July 11, 2019

Every day that passes I can see how there is more and more news, interviews and articles related to one of the current trends: artificial intelligence and its growing importance in trading.


After all, the cryptocurrency bubble is not something casual, but it gives us a clue as to the direction this world is heading at a forced pace, in a way that is almost similar to the Bitcoin quote: to a standardized world, which is just another way of calling a virtual world.

There are those who say that this type of trading has ended with the possibility of obtaining benefits in certain sectors such as commodities.

It seems that robots and trading algorithms are destroying the hedge fund industry.

That is what the manager says of what was, perhaps, the main commodity hedge fund of the entire market, Stephen Jamison and his Koppenberg Macro Commodity Fund.

In fact, the situation seems to be so bad that the same manager has decided to close the fund with the end of this month.

Interestingly, a few years ago we had these funds were quite fashionable, including the aforementioned, as we can find, also in Reuters, which back in 2013 reminded us that "raw materials niche hedge funds beat the market with Jamison doubling the capital invested. "

That is, just over 4 years ago, this fund was beating the market and the manager was overwhelmed by so many people who wanted to invest in it.

It was a success story.

In the years after the collapse of 2008, commodities performed quite well, from a bullish point of view, with a great recovery of oil and energy and with the large metals bull market that ended in 2011. That is why the Jamison fund obtained returns of 18% in 2010, 8% in 2011 and 2.7% in 2012.

With these data people were willing to participate in the "success of raw materials" and therefore, the Jamison fund was one of the most desired.

Well, this case is for us to see how what seemed a few years ago the best strategy in the world, ended up resulting in a fiasco to the point of closing.


First of all, let's see a little why this hedge fund is supposed to be successful.

Well, if we look at the original Reuters article, the message we can get is that this hedge fund was performing well when the commodities market was a bull market, that is, like the one from 2009 to 2011.

From this it is understood that this Jamison applied trading strategies in the medium or long term to take advantage of the bull market of raw materials, hence the good results in those years.

However, if we look at the graph of raw materials in recent years, and since 2011, we see that it has been a disastrous market to go long, with a very marked downward trend, something we know very well if we have notions of what has happened these years with gold, oil or grains.

End of hedge funds by algorithms?

I must admit that the recent article on the closing of this hedge fund is somewhat confusing, or at least it seems to me.

In the end you do not know whether to think if what this manager has is a typical tantrum from which he has not known how raw materials have had or were going to have a bear market or if there is any truth in the fact that the "fault" of the failure of his strategy is of the "robots".

Let's see if we can see a little more about the subject.

Apparently, according to this hedge fund manager the problem lies in two issues:

1. That machines and artificial intelligence have made success impossible in short-term trading strategies

 2. That the raw materials do not offer any advantage for long-term investments, since they also do not offer dividends, coupons or the appreciation in the price necessary to mitigate these problems.

Artificial intelligence in short-term trading of raw materials

Well, I'm going to try to dismantle this statement a bit, and I'm going to use a pretty logical point of view.

If this were true, artificial intelligence would not only have destroyed short-term trading in all commodities, but also in the rest of the markets, but it seems that there are a lot of hedge funds that are behaving well and have no plans close.

Then this statement can not be true.

That is, if other hedge funds that also apply short-term trading strategies do well, why does this go wrong and has to close?

Well, in this case Jamison could say that these other hedge funds invest in stocks and bonds and that these assets, by doing well in the long term, make those funds earn money.

Ok, but then what I would have to admit is that your fund does not make money because the long-term trend of raw materials is not good, at least in the last 5 years. That is, we would have to go directly to point 2.

This does not mean that in "artificial intelligence" or machines have no influence in the world of short-term trading. In fact, global dealers are almost totally automated, and every day that passes they use more robots and algorithms for risk control and execution of operations, but that does not mean that the people who manage the systems have disappeared. There are still many people in front of the computers in the offices, not already of the hedge funds, but of Goldman Sachs or J.P. Morgan

This is something that affects the dealer side of the market much more, of course.

On the trader side, both wholesalers (hedge funds) and retailers (individuals), the trading of "algorithms" and "artificial intelligence" is as available to use as it can be for any opponent.

That is, any hedge fund that can hire a good team of developers can devise automated trading strategies in the short term.

But according to my knowledge and experience the great majority of algorithmic strategies that are tried to put in service of the particular traders end in failure, but not because those algorithms or robots are not complex.


They fail because of the simple fact that in order to earn frequent trading from the trader's side, it is necessary to spin very, very fine, and that is something that not even most hedge funds can do.

Another thing is if we go to the dealers side.

In this sense, it would be good to know what kind of short-term trading strategies this man talked about. I would like to know if your hedge fund applied short-term trading as a dealer or as a trader.

The difference between those two points is like night to day.

Regarding point 2, here we have a fairly logical point of why hedge funds or any fund, in fact, that have invested in raw materials have had a very bad years, hence many are thinking about closing.

You just have to take a look at the markets of grains or metals to realize that those markets have been the worst to invest in recent years, and without returns.

Of course, trading in commodities is complicated.

This is not going to give you dividends or long-term coupons.

If you want those dividends and coupons then I always say the same: buy mining, energy or related to raw materials in general.

If what you want is to trade in those markets using futures and other instruments, then obviously you will need volatility and strong movements to the upside, because otherwise you will be playing a totally losing game. For example, if we look at the oil chart of the last 5 years, we can see that it has been a complete disaster, at least from the point of view of the "commodities funds".

Now, if those funds are said to be "traders" then, why did not they operate the short side of the market in 2014, when they could have done a "killing"?

Well, for the simple reason that the short side is very complicated to operate, and those hedge funds just want to trade the long.

Come on, I suppose, because if they were traders that operated both sides of the market they would have no excuse, since they had a 70% drop in a matter of two years and that in trading is a lot.


Regarding the fact that things have not gone very well in the last two years, it is understandable.

As I said, many of the commodities markets are dead, such as gold, corn, or with important falls such as sugar or natural gas.

Thus it is complicated to trade upwards, much less "invest" in those futures. However, and curiously, one of the main markets for raw materials, oil, has had a bull market quite appreciable for almost a year, bouncing from 42 euros to 66 today (July 11, 2019); more than 50% increase.

In this case I even decided to make an entry from $ 54 since I saw a possibility that the next break had some strength, and it was.

From that approximate price and applying bullish trading strategies, I have maintained the position until today, with a rise of more than 20%, which is not bad in a market that has been so depressed in recent years.

There is no doubt that automation and standardization have to do with the virtual, as opposed to the real and differentiated.

Here the artificial intelligence trading only confirms that the world of trading, like that of finance, and all economic sectors - and not so economic - of the world, are converging towards a more predictable model, to call it A) Yes.

Well, I have to admit that this field of "artificial intelligence" does not stop confusing me a little.

When we read about artificial intelligence, it is logical that we think that we are referring to a kind of Skynet, or something similar to what happened in the pre-Matrix world; which I do not rule out, of course.

However, when I read those articles about this supposed intelligence it is not what I deduce at first glance.

It just gives me the feeling that they are talking about more advanced applications of automation algorithms, not that we are really talking about a kind of "artificial intelligence".

The problem with artificial intelligence - let's call it that - is that, unlike the "real" one, it is not a genuine intelligence, but an intelligence that can only copy, not create. That is, it is an intelligence that is excellent in copying, in standardization, in automation, but never in creation.

I do not know if it does not happen to you that when you read about this topic, whether in the field of trading or someone else's, it seems to you that we are really talking about automated processes that are far from being a true "intelligence", in the creative sense of the word.

For example, in an article by itprotoday entitled "How artificial intelligence is going to shake the foundations of the trading industry", one of the experts on the subject, the CEO of Botworx.ai, Mahi de Silva, says things like:

 "The first benefit of computerized trading is its speed. Some aspects of financial trading are already making significant use of automation and pushing the speed advantage to billions of dollars. "

"The trading of futures, stocks and bonds, is strongly automated ..."

That is to say, that the same experts of the subject recognize that all this topic of the "artificial intelligence" is not but a massive application and more and more deep of the processes of mechanical automation.

At the same time, for example, in that article they speak of "learning" of machines, as if to say that these machines learn, as they have their own intelligence.

However, the latter does not agree with the previous, that is, with the function "automate" as the basis of this process IA or AI (Artificial intelligence).

I will try to explain a bit above what is more or less one of these projects of "artificial intelligence". With this I hope you can see that we are not really talking about intelligence in the creative sense but in the automation sense, which, in reality, has little intelligence.

In a Techemergence article, Alex Lu, CEO and founder of Kavout, an artificial intelligence company for trading said something like this:

 "We can do this today in natural language processing, which means we can make a computer understand the meaning and semantics of what a certain person says. This could be something positive or negative for certain companies, and that is something we call "sentiment analysis". We are building something called a measure of feeling, which means that we are making use of all the data we collect from traders, news, blogs, etc., and we are putting them together in something we call "measure of feeling." For example, we collect the data of the insiders, in such a way that we know which company or CEO is selling these or those actions. If we put these two datasets together, we can have a more accurate approximation of what people are thinking about actions "

As we can see it is a very interesting statement. And from a trader's point of view it makes a lot of sense.

That is, build a perfect prediction machine to be able to know with better probability what is going to happen in the market.

Everything is quite praiseworthy and logical, and yet this affirmation: the fact of wanting to build this financial weapon is nothing other than trying to apply algorithmic processes of data collection and ordering that is not very different from the one that has been done throughout life, only that in this case it would be done with a much greater complexity.

What the team of this company will try to do is try to apply some tracking system of the sales of the CEOs with some market information provider, and do something similar with the content that is put in the blogs on this topic, or in the news, which will have its complexity, obviously, but that will not stop being an algorithm.

With this, we can see that in reality we are facing a massive increase in automation in all social issues.

There is no doubt that with this issue we are talking about a technological improvement that allows us to produce more and with less resources, not very different, in fact, from the best technologies of the past, such as the printing press, … .

However, inversely, there is a disturbing effect in all this, which is none other than the inexorable path to a mechanized life: that is, to a life that in reality would be nothing but the absence of life.

That's why this is something that does not have to be good for trading or for trading professionals, for example, chartists.

This is quite obvious, because with the automation of processes and the ability to write certain algorithms in computer language makes the analysis of the graphics become a bit obsolete, to put it in some way, leading to the implementation one at a time, greater robotization of all productive issues, not just trading.

When your computers are able to see these "parameters" and get down to work, you no longer need those real "traders".

However, here is something that reminds me that not everything is as simple as it seems, because here I have to comment on two things.

It was never easy to earn money in trading with the simple observation of graphs and "indicators".

The fact that machines can make more and better automatic trading systems, does not mean that they will be winners where real traders were not before.

Where the machines are going to have an indisputable advantage is in the market makers' arena, that is, those that some call "traders" but in reality they are not, because those "equity" or "futures" traders what many people talk about are not JP employees Morgan or Goldman Sachs, whose main function was, and still is in some cases, but less each time, to efficiently manage the balance of the spread.

Spread that, of course, favors the house.

The real traders have to fight against that spread, as well as against the market, which makes things much more complicated.

From the point of view of the real traders, this automated trading is not very good, because it means that the market makers will become more efficient when making a more favorable spread for the houses.

Honestly, this artificial trading is something that I do not like much, mostly because in the end everything has a price, and radical automation, which is what we are witnessing today in human society, is nothing other than death of what is genuine and unique; and that is something that, obviously, is not very good.

I do not know if I explain.

See you on the flip side ... September.

Have a wonderful summer.

Best Regards,

Deo Talaverano.

Chief Market Strategist DHF.

George Town. Cayman Islands.