Listen to MACD and do the opposite.

[Wary]

I've just manually simulated MACD-based trading for the last year.

Took 3 days intervals chart from bitcoinwisdom.
We are supposed to sell when the signal line (the blue one) goes below MACD line(the brown one) and buy - when it goes above. Right?

For the last year there was only one pair of crossings: it went below at June 14 and crossed back at Aug 19
The sell price (middle of candle for June 14) was 100, and buy price for Aug 19 was 105.
So I've ended up with 5% less BTC than started.
3d-MACD trade gives you 5% of loss.
 
I've done the same with 1d intervals.
There were 5 pairs of crossings. Here are the results:

sold       bough    prices     gain

Apr 16   Apr 23   70-130    -46%
May 2   May 23  112-126   -11%
June 8   Jul 16    114-92    +24%
Dec 16  Jan 3     780-795   -2%
Jan 26   Mar 7    811-652   +24%

Totally 27% of loss.

WTF?!!!  Is there something I'm not getting about this MACD stuff?

[smoothie]

I've just manually simulated MACD-based trading for the last year.

Took 3 days intervals chart from bitcoinwisdom.
We are supposed to sell when the signal line (the blue one) goes below MACD line(the brown one) and buy - when it goes above. Right?

For the last year there was only one pair of crossings: it went below at June 14 and crossed back at Aug 19
The sell price (middle of candle for June 14) was 100, and buy price for Aug 19 was 105.
So I've ended up with 5% less BTC than started.
3d-MACD trade gives you 5% of loss.
 
I've done the same with 1d intervals.
There were 5 pairs of crossings. Here are the results:

sold       bough    prices     gain

Apr 16   Apr 23   70-130    -46%
May 2   May 23  112-126   -11%
June 8   Jul 16    114-92    +24%
Dec 16  Jan 3     780-795   -2%
Jan 26   Mar 7    811-652   +24%

Totally 27% of loss.

WTF?!!!  Is there something I'm not getting about this MACD stuff?

You forgot to read the "disclaimer" that said "This indicator doesn't guarantee a profit."

[Dalmar]

You can anticipate crossovers on the shorter time frames and in conjunction with other indicators. That April loss could have been avoided by using a stop and seeing that the rally obviously exhausted on the RSI.

[Hpotsirc]

I am quite new to TA myself but as far as i understand it a MACD crossover gives a buy/sell signal but this shouldn't be used a alone. You should rather use several indicators of different types and trade when most or all of them agree on a signal.

[Wary]

You can anticipate crossovers on the shorter time frames

Anticipate? Do you mean one shouldn't wait for the crossover, but buy/sell earlier? Say, when distance between the lines reduced to, say, 1/3 of maximum.

That April loss could have been avoided by using a stop and seeing that the rally obviously exhausted on the RSI.

Was it showing on RSI? The spells of green daily candles weren't becoming shorter: gggggRRggggRggggggggRgggCRASH

[Wary]

You forgot to read the "disclaimer" that said "This indicator doesn't guarantee a profit."

It does guarantee the profit, it just doesn't guarantee that it's you who'll get it

[chessnut]

WTF?!!!  Is there something I'm not getting about this MACD stuff?

rule of thumb, if a computer can do it, you will lose your money.

[Wary]

You should rather use several indicators of different types and trade when most or all of them agree on a signal.

Agree. It just seems that this specific indicatog gives you signal when it's too late.

rule of thumb, if a computer can do it, you will lose your money.

Wow! Well said!

[Hpotsirc]

Agree. It just seems that this specific indicatog gives you signal when it's too late.

Well i guess that is due to the nature of the MACD since it is trend following. But i think this problem can be ironed out by tuning. Have you tried using other combinations? On Bitcoinwisdom you can set them under Settings -> Indicator Parameters. Maybe you can improve your results by playing around with this a little bit. A quite "famous" combination seems to be 10/21.
EDIT:I also remember reading quite an important fact about TA: Somebody said that they will never be right 100% of the time, so a good indicator will propably just give you good signals somewhere between 60%-90% of the time (Please note that these numbers are made up by me!). Thats the reason why you always need to set up a drop-loss threshold for each trade depending on how much money/coins you are willing to risk.

[RyNinDaCleM]

When back testing like that, use the close price since that's when the generated signal should be acted upon.
Second, it's not going to guarantee profits but it does tend to keep you in when the trend is up and get you out when the trend is down. Using daily (or even more so the 3-day), the price will definitely bottom and turn before the MACD crosses up and similarly top before it crosses down, so yes, you may end up selling high and buying higher, but IF it had continued down for a longer period of time, you would be out effectively preserving capital. It's more of a safety than the holy grail.

If you want to front run the MACD, there are better ways than trading when it gets close. This is a good way to lose money or your precious BTC.

And as others have said, one indicator alone will give false signals especially during sideways consolidations. If you want to range trade at these times, then use a shorter time frame like 6hr or 4 hr. But you will still have losses here and there due to volatility whipping the price around faster than the indicator can keep up.

[chabmarwen]

I also did the backtesting for the 4h MACD and the 1h MACD, and the conclusion was so obvious !
The 4h chart gave me a small profit (considering going fiat when a bearish crossover occured-and not going short on bitfinex)
Whereas the 1h chart gave me huge profit. I did this backtesting from september til january.
At that time, I didn't know anything about margin trading. Considering these new options (leverage+possibility of shorting), it was obvious that you could be a winner by just following MACD 1h crossovers. You can ofcourse have a better strategy if you add other indicators and add fundamental analysis to your strategy.

I'd like to add that considering the fluctuations of bitcoin, especially during bubbles (like in April or November), using MACD in such big timeframes is quite irrelevant. I didn't try to go in 15m timeframe because I think you will get eaten by fees and swaps. Finally, you will get more often wrong than right, but you winnings will be much more important and erase your losses.

Hope I did help

[knightcoin]

off topic ( signed out )

waves ...

#define

Supercalifragilisticexpialidocious
https://www.youtube.com/watch?v=t8dZkx3tITk

[Wary]

Everybody, thanks a lot !!!

[seleme]

I also did the backtesting for the 4h MACD and the 1h MACD, and the conclusion was so obvious !
The 4h chart gave me a small profit (considering going fiat when a bearish crossover occured-and not going short on bitfinex)
Whereas the 1h chart gave me huge profit. I did this backtesting from september til january.
At that time, I didn't know anything about margin trading. Considering these new options (leverage+possibility of shorting), it was obvious that you could be a winner by just following MACD 1h crossovers. You can ofcourse have a better strategy if you add other indicators and add fundamental analysis to your strategy.

I'd like to add that considering the fluctuations of bitcoin, especially during bubbles (like in April or November), using MACD in such big timeframes is quite irrelevant. I didn't try to go in 15m timeframe because I think you will get eaten by fees and swaps. Finally, you will get more often wrong than right, but you winnings will be much more important and erase your losses.

Hope I did help

Any indicator/bot under the sun would give you a profit in that period. It was period of huge bubble, then huge crashes. Just go to cryptotrader.com and backtest any bot you find there for that period and it will be big winnings.

It's long flat markets where the value of indicators comes to shine.