HFT Trading Analysis and Effects Observed
HFT Trading Analysis
Guys we would like to preface this piece by saying the following. Over the past two and a half weeks (and for the next several weeks), we have been testing various HFT / intraday trading algos on the Evolution 4.0 system. We have jokingly called it the "final exam" for the Evolution 4.0 system. However, without revealing too much as to how the Evolution 4.0 system works here is some information that we would like to share.
Firstly, for those that have following our recommendations in so far as 30/60/90 day "extended swing" trades, we will say... so far so great. Many of the recommendations such as $TASR, $RGR, $UWTI, $OLN, and even our index plays $SPXL and $UDOW have been performing largely as per our forecasts.
For those that have been in some of the metals plays such as $JNUG, $USLV, $NUGT and others, what we would like to say is this: Cost of doing business. We know that metals are breaking free of anglo western banking system manipulation and as such, there is going to be that level of volatility both to the upside and the downside. For those that have been in those plays for 30/60/90 day runs its going to be slightly less visible as compared to those intraday trading said ETF's. We will be the first to say that one of the by products of our intraday "30 day portfolio" self imposed challenge has been the ability to really study the movements of some of these financial instruments. We will be publishing detailed reviews and findings in this coming month's edition of Wirehaus on what we have discovered.
Along those lines, we would also like to remind everyone that reads our materials: The "30 day portfolio challenge" has been as much a test of various algorithms and programmatics we have for Evolution 4.0 as much as it has been a "final exam". In fact, given how far the Evolution 4.0 system has come, we wanted to ensure that the intra-day trading systems were "all systems go" prior to saying "thats it, we have a commercially viable financial forecasting system". As such, in no way are we encouraging those who follow our Twitter feed or others to "trade along with us" using the same trading style. We have basically been using a "kamakazi trading" approach on a simulated account to dial in various programatics on our system as well as add data to the learning algo component of our software.
As the slide at the beginning of our MarketUpdate pieces says: Please consider your own financial circumstances, risk tolerances, investment goals, etc prior to acting on any of the Evolution Consulting advice or information. We mean that. As such, we are happy to answer any questions that our readers financial professionals might have in regards to ideas that we share.
With that said:
One of the first things that we spoke about during our test of concept phase in the so called "alternative news" community was the potential dangers of HFT (High Frequency Trading) on modern capital markets. For those that are not familiar with the concept of HFT machines, we encourage you to do a bit of Googling. In the interest of keeping this piece to normal commentary length we are going to only mention some specific discoveries we have recently made.
Firstly in terms of HFT machines, there are effectively several, however the kind that we are going to speak about here are not the kind that "scalp" orders from the NYSE (or respective exchange's order books). In other words, machines that are designed to take advantage of slight imbalances / imperfections in equity pricing to make hundreds or thousands of trades in rapid succession. Instead we are going to talk about "learning algos" that trade the markets as a human trader would. In other words "front running" or using patterning / learnings algos to forecast and predict individual equity, sector, or index movements.
Once again, without going into excessive detail as to how the Evolution 4.0 system works, we have started out by modifying several existing algos that are loosely based upon the very same proprietary algos that are used by many of the biggest names on Wall Street. Our discovery: On no uncertain terms, they are obsolete. These are algos that were designed as recently as the mid 2000's. They simply are not fast enough to keep up with ETF/ETN offerings such as $JNUG, $NUGT and other ultra volatile ETF/ETN offerings (amongst others). What does this prove? Something that we have been saying since the very early days of our test of concept phase: The speed of information and the speed of transactionary business are the two greatest (and often unspoken) variables in modern capital markets, period.
While we will not disclose the names of the algos that we have used or the sources of said algos, what we will say is this: Many of even the "household names" on wall street are trading with algos that simply cannot keep up with the speed of information in todays modern capital markets. Once again, several of the ones we have tried thus far are either A: "out of the box" algo solutions, or B: "in-office modified" variants thereof. The first one or two that we tried (modified versions) had us up over 35 percent in a few days, however, they were then unable to keep up with rapidly changing market conditions combined with the ultra volatile nature of commodity based ETF and ETN offerings.
The second set we tried were "out of the box" solutions that are very, very similar to the algorithms used in major bank / Wirehaus trading machines. They too were simply too slow even with current dataset "knowledge". In fact, the "chasing the market" effect was almost laughable. Now these did work all right on single leverage ETF and individual equities but we are designing the Evolution 4.0 system well beyond what it needs to be given current trading / market conditions. Using the unmodified solutions (including a mis-install which was a mistake on our part we freely admit) we went from up 35% on our starting capital to down 30% inside of only a few days. In fact we have actually researched if there was intentional sabotage of said algorithms. In other words, it was that bad.
After completely removing our "opposite day" algos as we comically called them, we have returned to a base set that has been fairly consistent (with a few errors) in previous experiments. Fortunately we do have several "from the ground up" built algos that are unique and proprietary to the Evolution 4.0 and Evolution Ultra system and we will be installing those soon. (Actually in the process of installing them as this piece is being written).
However, without getting into any more detail there, what we will say is this: Will machines be directly responsible for "crashing the market" as many have said? Based upon our research, no. However, it must be said that machines (HFT of all varieties) do add a noticeable amount of volatility to individual equities, ETF/ETN's, and by default major indexes.
Keeping in mind that many of the algos that are being used by major banks today are programmed to operate within a "given normal" as it is often called, what happens when that given normal changes? In other words, what happens in the event of a retail level or institutional level "panic sell off". That is to say, what happens when market conditions are all of a sudden many deviations away from standard? Based upon our research with some of the "out of the box" solutions, the answer is a "magnifier effect". Basically take any volatility that would be associated with various economic events of major magnitude, then realize that HFT trading will exacerbate those movements by many orders of magnitude. How much exactly is difficult to say, though with the increasing popularity of HFT trading (both private and institutional level), we will say "significant". Even more significant than any effects that may have been seen in 2008 during the post sub prime mortgage meltdown. (Once again, due to the number of individuals and firms using said "black box" trading systems).
In the next several trading sessions we will be installing and testing a few of the all original algos that have been designed exclusively for Evolution 4.0 (more so for Evolution Ultra). Again, without going into detail, some of the major flaws that even many large banking / trading / financial institutions algos often contain, have been removed entirely from the ones now installed in the Evolution 4.0 system.
As the weather warms, we shall use a motorsports / auto racing analogy:
The 30 day portfolio challenge is as much about dialing in our "qualifying and race set ups" for Evolution 4.0 / Evolution Ultra, as much as it is about simply doubling a portfolio. In addition to being handy for in-office trading of Evolution Consulting funds, the HFT / High speed trading algos will also help contribute to extended swing trading as well as overall economic forecasting information that we are able to share. Given current market conditions, we have become tremendous proponents of the 30/60/90 day approach to trading capital markets. We once again remind our readers that our intra day trading is purely experimental and in no way encourage every investor to do so. Those who are experienced with intraday trading / high speed trading are welcome however to follow along on this page as well as on our Twitter account.
We would also like to say a very sincere thank you to those who have been subscribing to our materials, listening to our podcast, as well as watching our new MarketUpdate segments on YouTube. As we so often say, being born into this world is not an option, though prospering in it is. We salute the early adaptors who choose to prosper rather than complain.
