Measuring Toxic Flow for Trading & Risk Management

A common theme of microstructure modeling is that trade flow is often predictive of market direction.  One concept in particular that has gained traction is flow toxicity, i.e. flow where resting orders tend to be filled more quickly than expected, while aggressive orders rarely get filled at all, due to the participation of informed traders trading against uninformed traders.  The fundamental insight from microstructure research is that the order arrival process is informative of subsequent price moves in general and toxic flow in particular.  This is turn has led researchers to try to measure the probability of informed trading  (PIN).  One recent attempt to model flow toxicity, the Volume-Synchronized Probability of Informed Trading (VPIN)metric, seeks to estimate PIN based on volume imbalance and trade intensity.  A major advantage of this approach is that it does not require the estimation of unobservable parameters and, additionally, updating VPIN in trade time rather than clock time improves its predictive power.  VPIN has potential applications both in high frequency trading strategies, but also in risk management, since highly toxic flow is likely to lead to the withdrawal of liquidity providers, setting up the conditions for a flash-crash” type of market breakdown.

The procedure for estimating VPIN is as follows.  We begin by grouping sequential trades into equal volume buckets of size V.  If the last trade needed to complete a bucket was for a size greater than needed, the excess size is given to the next bucket.  Then we classify trades within each bucket into two volume groups:  Buys (V(t)B) and Sells (V(t)S), with V = V(t)B + V(t)S
The Volume-Synchronized Probability of Informed Trading is then derived as:

risk management

Typically one might choose to estimate VPIN using a moving average over n buckets, with n being in the range of 50 to 100.

Another related statistic of interest is the single-period signed VPIN. This will take a value of between -1 and =1, depending on the proportion of buying to selling during a single period t.

Toxic Flow

Fig 1. Single-Period Signed VPIN for the ES Futures Contract

It turns out that quote revisions condition strongly on the signed VPIN. For example, in tests of the ES futures contract, we found that the change in the midprice from one volume bucket the next  was highly correlated to the prior bucket’s signed VPIN, with a coefficient of 0.5.  In other words, market participants offering liquidity will adjust their quotes in a way that directly reflects the direction and intensity of toxic flow, which is perhaps hardly surprising.

Of greater interest is the finding that there is a small but statistically significant dependency of price changes, as measured by first buy (sell) trade price to last sell (buy) trade price, on the prior period’s signed VPIN.  The correlation is positive, meaning that strongly toxic flow in one direction has a tendency  to push prices in the same direction during the subsequent period. Moreover, the single period signed VPIN turns out to be somewhat predictable, since its autocorrelations are statistically significant at two or more lags.  A simple linear auto-regression ARMMA(2,1) model produces an R-square of around 7%, which is small, but statistically significant.

A more useful model, however , can be constructed by introducing the idea of Markov states and allowing the regression model to assume different parameter values (and error variances) in each state.  In the Markov-state framework, the system transitions from one state to another with conditional probabilities that are estimated in the model.

SSALGOTRADING AD

An example of such a model  for the signed VPIN in ES is shown below. Note that the model R-square is over 27%, around 4x larger than for a standard linear ARMA model.

We can describe the regime-switching model in the following terms.  In the regime 1 state  the model has two significant autoregressive terms and one significant moving average term (ARMA(2,1)).  The AR1 term is large and positive, suggesting that trends in VPIN tend to be reinforced from one period to the next. In other words, this is a momentum state. In the regime 2 state the AR2 term is not significant and the AR1 term is large and negative, suggesting that changes in VPIN in one period tend to be reversed in the following period, i.e. this is a mean-reversion state.

The state transition probabilities indicate that the system is in mean-reversion mode for the majority of the time, approximately around 2 periods out of 3.  During these periods, excessive flow in one direction during one period tends to be corrected in the
ensuring period.  But in the less frequently occurring state 1, excess flow in one direction tends to produce even more flow in the same direction in the following period.  This first state, then, may be regarded as the regime characterized by toxic flow.

Markov State Regime-Switching Model

Markov Transition Probabilities

P(.|1)       P(.|2)

P(1|.)        0.54916      0.27782

P(2|.)       0.45084      0.7221

Regime 1:

AR1           1.35502    0.02657   50.998        0

AR2         -0.33687    0.02354   -14.311        0

MA1          0.83662    0.01679   49.828        0

Error Variance^(1/2)           0.36294     0.0058

Regime 2:

AR1      -0.68268    0.08479    -8.051        0

AR2       0.00548    0.01854    0.296    0.767

MA1     -0.70513    0.08436    -8.359        0

Error Variance^(1/2)           0.42281     0.0016

Log Likelihood = -33390.6

Schwarz Criterion = -33445.7

Hannan-Quinn Criterion = -33414.6

Akaike Criterion = -33400.6

Sum of Squares = 8955.38

R-Squared =  0.2753

R-Bar-Squared =  0.2752

Residual SD =  0.3847

Residual Skewness = -0.0194

Residual Kurtosis =  2.5332

Jarque-Bera Test = 553.472     {0}

Box-Pierce (residuals):         Q(9) = 13.9395 {0.124}

Box-Pierce (squared residuals): Q(12) = 743.161     {0}

 

A Simple Trading Strategy

One way to try to monetize the predictability of the VPIN model is to use the forecasts to take directional positions in the ES
contract.  In this simple simulation we assume that we enter a long (short) position at the first buy (sell) price if the forecast VPIN exceeds some threshold value 0.1  (-0.1).  The simulation assumes that we exit the position at the end of the current volume bucket, at the last sell (buy) trade price in the bucket.

This simple strategy made 1024 trades over a 5-day period from 8/8 to 8/14, 90% of which were profitable, for a total of $7,675 – i.e. around ½ tick per trade.

The simulation is, of course, unrealistically simplistic, but it does give an indication of the prospects for  more realistic version of the strategy in which, for example, we might rest an order on one side of the book, depending on our VPIN forecast.

informed trading

Figure 2 – Cumulative Trade PL

References

Easley, D., Lopez de Prado, M., O’Hara, M., Flow Toxicity and Volatility in a High frequency World, Johnson School Research paper Series # 09-2011, 2011

Easley, D. and M. O‟Hara (1987), “Price, Trade Size, and Information in Securities Markets”, Journal of Financial Economics, 19.

Easley, D. and M. O‟Hara (1992a), “Adverse Selection and Large Trade Volume: The Implications for Market Efficiency”,
Journal of Financial and Quantitative Analysis, 27(2), June, 185-208.

Easley, D. and M. O‟Hara (1992b), “Time and the process of security price adjustment”, Journal of Finance, 47, 576-605.

 

Hiring High Frequency Quant/Traders

I am hiring in Chicago for exceptional HF Quant/Traders in Equities, F/X, Futures & Fixed Income.  Remuneration for these roles, which will be dependent on qualifications and experience, will be in line with the highest market levels.

Role
Working closely with team members including developers, traders and quantitative researchers, the central focus of the role will be to research and develop high frequency trading strategies in equities, fixed income, foreign exchange and related commodities markets.

Responsibilities
The analyst will have responsibility of taking an idea from initial conception through research, testing and implementation.  The work will entail:

  • Formulation of mathematical and econometric models for market microstructure
  • Data collation, normalization and analysis
  • Model prototyping and programming
  • Strategy development, simulation, back-testing and implementation
  • Execution strategy & algorithms

Qualifications & Experience

  • Minimum 5 years in quantitative research with a leading proprietary trading firm, hedge fund, or investment bank
  • In-depth knowledge of Equities, F/X and/or futures markets, products and operational infrastructure
  • High frequency data management & data mining techniques
  • Microstructure modeling
  • High frequency econometrics (cointegration, VAR,error correction models, GARCH, panel data models, etc.)
  • Machine learning, signal processing, state space modeling and pattern recognition
  • Trade execution and algorithmic trading
  • PhD in Physics/Math/Engineering, Finance/Economics/Statistics
  • Expert programming skills in Java, Matlab/Mathematica essential
  • Must be US Citizen or Permanent Resident

Send your resume to: jkinlay at systematic-strategies.com.

No recruiters please.

Alpha Spectral Analysis

One of the questions of interest is the optimal sampling frequency to use for extracting the alpha signal from an alpha generation function.  We can use Fourier transforms to help identify the cyclical behavior of the strategy alpha and hence determine the best time-frames for sampling and trading.  Typically, these spectral analysis techniques will highlight several different cycle lengths where the alpha signal is strongest.

The spectral density of the combined alpha signals across twelve pairs of stocks is shown in Fig. 1 below.  It is clear that the strongest signals occur in the shorter frequencies with cycles of up to several hundred seconds. Focusing on the density within
this time frame, we can identify in Fig. 2 several frequency cycles where the alpha signal appears strongest. These are around 50, 80, 160, 190, and 230 seconds.  The cycle with the strongest signal appears to be around 228 secs, as illustrated in Fig. 3.  The signals at cycles of 54 & 80 (Fig. 4), and 158 & 185/195 (Fig. 5) secs appear to be of approximately equal strength.
There is some variation in the individual pattern for of the power spectra for each pair, but the findings are broadly comparable, and indicate that strategies should be designed for sampling frequencies at around these time intervals.

power spectrum

Fig. 1 Alpha Power Spectrum

 

power spectrum

Fig.2

power spectrumFig. 3

power spectrumFig. 4

power spectrumFig. 5

PRINCIPAL COMPONENTS ANALYSIS OF ALPHA POWER SPECTRUM
If we look at the correlation surface of the power spectra of the twelve pairs some clear patterns emerge (see Fig 6):

spectral analysisFig. 6

Focusing on the off-diagonal elements, it is clear that the power spectrum of each pair is perfectly correlated with the power spectrum of its conjugate.   So, for instance the power spectrum of the Stock1-Stock3 pair is exactly correlated with the spectrum for its converse, Stock3-Stock1.

SSALGOTRADING AD

But it is also clear that there are many other significant correlations between non-conjugate pairs.  For example, the correlation between the power spectra for Stock1-Stock2 vs Stock2-Stock3 is 0.72, while the correlation of the power spectra of Stock1-Stock2 and Stock2-Stock4 is 0.69.

We can further analyze the alpha power spectrum using PCA to expose the underlying factor structure.  As shown in Fig. 7, the first two principal components account for around 87% of the variance in the alpha power spectrum, and the first four components account for over 98% of the total variation.

PCA Analysis of Power Spectra
PCA Analysis of Power Spectra

Fig. 7

Stock3 dominates PC-1 with loadings of 0.52 for Stock3-Stock4, 0.64 for Stock3-Stock2, 0.29 for Stock1-Stock3 and 0.26 for Stock4-Stock3.  Stock3 is also highly influential in PC-2 with loadings of -0.64 for Stock3-Stock4 and 0.67 for Stock3-Stock2 and again in PC-3 with a loading of -0.60 for Stock3-Stock1.  Stock4 plays a major role in the makeup of PC-3, with the highest loading of 0.74 for Stock4-Stock2.

spectral analysis

Fig. 8  PCA Analysis of Power Spectra

Master’s in High Frequency Finance

I have been discussing with some potential academic partners the concept for a new graduate program in High Frequency Finance.  The idea is to take the concept of the Computational Finance program developed in the 1990s and update it to meet the needs of students in the 2010s.

The program will offer a thorough grounding in the modeling concepts, trading strategies and risk management procedures currently in use by leading investment banks, proprietary trading firms and hedge funds in US and international financial markets.  Students will also learn the necessary programming and systems design skills to enable them to make an effective contribution as quantitative analysts, traders, risk managers and developers.

I would be interested in feedback and suggestions as to the proposed content of the program.

The Information Content of the Pre- and Post-Market Trading Sessions

I apologize in advance for this rather “wonkish” post, which is aimed chiefly at the high frequency fraternity, or those at least who trade intra-day, in the equity markets.  Such minutiae are the lot of those engaged in high frequency trading.  I promise that my next post will be of more general interest.

Pre- and Post Market Sessions

The pre-market session in US equities runs from 8:00 AM ET, while the post-market session runs until 8:00 PM ET.  The question arises whether these sessions are worth trading, or at the very least, offer a source of data (quotes, trades) that might be relevant to trading the regular session, which of course runs from 9:30 AM to 4:00 PM ET.  Even if liquidity is thin and trades infrequent, and opportunities in the pre- and post-market very limited, it might be that we can improve our trading models by taking into account such information as these sessions do provide, even if we only ever plan to trade during regular trading hours.

It is somewhat challenging to discuss this in great detail, because HFT equity trading is very much in the core competencies of my firm, Systematic Strategies.  However, I hope to offer some ideas, at least, that some readers may find useful.

SSALGOTRADING AD

 

A Tale of Two Pharmaceutical Stocks

In what follows I am going to make use of two examples from the pharmaceutical industry: Alexion Pharmaceuticals, Inc. (ALXN), which has a market cap of $35Bn and trades around 800,000 shares daily, and Pfizer Inc. (PFE), which has a market cap of over $200Bn and trades close to 50M shares a day.

Let’s start by looking at a system trading ALXN during regular market hours.  The system isn’t high frequency, but trades around 1-2 times a day, on average.  The strategy equity curve from 2015 to April 2016 is not at all impressive.

 

ALXN Regular

ALXN – Regular Session Only

 

But look at the equity curve for the same strategy when we allow it to run on the pre- and post-market sessions, in addition to regular trading hours.  Clearly the change in the trading hours utilized by the strategy has made a huge improvement in the total gain and risk-adjusted returns.

 

ALEXN with pre-market

ALXN – with Pre- and Post-Market Sessions

 

The PFE system trades much more frequently, around 4 times a day, but the story is somewhat similar in terms of how including the pre- and post-market sessions appears to improve its performance.

PFE Regular

PFE – Regular Session Only

PFE with premarket

PFE – with Pre- and Post-Market Sessions

 

Improving Trading Performance

In both cases, clearly, the trading performance of the strategies has improved significantly with the inclusion of the out-of-hours sessions.  In the case of ALXN, we see a modest increase of around 10% in the total number of trades, but in the case of PFE the increase in trading activity is much more marked – around 30%, or more.

The first important question to ask is when these additional trades are occurring.  Assuming that most of them take place during the pre- or post-market, our concern might be whether there is likely to be sufficient liquidity to facilitate trades of the frequency and size we wish to execute.  Of various possible hypotheses, some negative, other positive, we might consider the following:

(a) Bad ticks in the market data feed during out-of-hours sessions give rise to apparently highly profitable “phantom” trades

(b) The market data is valid, but the trades are done in such low volume as to be insignificant for practical purposes (i.e. trades were done for a few hundred lots and additional liquidity is unlikely to be available)

(c) Out-of-hours sessions enable the system to improve profitability by entering or exiting positions in a more timely manner than by trading the regular session alone

(d) Out-of-hours market data improves the accuracy of model forecasts, facilitating a larger number of trades, and/or more profitable trades, during regular market hours

An analysis of the trading activity for the two systems provides important insight as to which of the possible explanations might be correct.


ALXN Analysis

(click to enlarge)

Dealing first with ALXN, we that, indeed, an additional 11% of trades are entered or exited out-of-hours.  However, these additional trades account for somewhere between 17% (on exit) and 20% (on entry) of the total profits.  Furthermore, the size of the average entry trade during the post-market session and of the average exit trade in the pre-market session is more than double that of the average trade entered or exited during regular market hours. That gives concerns that some of the apparent increase in profits may be due to bad ticks at prices away from the market, allowing the system enter or exit trades at unrealistically low or high prices.  Even if many of the trades are good, we will have concerns about the scalability of the strategy in out-of-hours trading, given the relatively poor liquidity in the stock. On the other hand, at least some of the uplift in profits arises from new trades occurring during the regular session. This suggests that, even if we are unable to execute many of the trading opportunities seen during pre- or post-market, the trades from those sessions provides useful additional data points for our model, enabling it to increase the number and/or profitability of trades in the regular session.

Next we turn to PFE.  We can see straight away that, while the proportion of trades occurring during out-of-hours sessions is around 23%, those trades now account for over 50% of the total profits.  Furthermore, the average PL for trades executed on entry post-market, and on exit pre-market, is more than 4x the average for trades entered or exited during normal market hours.  Despite the much better liquidity in PFE compared to ALXN, this is a huge concern – we might expect to see significant discrepancies occurring between theoretical and actual performance of the strategy, due to the very high dependency on out-of-hours trading.

PFE Analysis

(click to enlarge)

As we dig further into the analysis, we do indeed find evidence that bad data ticks play a disproportionate role.  For example, this trade in PFE which apparently occurred at around 16:10 on 4/6 was almost certainly a phantom trade resulting from a bad data point. It turns out that, for whatever reason, such bad ticks are a common occurrence in the stock and account for a large proportion of the apparent profitability of out-of-hours trading in PFE.

 

PFE trade

 

CONCLUSION

We are, of course, only skimming the surface of the analysis that is typically carried out.  One would want to dig more deeply into ways in which the market data feed could be cleaned up and bad data ticks filtered out so as to generate fewer phantom trades.  One would also want to look at liquidity across the various venues where the stocks trade, including dark pools, in order to appraise the scalability of the strategies.

For now, the main message that I am seeking to communicate is that it is often well worthwhile considering trading in the pre- and post-market sessions, not only with a view to generating additional, profitable trading opportunities, but also to gather additional data points that can enhance trading profitability during regular market hours.

High Frequency Trading: Equities vs. Futures

A talented young system developer I know recently reached out to me with an interesting-looking equity curve for a high frequency strategy he had designed in E-mini futures:

Fig1

Pretty obviously, he had been making creative use of the “money management” techniques so beloved by futures systems designers.  I invited him to consider how it would feel to be trading a 1,000-lot E-mini position when the market took a 20 point dive.  A $100,000 intra-day drawdown might make the strategy look a little less appealing.  On the other hand, if you had already made millions of dollars in the strategy, you might no longer care so much.

SSALGOTRADING AD

A more important criticism of money management techniques is that they are typically highly path-dependent:  if you had started your strategy slightly closer to one of the drawdown periods that are almost unnoticeable on the chart, it could have catastrophic consequences for your trading account.  The only way to properly evaluate this, I advised, was to backtest the strategy over many hundreds of thousands of test-runs using Monte Carlo simulation.  That would reveal all too clearly that the risk of ruin was far larger than might appear from a single backtest.

Next, I asked him whether the strategy was entering and exiting passively, by posting bids and offers, or aggressively, by crossing the spread to sell at the bid and buy at the offer.  I had a pretty good idea what his answer would be, given the volume of trades in the strategy and, sure enough he confirmed the strategy was using passive entries and exits.  Leaving to one side the challenge of executing a trade for 1,000 contracts in this way, I instead ask him to show me the equity curve for a single contract in the underlying strategy, without the money-management enhancement. It was still very impressive.

Fig2

 

The Critical Fill Assumptions For Passive Strategies

But there is an underlying assumption built into these results, one that I have written about in previous posts: the fill rate.  Typically in a retail trading platform like Tradestation the assumption is made that your orders will be filled if a trade occurs at the limit price at which the system is attempting to execute.  This default assumption of a 100% fill rate is highly unrealistic.  The system’s orders have to compete for priority in the limit order book with the orders of many thousands of other traders, including HFT firms who are likely to beat you to the punch every time.  As a consequence, the actual fill rate is likely to be much lower: 10% to 20%, if you are lucky.  And many of those fills will be “toxic”:  buy orders will be the last to be filled just before the market  moves lower and sell orders will be the last to get filled just as the market moves higher. As a result, the actual performance of the strategy will be a very long way from the pretty picture shown in the chart of the hypothetical equity curve.

One way to get a handle on the problem is to make a much more conservative assumption, that your limit orders will only get filled when the market moves through them.  This can easily be achieved in a product like Tradestation by selecting the appropriate backtest option:

fig3

 

The strategy performance results often look very different when this much more conservative fill assumption is applied.  The outcome for this system was not at all unusual:

Fig4

 

Of course, the more conservative assumption applied here is also unrealistic:  many of the trading system’s sell orders would be filled at the limit price, even if the market failed to move higher (or lower in the case of a buy order).  Furthermore, even if they were not filled during the bar-interval in which they were issued, many limit orders posted by the system would be filled in subsequent bars.  But the reality is likely to be much closer to the outcome assuming a conservative fill-assumption than an optimistic one.    Put another way:  if the strategy demonstrates good performance under both pessimistic and optimistic fill assumptions there is a reasonable chance that it will perform well in practice, other considerations aside.

An Example of a HFT Equity Strategy

Let’s contrast the futures strategy with an example of a similar HFT strategy in equities.  Under the optimistic fill assumption the equity curve looks as follows:

Fig5

Under the more conservative fill assumption, the equity curve is obviously worse, but the strategy continues to produce excellent returns.  In other words, even if the market moves against the system on every single order, trading higher after a sell order is filled, or lower after a buy order is filled, the strategy continues to make money.

Fig6

Market Microstructure

There is a fundamental reason for the discrepancy in the behavior of the two strategies under different fill scenarios, which relates to the very different microstructure of futures vs. equity markets.   In the case of the E-mini strategy the average trade might be, say, $50, which is equivalent to only 4 ticks (each tick is worth $12.50).  So the average trade: tick size ratio is around 4:1, at best.  In an equity strategy with similar average trade the tick size might be as little as 1 cent.  For a futures strategy, crossing the spread to enter or exit a trade more than a handful of times (or missing several limit order entries or exits) will quickly eviscerate the profitability of the system.  A HFT system in equities, by contrast, will typically prove more robust, because of the smaller tick size.

Of course, there are many other challenges to high frequency equity trading that futures do not suffer from, such as the multiplicity of trading destinations.  This means that, for instance, in a consolidated market data feed your system is likely to see trading opportunities that simply won’t arise in practice due to latency effects in the feed.  So the profitability of HFT equity strategies is often overstated, when measured using a consolidated feed.  Futures, which are traded on a single exchange, don’t suffer from such difficulties.  And there are a host of other differences in the microstructure of futures vs equity markets that the analyst must take account of.  But, all that understood, in general I would counsel that equities make an easier starting point for HFT system development, compared to futures.

Reflections on Careers in Quantitative Finance

CMU’s MSCF Program

Carnegie Mellon’s Steve Shreve is out with an interesting post on careers in quantitative finance, with his commentary on the changing landscape in quantitative research and the implications for financial education.

I taught at Carnegie Mellon in the late 1990’s, including its excellent Master’s program in quantitative finance that Steve co-founded, with Sanjay Srivastava.  The program was revolutionary in many ways and was immediately successful and rapidly copied by rival graduate schools (I help to spread the word a little, at Cambridge).

Fig1The core of the program remains largely unchanged over the last 20 years, featuring Steve’s excellent foundation course in stochastic calculus;  but I am happy to see that the school has added many, new and highly relevant topics to the second year syllabus, including market microstructure, machine learning, algorithmic trading and statistical arbitrage.  This has moved the program in terms of its primary focus, which was originally financial engineering, to include coverage of subjects that are highly relevant to quantitative investment research and trading.

It was this combination of sound theoretical grounding with practitioner-oriented training that made the program so successful.  As I recall, every single graduate was successful in finding a job on Wall Street, often at salaries in excess of $200,000, a considerable sum in those days.  One of the key features of the program was that it combined theoretical concepts with practical training, using a simulated trading floor gifted by Thomson Reuters (a model later adopted btrading-floor-1y the ICMA centre at the University of Reading in the UK).  This enabled us to test students’ understanding of what they had been taught, using market simulation models that relied upon key theoretical ideas covered in the program.  The constant reinforcement of the theoretical with the practical made for a much deeper learning experience for most students and greatly facilitated their transition to Wall Street.

Masters in High Frequency Finance

While CMU’s program has certainly evolved and remains highly relevant to the recruitment needs of Wall Street firms, I still believe there is an opportunity for a program focused exclusively on high frequency finance, as previously described in this post.  The MHFF program would be more computer science oriented, with less emphasis placed on financial engineering topics.  So, for instance, students would learn about trading hardware and infrastructure, the principles of efficient algorithm design, as well as HFT trading techniques such as order layering and priority management.  The program would also cover HFT strategies such as latency arbitrage, market making, and statistical arbitrage.  Students would learn both lower level (C++, Java) and higher level (Matlab, R) programming languages and there is  a good case for a mandatory machine code programming course also.  Other core courses might include stochastic calculus and market microstructure.

Who would run such a program?  The ideal school would have a reputation for excellent in both finance and computer science. CMU is an obvious candidate, as is MIT, but there are many other excellent possibilities.

Careers

I’ve been involved in quantitative finance since the beginning:  I recall programming one of the first 68000 Assembler microcomputers in the 1980s, which was ultimately used for an F/X system at a major UK bank. The ensuing rapid proliferation of quantitative techniques in finance has been fueled by the ubiquity of cheap computing power, facilitating the deployment of quantitate techniques that would previously been impractical to implement due to their complexity.  A good example is the machine learning techniques that now pervade large swathes of the finance arena, from credit scoring to HFT trading.  When I first began working in that field in the early 2000’s it was necessary to assemble a fairly sizable cluster of cpus to handle the computation load. These days you can access comparable levels of computational power on a single server and, if you need more, you can easily scale up via Azure or EC2.

fig3It is this explosive growth in computing power  that has driven the development of quantitative finance in both the financial engineering and quantitative investment disciplines. As the same time, the huge reduction in the cost of computing power has leveled the playing field and lowered barriers to entry.  What was once the exclusive preserve of the sell-side has now become readily available to many buy-side firms.  As a consequence, much of the growth in employment opportunities in quantitative finance over the last 20 years has been on the buy-side, with the arrival of quantitative hedge funds and proprietary trading firms, including my own, Systematic Strategies.  This trend has a long way to play out so that, when also taking into consideration the increasing restrictions that sell-side firms face in terms of their proprietary trading activity, I am inclined to believe that the buy-side will offer the best employment opportunities for quantitative financiers over the next decade.

It was often said that hedge fund managers are typically in their 30’s or 40’s when they make the move to the buy-side. That has changed in the last 15 years, again driven by the developments in technology.  These days you are more likely to find the critically important technical skills in younger candidates, in their late 20’s or early 30’s.  My advice to those looking for a career in quantitative finance, who are unable to find the right job opportunity, would be: do what every other young person in Silicon Valley is doing:  join a startup, or start one yourself.

 

Making Money with High Frequency Trading

There is no standard definition of high frequency trading, nor a single type of strategy associated with it. Some strategies generate returns, not by taking any kind of view on market direction, but simply by earning Exchange rebates. In other cases the strategy might try to trade ahead of the news as it flows through the market, from stock to stock (or market to market).  Perhaps the most common and successful approach to HFT is market making, where one tries to earn (some fraction of) the spread by constantly quoting both sides of the market.  In the latter approach, which involves processing vast numbers of order messages and other market data in order to decide whether to quote (or pull a quote), latency is of utmost importance.  I would tend to argue that HFT market making owes its success as much, or more, to computer science than it does to trading or microstructure theory.

By contrast, Systematic Strategies’s approach to HFT has always been model-driven.  We are unable to outgun firms like Citadel or Getco in terms of their speed of execution; so, instead, we focus on developing theoretical models of market behavior, on the assumption that we are more likely to identify a source of true competitive advantage that way.  This leads to slower, less latency-sensitive strategies (the models have to be re-estimated or recomputed in real time), but which may nonetheless trade hundreds of times a day.

A good example is provided by our high frequency scalping strategy in Corn futures, which trades around 100-200 times a day, with a win rate of over 80%.

Corn Monthly PNL EC

 

One of the most important considerations in engineering a HFT strategy of this kind is to identify a suitable bar frequency.  We find that our approach works best using data at frequencies of 1-5 minutes, trading at latencies of around 1 millisec, whereas other firms are reacting to data tick-by-tick, with latencies measured in microseconds.

Often strategies are built using only data derived from with a single market, based on indicators involving price action, pattern trading rules, volume or volatility signals.  In other cases, however, signals are derived from other, related markets: the VXX-ES-TY complex would be a typical example of this kind of inter-market approach.

SSALGOTRADING AD

When we build strategies we often start by using a simple retail platform like TradeStation or MultiCharts.  We know that if the strategy can make money on a platform with retail levels of order and market data latency (and commission rates), then it should perform well when we transfer it to a production environment, with much lower latencies and costs.  We might be able to trade only 1-2 contracts in TradeStation, but in production we might aim to scale that up to 10-15 contract per trade, or more, depending on liquidity.  For that reason we prefer to trade only intraday, when market liquidity is deepest; but we often find sufficient levels of liquidity to make trading worthwhile 1-2 hours before the open of the day session.

Generally, while we look for outside money for our lower frequency hedge fund strategies, we tend not to do so for our HFT strategies.  After all, what’s the point?  Each strategy has limited capacity and typically requires no more than a $100,000 account, at most.  And besides, with Sharpe Ratios that are typically in double-digits, it’s usually in our economic interest to use all of the capacity ourselves.  Nor do we tend to license strategies to other trading firms.  Again, why would we?  If the strategies work, we can earn far more from trading rather than licensing them.

We have, occasionally, developed strategies for other firms for markets in which we have no interest (the KOSPI springs to mind).  But these cases tend to be the exception, rather than the rule.