-35-5 came earlier, resulting in less profit compared to 12-26-9.
這就引出了MACD參數最佳化最常見的坑：過度擬合。
This highlights the most common pitfall in MACD optimization: overfitting.
很多人在回測時為了讓參數看起來完美，會刻意調整到完全符合過去的盤面。
Many people, during backtesting, deliberately tune parameters to perfectly fit past market conditions.
這根本就是看著答案寫考卷，回測數據再漂亮也沒用，一到實盤立馬失效。
It's like writing an exam with the answer key; no matter how perfect the backtest looks, it fails immediately in live trading.
我的建議是這樣：新手先用12-26-9觀察一段時間，習慣了再考慮調整。
My advice is: beginners should first use 12-26-9 for a period, and only consider adjusting after getting used to it.
如果決定換參數，一定要做充分的回測，但回測時要注意別陷入過度擬合的陷阱。
If you decide to change parameters, be sure to do thorough backtesting, but watch out for overfitting traps.
找到一組適合自己交易風格的參數後，就堅持用一段時間，別頻繁換來換去。
Once you find a set that fits your trading style, stick with it for a while and avoid frequent changes.
有些人會同時看兩組MACD來對比，這也可以，但訊號變多意味著判斷難度也上升。
Some traders compare two MACD setups simultaneously, which is fine, but more signals mean increased difficulty in judgment.
最後想說的是，MACD只是工具，參數調整也只是優化工具的過程。
Finally, remember that MACD is just a tool; parameter adjustment is merely part of optimizing that tool.
真正重要的還是你的交易策略和風險管理。
What truly matters are your trading strategy and risk management.
再完美的MACD參數最佳化也替代不了紀律和耐心。
Even the most perfect MACD optimization cannot replace discipline and patience.