FX Quant and Positioning Weekly

Size: px

Start display at page:

Download "FX Quant and Positioning Weekly"

Merilyn Rice
6 years ago
Views:

1 November 11, 2013 FX Quant and Positioning Weekly Karl Steiner Dag Müller Anders Söderberg

2 Content Main conclusions. 3 Risk appetite index... 4 Speculative positioning and sentiment Table and summary Charts of percentile rank development Price based indicators T3 systematic trend model FX-o-meters: trend, stretch and volatility FX market environment Volatility 11 Correlation 12 Appendix Speculative position charts T3 performance Short-term fair values Volatility charts Risk appetite index and the factors.. 27 Explanations FX QUANT AND POSITIONING WEEKLY 2

3 Main conclusions Spec s cut excessive bullish EUR position by more than half Speculative positioning (based on CFTC s COT reports) Speculators slashed their previously excessively bullish EUR position by more than half. The change was the biggest since May 2011 and 4 th biggest ever and came as EMU inflation surprised on the downside sparking (which turned out to be correct) expectations of a cut by the ECB. As USD was favored over of all currencies except MXN the net USD position swung to a net long after spec s being short USD the past four weeks. The NOK speculative positioning proxy once again foresaw a sharp EUR/NOK rise. As we wrote in the report last week our speculative positioning proxy for NOK indicated a switch from short to (a small) long NOK positioning. This has now occurred seven times in 2013 and all those occasions have been followed by a rise in EUR/NOK. Short-term fair values The largest misvaluations versus short-term fair values are currently seen in EUR/SEK and USD/CAD. EUR/SEK closed at 1.8 standard deviations above its STFV at on Friday. Thus there is a significant downside near-term valuation pressure, though no correction signal was generated in our trading set up (signal on breaches of +/-2 standard deviations). However since September there have been five occasions were a correction has been triggered already on breaches of +/-1.8 standard deviations (see chart on p. 22). Besides EUR/SEK also USD/CAD is at 1.8 standard deviations above its STFV and should hence experience valuation pressure on the downside. Price based indicators FX-O-meters: No currency pairs are severely long-term stretched however EUR/GBP, EUR/NOK and USD/NOK render extra attention. Trends are still weak as measured by our trend-o-meter though volatility made a bit of a comeback last week. This is not an optimal environment for trend following strategies; rather contrarian or range trading strategies tend to work better in this sort of environment. T3 trend model: For this week our systematic setup recommends going long USD/NOK and USD/CAD. However due to the market environment, as mentioned above, we recommend underweighting trend strategies compared to a normal weighting in a portfolio. SEB Risk appetite index Fresh yearly high. RAI rose to a fresh yearly high at on Wednesday but then fell back slightly. RAI has been in a rising trend since mid-2012 where the initial move was quite sharp but recently it has been much more of a grind. However the risk neutral level of 100 is close a level not seen since July FX QUANT AND POSITIONING WEEKLY 3

4 SEB Risk Appetite Index Fresh yearly high on Wednesday RAI 1y Last week RAI rose to a fresh yearly high on Wednesday but then fell back Thursday as ECB surprisingly cut rates. Friday, following higher than expected NFP, it rose slightly again. Day Level High/Low Mon Tue Wed High Thu Low Fri RAI 4y RAI 20d Looking ahead The EMBI spread trend higher since around Oct 20. A turnaround in that trend is probably needed for RAI to make a sustainable attempt above the risk neutral 100 level. RAI has been in a rising trend since mid The initial move was sharply higher but has been much more of a grind lately. Still the lows are higher and the tops as well indicating that the long-term trend for higher RAI persist FX QUANT AND POSITIONING WEEKLY 4

5 Speculative positioning and sentiment for the period October 29- November 5 Main points Data released with normal lag (3 days) Speculators decreased their FX exposure slightly Net long EUR position more than slashed in half in the biggest change since May Aggregated USD position switched to net long after specs being short USD for four weeks. Net non commercial positions Percentile ranks Current Previous Change Current Change AUD 25,067 23,198 8% CAD 18,002 15,237 18% CHF 8,095 11,451 29% EUR 33,143 70,617 53% 82 0 GBP 2,392 10, % 63 4 JPY 73,792 62,395 18% MXN 5,180 4,896 6% NZD 9,708 10,625 9% Net USD 63,127 6, % th percentile 25-50th percentile 50-75th percentile th percentile Source: CFTC. Percentile ranks calculated using data one year back. Summary 4th biggest drop in EUR positioning ever and biggest since May 2011 as specs slashed their net long EUR position by more than half. Bear in mind that this was before the ECB cut! The change brought the percentile rank from 98 to 84 thus indicating that the excessively bullish position is at far more normal levels. Specs became bullish the USD. The positioning changes in all currencies except the MXN indicate favorable USD sentiment during the period. This resulted in a switch into an aggregated net long USD position after being short only four weeks. Normalization of the excessively bearish MXN positioning. Using our percentile rank measure MXN was the only currency with excessive bearish positioning (at the 6th percentile) in the period before this report. Accordingly MXN managed to receive an increase in its net long position even as the general sentiment was pro USD. All in all speculators decreased their FX exposure slightly. The total amount of FX contracts held by speculators decreased by 2% after increasing in the past two weeks. The biggest contributor to the decrease was EUR where the change in net position was mostly due to reduced long contracts FX QUANT AND POSITIONING WEEKLY 5

6 Speculative positions Percentile rank development: High beta currencies European currencies Safe haven currencies USD vs. EUR FX QUANT AND POSITIONING WEEKLY 6

7 T3 systematic trend model Long USD/NOK and long USD/CAD FX-o-meters sorted by strength of trend CCY Pair Trend Stretch Vol 20d rank 11-Nov M A(200d) USDN OK EURGBP USDCAD EURUSD USDSEK EUR N OK AUDUSD NOKSEK NZDUSD The three most extreme readings in each category is highlighted by being bold and colored: Trend Stretch Excess volatility * Explanation of the trend, stretch, vol and 20d rank on page 24 USD/NOK higher is the strongest trend (0.84 st dev) and as the stretch and excess volatility is below the model thresholds a recommendation is triggered. However the stretch (at 1.77 st dev) is quite large and should be monitored. USD/CAD has the third strongest trend (+0.66 st dev) and a stretch (1.3) well below the threshold at 2 and also a low excess volatility (-0.1). Thus a long recommendation is triggered for a second week in a row. EUR/GBP has the second strongest trend (-0.8 st dev) but the excess volatility is not low enough for a recommendation to be triggered (the stretch at -1.9 is also quite near the threshold). Recommended stop less levels: USD/NOK and USD/CAD EURSEK EURJP Y CCY Entry Stop loss EURP LN USD/NOK USDJP Y USD/CAD GB P USD Exit: On Friday at the ECB fix (14.15) EURCAD EURA UD EURNZD Average Bloomberg indicative quotes at 10:00 CET. Previous week: : Long USD/CAD gained 0.4 % ( ) while both the long NOK/SEK and long USD/JPY was stopped out: NOK/SEK -0.9% ( ) and USD/JPY -0.9% ( ). Return Year to date: -3.1% Since inception (2012): 13.1% FX QUANT AND POSITIONING WEEKLY 7

8 T3 systematic trend model Post and pre trade analysis Entry Exit P/L % The trend continued during the week. Though the optimal exit would have been later on Friday. New trades Entry Exit P/L % The trend unfortunately did not continue and event went as far as our stop loss level during the week. Entry Exit P/L % The development was a bit hesitant in the beginning of the week ahead of important numbers on Thu and Fri. Thu it turned strongly positive following ECB and US GDP but a stop loss rally during the NY session carried as far as to stop us out. It proved to be a too tight stop loss as USD/JPY revered on Friday and rose to what would have been positive levels FX QUANT AND POSITIONING WEEKLY 8

9 FX O-meters Spot EURGBP EURUSD AUDUSD NOKSEK NZDUSD EURJPY EURCAD EURAUD EURNZD GBPUSD USDJPY EURPLN EURSEK EURNOK USDSEK USDCAD USDNOK SEB FX Trend-o-meter SEB FX Stretch-o-meter EURGBP NOKSEK AUDUSD EURPLN EURNZD NZDUSD USDJPY EURUSD EURAUD USDSEK EURJPY EURCAD USDCAD GBPUSD EURSEK USDNOK EURNOK Trend compared to previous week EURGBP EURUSD AUDUSD NOKSEK NZDUSD EURJPY EURCAD EURAUD EURNZD GBPUSD USDJPY EURPLN EURSEK EURNOK USDSEK USDCAD USDNOK Previous week Current FX QUANT AND POSITIONING WEEKLY 9

10 FX market environment Average O-meter scores Average Trend-o-meter score Average Stretch-o-meter score Standard deviations y = -4E-05x R2 = y = x R2 = Average Vol-o-meter score y = x R2 = Average trend score still well below trend. Markets players seem to have low conviction. This is not a market optimal for trend strategies. Contrarian strategies tend to work better in this kind of environment. Jun-12 Jul-12 Aug-12 Sep-12 Oct-12 Nov-12 Dec-12 Jan-13 Feb-13 Mar-13 Apr-13 May-13 Jun-13 Jul-13 Aug-13 Sep-13 Oct-13 Nov-13 Excess volatility Jun-12 Jul-12 Aug-12 Sep-12 Oct-12 Nov-12 Dec-12 Jan-13 Feb-13 Mar-13 Apr-13 May-13 Jun-13 Jul-13 Aug-13 Sep-13 Oct-13 Nov-13 Jun-12 Jul-12 Aug-12 Sep-12 Oct-12 Nov-12 Dec-12 Jan-13 Feb-13 Mar-13 Apr-13 May-13 Jun-13 Jul-13 Aug-13 Sep-13 Oct-13 Nov-13 Standard deviations FX QUANT AND POSITIONING WEEKLY 10

11 Volatility Implied volatilities Rising implied volatilities JPY low and Scandies high Levels and weekly change 1 month annualized implied volatility Level compared to 1y history 1 month annualized implied volatility (%) (ppts) (%) 11/11/2013 Max Median Min EURNOK NOKSEK EURGBP EURSEK USDNOK NZDUSD 11/4/ /11/2013. AUDUSD USDJPY USDSEK EURUSD USDCAD EURJPY EURJPY USDJPY NZDUSD USDSEK USDNOK AUDUSD EURUSD EURGBP EURSEK EURNOK NOKSEK USDCAD Positive change over the week, rhs Negative change over the week, rhs FX QUANT AND POSITIONING WEEKLY 11

12 Correlation: Currency pairs Cross currency correl increase for a third week while correl with risk is in steady falling trend EURUSD EURSEK EURNOK EURAUD USDJPY USDCAD GBPUSD AUDUSD USDSEK USDNOK EURUSD EURUSD EURSEK EURSEK EURNOK EURNOK EURCHF EURCHF USDJPY USDJPY USDCAD USDCAD GBPUSD GBPUSD AUDUSD AUDUSD USDSEK USDSEK USDNOK USDNOK NOKSEK NOKSEK USDCHF USDCHF EURJPY EURJPY EURCAD EURCAD EURGBP EURGBP NZDUSD NZDUSD EURPLN EURPLN MSCI W MSCI W VIX 1 VIX NOKSEK EURNZD EURJPY EURCAD EURGBP NZDUSD EURPLN MSCI W VIX High positive correlation Low positive correlation Low negative correlation High negative correlation Cross currency correlation overtime y = -7E-05x R2 = Jan 28-Jan 11-Feb 25-Feb 11-Mar 25-Mar 8-Apr 22-Apr 6-May 20-May 3-Jun 17-Jun 1-Jul 15-Jul 29-Jul 12-Aug 26-Aug 9-Sep 23-Sep 7-Oct 21-Oct 4-Nov Jan y = x R2 = Feb Currency and risk correlations overtime 14-Mar 14-Apr 14-May 14-Jun 14-Jul 14-Aug 14-Sep 14-Oct * MSCI world is used as proxy for risk. Rising (falling) MSCI world => Increasing (decreasing) risk appetite. I.e. a positive correlation indicates that the currency rate tend to increase when risk appetite is increasing and vice versa FX QUANT AND POSITIONING WEEKLY 12

13 Appendix: Speculative positions Speculative positions in AUD and CAD AUD positioning vs. USD CAD positioning vs. USD FX QUANT AND POSITIONING WEEKLY 13

14 Appendix: Speculative positions Speculative positions in CHF and EUR CHF positioning vs. USD EUR positioning vs. USD FX QUANT AND POSITIONING WEEKLY 14

15 Appendix: Speculative positions Speculative positions in GBP and JPY GBP positioning vs. USD JPY positioning vs. USD FX QUANT AND POSITIONING WEEKLY 15

16 Appendix: Speculative positions Speculative positions in MXN and NZD MXN positioning vs. USD NZD positioning vs. USD FX QUANT AND POSITIONING WEEKLY 16

17 Appendix: Speculative positions Speculative positions in USD and total amount of FX contracts held by specs FX QUANT AND POSITIONING WEEKLY 17

18 Appendix: Speculative positions Speculative positions (proxy) in SEK and NOK SEK positioning vs. EUR NOK positioning vs. EUR EUR/NOK spiked higher following a switch from short to long in the proxy again last week! Speculators are according to our proxy positioning very wary of long NOK positioning. As soon as a long position has been indicated by our proxy in 2013 a rise in EUR/NOK has followed right away or in the coming weeks (as shown in the chart above) FX QUANT AND POSITIONING WEEKLY 18

19 Appendix: Speculative positions Speculative positions in Crude Oil and Gold Crude Oil Gold FX QUANT AND POSITIONING WEEKLY 19

20 Appendix: Speculative positions Speculative positions in S&P 500 & Nikkei as of Sep 24 S&P 500 Nikkei FX QUANT AND POSITIONING WEEKLY 20

21 Appendix: T3 systematic trend model Performance P/L per trade Performance per trade Aggregate Average trade Avg winning Avg losing Winning Losing Winning % Winning ratio Highest trade Lowest trade Max adverse excursion * 2012 Inception* % -3.07% 0.16% -0.13% 1.18% 0.90% -0.83% -0.74% % 38% % 2.2% -2.4% -1.6% -3.7% Winning Losing Aggregate /2 1/9 1/23 1/30 2/6 2/13 2/20 2/27 4/2 4/16 5/7 5/14 5/21 5/28 6/4 6/11 6/18 6/25 7/9 7/30 8/20 8/27 8/27 9/10 10/8 10/22 11/5 11/12 12/3 3/5 3/25 4/15 5/13 9/16 9/30 10/7 10/14 10/21 10/28 11/4 11/ FX QUANT AND POSITIONING WEEKLY 21

22 Appendix: Short-term fair values* Scandies 1.97 (2.16*) * Signal used in basic trading setup FX QUANT AND POSITIONING WEEKLY 22

23 Appendix: Short-term fair values * USD FX QUANT AND POSITIONING WEEKLY 23

24 Appendix: Short-term fair values * EUR Source: Bloomberg and SEB FX QUANT AND POSITIONING WEEKLY 24

25 Appendix: Volatility EUR crosses EUR/USD EUR/SEK EUR/NOK EUR/GBP EUR/JPY NOK/SEK Source: Bloomberg and SEB FX QUANT AND POSITIONING WEEKLY 25

26 Appendix: Volatility USD crosses AUD/USD NZD/USD USD/JPY USD/CAD USD/SEK USD/NOK Source: Bloomberg and SEB FX QUANT AND POSITIONING WEEKLY 26

27 Appendix: Risk appetite index & its factors 27

28 Appendix: Explanations Percentiles and quartiles to analyze the speculative positions We apply percentiles to measure the significance of the latest net positions, speculative share of open interest and changes in these compared with history. One advantage with percentiles compared to other measures of dispersion is that it does not make any assumptions about the shape of the distribution (such as normality). The x th percentile of a data set is the value below which lie x% of the previous positions. E.g. if the current net position is the 90 th percentile then 90% of the previous net positions have been lower and 10% higher. We break down the distribution of the data into quartiles where: the first quartile is the 25 th percentile (lower quartile) below which one fourth of the data lie, the second quartile which is the median below which lays half the data (middle quartile) and the third quartile which is the 75 th percentile (upper quartile) below which three fourths of the data lie (and consequently 25% lie above). Frequncy th percentile 25-50th percentile 50-75th percentile th percentile Actual distribution Frequency The extreme points of the data set are represented by the 0 th and 100 th percentiles which are the min and max of the data. The range between these is one measure of the spread of data set. However, the inter quartile range, defined as the difference between the first quartile and third quartile, is often a more informative measure of the spread since it focuses on the spread where most observations have been (i.e. an objective and transparent way of excluding outliers). 0-22,149-17,676-13,203-8,730-4, ,690 9,164 13,637 18,110 22, FX QUANT AND POSITIONING WEEKLY 28

29 Appendix: Explanations T3 systematic trend model set-up A strong trend which is not stretched and without excess volatility is the dream scenario, as was stated in the SEB FX O-meter explanation on page 6. To identify such dream scenarios, i.e. trend trade candidates, we have formalized three rules. One to identify strong trends and two to falsify the indication (filters). For an entry to be signaled in a currency pair it has to be: One of the three strongest trends Not excessively stretch Not showing excessive volatility Entry, exit and risk management Entry signals are generated Mondays at CET Exit takes place Fridays on the ECB fix (14.15 CET) An intraweek stop loss level is recommended (we will publish a currency pair volatility adjusted stop loss (percentage) for each trade recommended by the strategy which may be used as a guidance) FX QUANT AND POSITIONING WEEKLY 29

30 Appendix: Explanations FX-o-meters Trend-o-meter This indicator measures the strength of the 1 month trend (measured in standard deviations). Stretch-o-meter This indicator measures the distance between the current exchange rate compared to the 200 day moving average. The distance is measured in standard deviations. A currency should be between -2 and +2 roughly 95% of the time. A positive (negative) value indicates that the current rate is above (below) the 200 day moving average. Vol-o-meter This indicator measures the 1 month volatility and compares it to the average volatility over the last 200 days. Excess volatility suggests that the risk is high. 20d rank Measures the rank of the current level where 1 indicates the highest level and 20 the lowest seen in a 20 day period. This is a complementary trend-o-meter. A currency in the upper price range is more likely to be trending higher than one that is in the middle of its 20 day price range. Overall: A strong trend which is not stretched and without excess volatility is the dream scenario. This is the sort of trend which our T3 systematic model tries to identify each week FX QUANT AND POSITIONING WEEKLY 30

Short-term fair value table

Short-term fair value table Macro & FICC Research Short-term fair values Tuesday 12 February 2019 Time for some Scandie strength? NOK with correction signal EUR/SEK correcting lower ahead of Rix. EUR/SEK has been trading in overbought