Trend Following Is Not Predicting

[bashatrader]

I almost titled this thread "Define Your Terms". With the subtitle "Some Quants are Idiots". Specifically I'm referring to a post in another that references this article:

http://www.priceactionlab.com/Blog/2015/09/trend-following-forecasts/

Notice how Harris and the Qusma blogger he references NEVER define exactly what he means by "forecast"....

Anticipation is a statement about the future which...

So ... common sense tells us that a sure thing is not a prediction. And common sense tells us that most anticipations don't count as predictions either.

I assume that all people that studied finance have taken a course in forecasting 101. I do not think anyone has taken a course in anticipation 101.

Hey guys, if you have taken anticipation 101 please raise your hands.

Covel thinks that trend following is not predicting because the one who is doing it does not do anything, the model does the job. He is a salesman of talk shows. When traders say it predicts, they mean the model makes forecasts. Qusma proved that and Harris explained it in simpler words. They are both top quants.

Forecasting def. from investopedia:

"The use of historic data to determine the direction of future trends."

Nice and simple definition. This guy here is doing a straw-man argument based on anticipation.

Kut2k2, anticipation is the mental state that accompanies prediction.

"Anticipation, or being enthusiastic, is an emotion involving pleasure, excitement, and sometimes anxiety in considering some expected or longed-for good event."

In AI, anticipation is the concept of an agent making decisions based on predictions, expectations, or beliefs about the future.

Kut2k2, stop playing with words like a kid.

 

[kut2k2]

Forecasting def. from investopedia:

"The use of historic data to determine the direction of future trends."

Nice and simple definition.

I totally agree. Forecasting is about future trends.

Guess what? Trend following has nothing to do with future trends. Trend following is all about the current/extant/present trend.

Follow the logic here, dude, you've made it nice and simple yourself.

 

[romik]

I like simple explanations. Trend following is basically hoping on continuation of a trend on whatever time frame is being traded. It's not predicting, it's holding your ballsack tightly and hoping the trend offers you ongoing continuation.

 

[marketsurfer]

The public also has no sell discipline. Nor do they trade in a double-digit number of markets to diversify.

Go shut up and color.

So you need to trade double digit markets to be a trend follower?

 

[marketsurfer]

I like simple explanations. Trend following is basically hoping on continuation of a trend on whatever time frame is being traded. It's not predicting, it's holding your ballsack tightly and hoping the trend offers you ongoing continuation.

Sounds like buy/ sell and hold with stops. Right?

 

[StatGuy]

Anticipation is a statement about the future which, while perhaps not a sure thing, still has a greater than 50% chance of occurring. They are largely based on past experiences, perhaps not personal experiences but the experiences of others like the anticipater, that If A Then B is a likely event.

I will grant the idea that trend followers anticipate. We anticipate, based on past experiences, that trading with the trend is profitable. I personally believe that trading with the trend is virtually a sure thing, the biggest uncertainty being whether the trend follower is detecting true trend or false trend (chop).

Actually, what you are describing is the conditional expectation of a particular event given past information. This is in essence what a statistical prediction is. You can say that the conditional probability of a particular event is > 0.50, but you are indeed making statements involving uncertainty. See conditional expectation and Baye's theorem.

 

[goodgoing]

Actually, what you are describing is the conditional expectation of a particular event given past information. This is in essence what a statistical prediction is. You can say that the conditional probability of a particular event is > 0.50, but you are indeed making statements involving uncertainty. See conditional expectation and Baye's theorem.

That was enough to make him go away.

 

[kut2k2]

Actually, what you are describing is the conditional expectation of a particular event given past information. This is in essence what a statistical prediction is. You can say that the conditional probability of a particular event is > 0.50, but you are indeed making statements involving uncertainty. See conditional expectation and Baye's theorem.

As my response to bashatrader above demonstrates, predicting aka forecasting is about future events, whereas trend following is about current events. Why this is so confusing to people is incredible to me. All trend following says is that if you trade with the current trend, you will profit. When has that not been true? Never. No need to predict anything, just find the current trend and trade with it.

The problem for most is that they are incompetent at finding the current trend, so losses mount, frustration rises and eventually hatred of TF rises as well. Instead of dealing with the issue --answering the challenge of finding the current trend-- it's easier to blame the whole idea of TF as being flawed or requiring some psychic ability to see the future. TF doesn't require psychic ability but it may require the uncommon skill of thinking outside of the box.

 

[panzerman]

The problem for most is that they are incompetent at finding the current trend...

Finding a current trend (however one defines it) is easy, and people are not incompetent at finding it. What makes it difficult to profit is that the beginning and end of trends are randomly distributed through time. Prices transition from trending to cycling and nobody knows when that transition will occur. The transitions are often sharp reversals in direction. If it were easy to profit over time from taking the meat out of the middle of a trend, any N-day moving average system would work.

 

[kut2k2]

Finding a current trend (however one defines it) is easy, and people are not incompetent at finding it. What makes it difficult to profit is that the beginning and end of trends are randomly distributed through time. Prices transition from trending to cycling and nobody knows when that transition will occur. The transitions are often sharp reversals in direction. If it were easy to profit over time from taking the meat out of the middle of a trend, any N-day moving average system would work.

People are incompetent at finding the current trend because they often find chop instead. Sure, strictly speaking, they can the trend when it exists but if they're also finding "trend" where none exists, that effectively means they're incompetent at finding the trend. Not being able to distinguish between trend and chop is like not being able to distinguish between healthy vegetables and toxic plants. How would anyone like to be stuck with that diet?

TF losses have nothing to do with the unpredictability of trend beginnings and endings. It's all about failing to distinguish between trending and chopping regimes, which MA crossovers and channel breakouts routinely fail to do.