AQR Hedge Fund moving into Trend Following

[JezLiberty]

Quote from Daal:

TF claim a uptick/downtick is more likely to be followed by another uptick/downtick. This defies economic theory where higher prices bring about supply and decrease demand and lower prices create demand and decrease supply, if markets werent mean reverting by nature, that is if they were more or less random but with a bias of going to whichever direction they recently headed you would have absolute gigantic mispricings(like a stock market with a normalized PE of 1000). The reason that doesnt happen is because economics kicks in and prices mean revert

Now you might claim you can capture the 'middle' moves with a system that worked historically, and perhaps that is true, whether it will keep working who knows, but the TF claim that higher/lower prices are more likely to be followed by higher/lower prices is not correct most of the time

Agree with what you are saying in terms of mean reverting nature of the markets.

However, I would tend to have a slightly different definition of underlying trend following concept
In my opinion, (and you can see how I explained it here) price market distributions are leptokurtic - which simply means that large trends will occur more frequently/at a larger magniture than the normal distribution would suggest. As a result, if you try to get on all the trends, you will necessarily ride all the huge ones. This will provide the bulk of the returns, offsetting any losses that might have occurred in the random/mean reverting periods where trends reverse too quickly.


So, an uptick/downtick is not more likely to be followed by another uptick/downtick (ie the market is more frequently mean reverting) but this is just one parameter of the equation (ie Win %)... You also have to take into account Avg Win / Avg Loss to calculate expectancy.

I believe that Trend Following's underlying concept can be summarised by:

if you follow all trends, despite the predominant mean reverting characteristic of the markets, big trends/wins will arise infrequently and produce more returns than the losses incurred most of the time (ie positive expectancy despite lower probability).

 

[makloda]

Quote from JezLiberty:

if you follow all trends, despite the predominant mean reverting characteristic of the markets, big trends/wins will arise infrequently and produce more returns than the losses incurred most of the time (ie positive expectancy despite lower probability).

On the one hand, you have those guys like Niederhoffer picking up nickles in front of steamrollers. On the other hand side you have guys sitting inside the steamrollers

 

[Daal]

Quote from makloda:

Nobody is saying that, you mix it up once again. No trendfollower will ever say "Buy what's going up because you will then have a higher percentage of winning trades than losing trades". I find it strange this argument keeps coming up, yet nobody in trendfollowing is ever claiming it.

You might want to re-read Trend Following, I remember of Covel's gurus claiming TF was like flipping a coin 20 times knowing its going heads 11/12 times or something like that. I read that claim all that time, the daytrading 'wizards' of the forum routinely claim its a bad idea to short a rising stock market because 'its going up you fool' as if the odds were against you

Quote from makloda:

Economic theories of higher (lower) prices generating more (less) supply do not contradict the profitability of long-term trend following, in fact trendfollowers can profit from long-term economic up and down cycles in commodity markets. How long does it take for new oil fields, copper mines or soy bean plantations to come to the market after prices went up 200%? 1 years? 3 years? More than enough time to ride a trend up and then back down again.

Then you are raising arguments against TF in all kinds of markets where there is almost instant feedback, such as stock markets(where companies can announce IPOs/Equity offerings,supply, at any time), bond markets(same thing). We saw that in early 2009 when corp and junk bond markets where closed but then a rally started and yields went down, that opened that market and a massive flood of issuance started quickly after. That wasnt enough to kill the rally but in my view it wasnt because of some silly TF principle, but simply because of multiple equilibria where the very fact that companies could borrow increased their value as it meant they could continue as a going concern

 

[Daal]

Quote from JezLiberty:

Agree with what you are saying in terms of mean reverting nature of the markets.

However, I would tend to have a slightly different definition of underlying trend following concept
In my opinion, (and you can see how I explained it here) price market distributions are leptokurtic - which simply means that large trends will occur more frequently/at a larger magniture than the normal distribution would suggest. As a result, if you try to get on all the trends, you will necessarily ride all the huge ones. This will provide the bulk of the returns, offsetting any losses that might have occurred in the random/mean reverting periods where trends reverse too quickly.


So, an uptick/downtick is not more likely to be followed by another uptick/downtick (ie the market is more frequently mean reverting) but this is just one parameter of the equation (ie Win %)... You also have to take into account Avg Win / Avg Loss to calculate expectancy.

I believe that Trend Following's underlying concept can be summarised by:

if you follow all trends, despite the predominant mean reverting characteristic of the markets, big trends/wins will arise infrequently and produce more returns than the losses incurred most of the time (ie positive expectancy despite lower probability).

I'm aware of the expectancy argument, in fact I had to point that out to Covel as he was relying in personal attacks to counter Niederhoffer anti TF reasoning. The way to know if there is significant positive expectancy(above the returns of treasuries) is to show the returns of all TF funds adjusted for suvivorship and other biases, not handpicking a few gurus and showing their track record like Covel does

 

[makloda]

Quote from Daal:

You might want to re-read Trend Following, I remember of Covel's gurus claiming TF was like flipping a coin 20 times knowing its going heads 11/12 times or something like that.

Could you please post the exact quote, couldn't find it in Covel's book.

All I found in Covel's book was "There is a random distribution between wins and losses for any given set of variables that defines an edge. In other words, based on the past performance of your edge, you may know that out of the next 20 trades, 12 will be winners and 8 will be losers." by Mark Douglas from "Trading in the Zone". You probably don't mean him as he can hardly be classified as a trend following guru.

Actually, I believe Covel does a reasonably good job at explaining the difference between probability and expectancy (chapter 10).

 

[makloda]

Quote from Daal:

is to show the returns of all TF funds adjusted for suvivorship and other biases

What investment/trading approaches (value investing, equity L/S, global macro etc.) come with this type of proof?

 

[Daal]

Quote from makloda:

Could you please post the exact quote, couldn't find it in Covel's book.

All I found in Covel's book was "There is a random distribution between wins and losses for any given set of variables that defines an edge. In other words, based on the past performance of your edge, you may know that out of the next 20 trades, 12 will be winners and 8 will be losers." by Mark Douglas from "Trading in the Zone". You probably don't mean him as he can hardly be classified as a trend following guru.

Actually, I believe Covel does a reasonably good job at explaining the difference between probability and expectancy (chapter 10).

Most people interpret the 'the trend is your friend' as if prices are more likely to follow their recent action, if that is not the case with Covel's gurus then I was incorrect but I would be surprised, Gartman seems to believe that, so does a number of ET TF cheerleaders, VN debunked that myth for the stock market.

Those who disagree with that definition should put forward a testable definition for the 'trend is your friend'

 

[Daal]

Quote from makloda:

What investment/trading approaches (value investing, equity L/S, global macro etc.) come with this type of proof?

If you correctly pick stocks or macro trades before they move in a lot in your favor, by definition you will beat the market(since you are antecipating it), there is no need for 'proof' of that. Its likely asking for proof that someone will get rich if win the lottery

Now if you ask where is the proof that a human can be trained or can learn how to beat those markets, then I would agree there is a need for proof and I dont know where it is

 

[makloda]

Quote from Daal:

Most people interpret the 'the trend is your friend' as if prices are more likely to follow their recent action, if that is not the case with Covel's gurus then I was incorrect but I would be surprised, Gartman seems to believe that, so does a number of ET TF cheerleaders, VN debunked that myth for the stock market.

IMO, this "trend" or "direction" non-sense isn't half as important as the idea of volatility-based position sizing, letting winners run and cutting losing positions relentlessly (without averaging down). I believe it was David Harding from Winton Capital who said something along the lines of:

"You could even enter in a random direction, we tested it. If you just let your few winners run and cut your losing positions fast then you would have made money historically". I believe it was in one of his interviews somewhere.

 

[makloda]

Quote from Daal:

If you correctly pick stocks or macro trades before they move in a lot in your favor, by definition you will beat the market(since you are antecipating it)

Isn't a similar argument valid for trend following? If you can ride a couple of big trends and otherwise cut your losing positions fast you will likely beat the market, as you are (historically) able to dodge tail risk.

Why is rigorous, scientific proof required for trendfollowing (survivor-ship bias free etc.), but all other trading approaches get the benefit of the doubt. I find that strange.