this article is the result of my studies of the effects of HFT, or High Frequency Trading, on price (you can review my articles on HFT on this blog and the research paper on HFT & Program Trading published for SIAT/IFTA on Feb 2013), combined with the knowledge of how program trading algorithms enter the markets on the smaller timeframes.
If you read my personal profile, you will now that I am interested in and enjoy research, but being very practical, the objective is to add to my trading techniques and bottom line, and those of other people.
In fact, while I cannot deny the satisfaction I gain in exercising my intellectual curiosity and interests, I mainly study trading, algorithms, techniques, methods, psychology because I want to find new patterns and information I can use to improve my trading entries, executions and continuously update my view of what is really going on in the market (.. and these days there is really a lot going on!). Indeed, this blog and article is not about me, but how YOU can exploit Program Trading to make money in the market, and how you avoid, and/or learn to read HFT.
What is High Frequency Trading or HFT? For those who do not know exactly what this class of trading algorithms is all about, I will offer one among commonly accepted definitions. There are several varieties of HFT algorithms and also the definitions vary quite a lot, however the following attributes are common to all the types of HFT algorithms:
- market positions are acquired and removed from the market very quickly, often hundreds or even thousands of time during a trading day
- orders transmission to the electronic exchanges take place with a very small latency (which requires dedicated computer appliances directly connected to the exchanges)
- algorithms manage a very small inventory of stocks, futures, lots or, generally, financial instruments
- every positions is typically closed in the space of few minutes or seconds
- rarely the position is carried overnight
- orders to trades ratio is very high, e.g. a very high number of orders is commonly canceled compared to the number of executed trades
In May 2013 I have presented a work on the effects of HFT on the markets and Program Trading (note: studying the effects of Program Trading on price and acting on the related observations is my trading edge) for SIAT/IFTA at the largest Investment and Trading event in Italy, the IT Forum of Rimini. In that occasion I have touched, among other things, on the way HFT negetively affects market structure of price. My speech on HFT and its dangers has been covered by “Il Sole 24 ore”, the most influential financial journal in Italy.
In the present article, on the other hand, I want to touch on another important effect that HFT algorithms have on markets and, specifically, how HFT represents an obstacle to other classes of algorithms like, for instance, Program Trading on small timeframes (below 240min). Maybe you already start seeing the pattern here, are you? Yes, it is no more human versus machine, but rather machines trading, and fighting for profits, against other machines.
While this topic is certainly fascinating and intriguing, it should also suggest how dangerous and subtle HFT can be (and generally is!), and this is why there is so much attention around it. HFT is even more dangerous when considered in light of the potential evolutions foreseen for this and other classes of algorithms. In fact, cheap and cheaper technology and faster computers are enabling adaptive automated trading methodologies with future generations of algorithms expected to be much more difficult to understand, explain, predict and control.
Leaving these worrisome considerations aside for a moment, I would like to show a recent example of how HFT impeded or obstacled an otherwise well formed setup from Program Trading algorithms on the 15mins (and lower timeframes). If you want to learn more, you can search about Program Trading on this blog and also watch the videos in the sections: forex, futures and stocks & ETFs on this blog. What I am about to show is a classic example of a price move caused by HFT (plus emotional participation) that had the same effect of going to ‘fish for stops’. However, as you will discover, stops are not the typical stops found below recent lows but, in the case in discussion, stops are related to a well formed setup created by Program Trading on the 15min timeframe.
The below picture (click on the picture to enlarge) shows a first traditional measured move modeling Program Trading activity on the S&P500 emini 15min chart. The Fibonacci trace is from A to B, and the entry setup at around 1,770 (C area), is respected to the pip. Price then moved higher and then again retraced towards the setup area but remained within the setup area (above the red line in D at 1,766.25).
After printing a new low in D, price remained above 1,766.25 and started moving higher. It is then when the “hell” came in, as soon as the FOMC minutes and news hit the wires and subsequently the markets (and you will see similar price dynamics for every type of planned and unplanned news event), there was a very quick spike lower. Everything happened in the space of one minute or so, as I will be clear when we will study the other chart, for the 1min timeframe below.
During that one minute or so, price has been pushed below the 1,766.25 stop level, violating the setup of Program Trading in the 15min timeframe. Before we get into the smaller timeframe let us keep studying the 15min chart above. The fist reaction to the FED has been a spike that caused prices to move marginally under 1,766.25 stopping Program Trading out, only on the 15min timeframe. Notice that Program Trading is active on all timeframes. After that, the orders supporting the “real“ and “logical“ direction of the market after FED announcement, i.e. tapering resulting in a reduction in $10B in monthly US bonds buyback, started flowing into the market.
Notice that I purportedly used quotes for the words real and logical, because what makes sense and appears the logical way to go is not always the best way to go, or the way price chooses in that specific market juncture. In the years I have seen hundreds of time the market moving in the direction opposite to what logic or common sense would have suggested. If you read my eBook “Key Concepts to Correct Trading Behavior” you will learn how strongly I discourage, new and experienced traders, in trading the news.
The order flow that set in after causing some troubles to Program Trading (by piercing the 1,766.25 stop in area D) pushed price higher and, not surprisingly pushed prices into first and second targets of the original ABC setup, in area C (with a long above 1,770, a stop below 1,766.25). The 1st target was at 1,793.75 and the second right at highs at 1,806. Both of them were hit. As we can see on the 15min chart above, although HFT algorithms managed to “break” the Program Trading 15min setup, when price was pushed higher Program Trading regained control. The proof is in the fact that market halted right at that 1,806 high and 2nd target, where Program Trading started taking profits out, thus stopping the move higher in an environment where HFT algorithms had faded out already.
Who wins? HFT always wins in the short term on the short timeframes. But once Program Trading,regains controls, in absence of news and HFT algorithms, it becames again the force that drives markets, in all timeframes. Before moving on to showing the exact levels of price and time when Program Trading regained control on the S&P500 emini, I want to elaborate a bit on where the S&P500 emini could be moving in the coming day (I took me 2 days to think and prepare this long, but hope interesting article and S&p500 has already made its move and it is now trading in extensions on the 15min chart, very aggressively!).
After touching highs at 1,806 there were two options for the S&P500 emini on the 15min chart above. The first one to retrace into the following setup in the G area, traced from highs in B to highs in F, or a so called extension long. So in this case Program Trading (and traders that use the Program Trading modeling edge) would enter this market at around 1,796.50 with a stop below 1,794. This is the next setup long with targets in the high 1,800s and above.
Those traders and analysists, like the author, who follow Program Trading footprints on price of the S&P500 emini know that this market likes to trade in traditional longs. So option two that must be considered could be, after piercing the 1793.75 level on the downside, a move of price into the 1,784 area H, for a bounce and a continuation higher. Knowing the main rules Program Trading uses on S&P500 emini futures can also help in objectively establishing when the possibility is there and probability is high for a complete reversal of the price move generated by the FED during the FOMC day (I saw that happening several times, too). In the case of the example being presented, that failure level to watch is the 1,778 level (area I).
In the following I want to comment and try to show you the other face of the coin. What Program Trading did after the 15min setup failed, due to HFT pushing prices below that 1,766.25 stop level? We cannot say if Program Trading on the 15min has attempted to trade the short side, after that failure. In fact, very often Program Trading is not active on certain timeframes during planned news announcements, to exactly avoid tricks by HFT and emotional behavior (there are also other times of the day and days of the week when Program Trading is characterized by a very low activity, or is deactivated, even on larger timeframes).
But on the 1min timeframe (click on the below chart to enlarge) it can be shown that other instances of Program Trading have tried to regain control of the markets, as soon as they could. After the first trust higher into the high in B (please refer to the below 1min chart), we witnessed profit taking into C and some programs apparently tried to pick price there. We could only observe it, after the fact, as programs used the anchor around the A area and traced to highs in B to obtain the new setup in the C area (entry above 1,780.75 with a stop below 1,778.75). The setup eventually got into the first target at 1,792.75 in the Cx area, for a 12 points gain over a 2.5 points risk, almost a R/R of 5. Again, there is no way this setup could be traded because the anchor A was only known after the fact.
The disciplined trader using the Program Trading footprints edge would have then traced from lows in C, previous setup area, to highs of the swing move in area E. That would have produced a setup area well below the F area, where price stopped and reversed and where the next long setup actually too place.
The fact that price stopped before, around the F area, respecting a long setup that would have been identified by using an anchor in D (and not C), i.e. tracing from D to highs in E, shows that Program Trading had to change anchor and means that my interpretation is that it did not yet regained control on the market.
In fact, again there was no way to anticipate that algorithms would have used D instead of C and where exactly, the actual setup would materialize. Had we had a way to trade the setup in F (entry at 1,790.75, stop below 1,789.25), we would have witnessed our trade to successfully hit the first target at 1,798.50 (Fx area), again another 5 R/R trade. An again I want to stress that there was no way to trade that observing Program Trading activity.
But at the next setup things start changing, Program Trading clearly regains control over the market on the 1min timeframe (Note: Program Trading works well on high volume markets and the 1min timeframe is a very high volume chart during an FOMC day, which makes it great for trading after HFT fades out, but it is also an ultra-fast market to trade, which makes us more prone to mistakes). How do we know Program Trading is now in control?
Likewise before, let us observe where the next setup is projected by Program Trading by tracing from F to G (new highs), which gives us the entry at H. This setup (entry at 1,795.75 and stop below 1,794), could now be traded, because price actually retraced into it. This would have allowed both Program Trading and professional traders to take a position at those levels for a ride into the first target at 1,803 (Hx area) for a 7.25 points gain.
I have represented the arrow pointing at the long setup in the area indicated with H in green color, symbolizing Program Trading regaining control on the market. On the contrary, the other arrows pointing to the previous long setup areas in C and F are colored in yellow/orange, symbolizing the class of Program Trading not yet back in control of the S&P500 emini futures market.
The following setup long in the sequence should now be evident and its presence should not surprise the reader. After hitting the target in the Hx area and printing a new high in correspondence to the area tagged with I, price retraced into the new long setup, indicated by the area L. This long setup could be identified in advance by tracing the Fibonacci study from H to I. By looking right well before price got into the L area, the trader could have been inputted a limit order for a long entry at 1,800.75 with a stop below 1,799, or a 2 points market stop order, and reap a 5.75 points gain (almost an R/R of 3) at 1,806.50, in correspondence of the target area indicated by Lx.
Before I close this long, and I hope also interesting post, I want to briefly discuss two points:
- what can we take away out of this long discussion?
- how do we actively and practically use this information as traders?
As per point 1) above, the recent example that presented on December 18th, 2013 shows how disturbing an effect HFT algorithms can have on price. It has been proven that, not only HFT changes market structure, but it can be driven to be of obstacle to Program Trading on smaller timeframes. In the above example I showed how HFT can provoke a failure in an otherwise well formed Program Trading setup. One of the characteristics of Program Trading is to help driving the natural flow of the markets and, differently from HFT which subtracts liquidity from the markets, Program Trading injects liquidity at certain points (which can be foreseen on all timeframes, similarly to what has been showed on the 1min chart above) thus contributing to maintain an high level of market efficiency or, more generally, an “healthy” price structure.
Such price structure somehow adheres to or is produced by what I call the “common psychology of the market”, or more precisely the psychology of market traders taken as a group. In fact, the overall behavior of the average trader who brings his/her own “every day” psychology into the game, produces the same effects on price than those enforced by Program Trading.
This is one of the reasons very few know or speak about Program Trading. This powerful class of algorithms results “invisible” to the majority, due to the fact that it works along and accompanies the natural flow of the markets, the flow implied by the traders group psychology. I showed that during important planned news event, like the FOMC meeting on December 18th, 2013, HFT programs can produce spikes that temporarily produce a failure in a sequence of price measured moves under the control of Program Trading.
I have also showed how Program Trading, which controls markets characterized by high volume (e.g. major indexes, forex majors, highly traded commodities, etc.), can regain control over price after news (and HFT algorithms) fade. The evidence has been showed on a smaller timeframe (1min), but that actually happens regularly on all timeframes. HFT algorithms can have negative impacts on Program Trading setups, but only temporarily. Program Trading is the real force in the markets, which is continuosly being applied to them. While HFT affects Program Trading temporarily on the smaller timeframes, news events related to market fundamentals, e.g. FED decision or economic downturn with effects on the USD Dollar and forex majors or significant geopolitical changes (e.g. China and Japan trading directly with one another without using the US Dollar reserve currency), can affect price on the larger timeframes (daily and weekly). However and whenever that happens, Program Trading will quickly adapt and then move forward into the driving seat again. Therefore, I believe that learning to spot Algorithmic Trading and, especially Program Trading footprints on price, represents probably the best edge a retail trader can have in modern markets, where 80% of volume is traded by machines and algorithms.
As per point 2) above: how do we apply the awareness and knowledge presented in this article? Besides the importance of learning how to recognize algorithms footprints on price and understand what program trading is doing on the different timeframes, I am a proponent of learning to guard ourselves and our trading from HFT. This point is especially important for those traders who like day trading using very small timeframe. However if you want to get very good or stellar R/R even up to 100 (and this can surely be obtained on long currency pair stretches), you can and will need to time your swing setups
in small timeframes too. Thus even if you are a swing trader, you may not be immune to the effects of HFT and you should master anyway how to deal with this class of algorithms .
The best advice I can give to readers is to not trade during planned news events. In fact, we do not know how emotional traders and HFT algorithms are going to interpret and react to the next piece of news. Neither we should assume that, if a 15min Program Trading setup is in place, the effects on price will necessarily be those showed in the above example. As mentioned, I have witnessed hundreds of news events where price went to the opposite direction of what logic and common sense would have suggested. Sometimes, maybe, just to reverse the very next day or during the subsequent days, and stat a move in the direction where common sense would have initially expected the price to go to.
If you are a daytrader and you are compelled to trade during news events (?), you can apply different strategies to deal with the presence of HFT. In the case of the above example, for instance, a trader could have reentered long after price moved above the highs of the one minute bar just before the red “incriminated” bar in the 1min chart I discussed above. The incriminated red bar is the one that broke the Program Trading setup identified on the 15min chart. The stop of the hypothetical trade would have been below recent lows on the 1min chart. This is not a strategy that I would necessarily prefer playing or recommend, because traders can still make lots of money in the markets even just taking setups “inspired” by Program Trading on the daily or 4 hour charts, as I show since April 2012 on this blog. But by executing this strategy, the trader might have enjoyed a R/R of 4 to 5, in a very volatile day, by just waiting for the end of the day.
What I actually hope you will take away from reading this article is that modern markets are much more complex and “crowded” than we would expect and, although coming from the right perspective price is not difficult to read and understand (note: spotting Program Trading footprints is not difficult, but rules and timing have to be learned), price cannot be studied and interpreted, as well as, effectively traded using traditional technical analysis or methods that do not take into account the overwhelming participation from algorithmic trading.
If you like this article, and you think it brings something new into the picture, please do not simply share it with friends, readers and fellow traders. Make it VIRAL! 🙂
Thank you (use the buttons below the article). Sharing is caring…
Notice that price projections in my method are not from/to random levels and do not follow Elliot Wave, DiNapoli levels or other methods, but rather a proprietary way to frame price based on modeling the effects of Program and Algorithmic Trading on price. My method helps “level plain” the trading game showing what banks, institutions and big hedge funds are actually doing (and not what they are saying they are doing). Follow me to learn how to spot the Algorithmic Trading footprints.
I send a free Newsletter in the weekend and provide updates throughout the week. The newsletter typically includes 3 video reviews for (1) EUR/USD, Dollar Index, S&P500 e-mini and Gold e-mini; (2) the Japanese Yen majors, i.e. USD/JPY, EUR/JPY and GBP/JPY; (3) the other majors: GBP/USD, AUD/USD, USD/CAD. Please, register here to receive the free weekly newsletter.
On the other hand, if you want to receive my videos and daily and weekly reviews, you can subscribe my free newsletter.
Should you decide to operate based on this information you are invited to do your own due diligence, consult a registered trading professional, as well as, understand the risks involved. This information is for educational purpose only. Please read the Disclaimer and accept all involved risks.
Have a great day.
Till the next post
Giuseppe Basile, ~the FibStalker