First full quarter since the topic inception is over. Portfolio is 4.21% down in EUR terms, 2.86% up in USD terms from 10/02/2017 (50% USD, 50% EUR portfolio). PNL from trading is roughly -8200 USD (-0.8%). It’s confusing to analyse the results in this way as annualized SR in USD is 1.25 and in EUR -1.56. Currency fluctuation free SR has been -0.5. I will simplify the exercise and keep it in USD for measuring Sharpe ratio and keep hedging volatility outside the calculations.
99%-VAR target was set at 1%. The target was not met with zero instances of 1% daily losses over 100 days, 4 instances of 5000 USD losses. Annualized 30d vol varied from 4% to 8% and an average of around 5.5%. The reason vol target was undershot is that position sizes were picked with a rough idea that the portfolio would contain around 20 trades, but as the number of hours I spent on this was quite little, the number of trades open at month-ends varied between 8 and 12.
Total of 19 trades were done over two quarters, 10 winning with 3600 USD average profit, 9 losing with -4900 USD average loss, 8 trades are still open. I looked through the worst trades again. One of the conclusions I came to is that managing four-legged trades is too complicated and things should be kept simple.
Short EURNOK: shorted on 28/03/2017 at around 9.22 with 8.80 target, currently 9.54. NOK looked cheap since nov-16 when it failed to respond to OPEC meeting. It traded around 9.40 in summer 2016 when further NOK cuts were still deliberated and oil traded hovered around 50. 9.40 -> 9.22 wasn’t sufficient to account for improvements in econ outlook, imo. Did not enter before 28/03/2017 because was concerned that a dip in oil could weaken it further. Losses came in 2 waves: 9.11 on 14/04/2017 to 9.52 on 04/05/2017. This coincided with another dip in oil prices. It slightly appreciated when oil r, but the second wave started around 22/05/2017 from 9.37 when oil started sliding again to 9.54 now. Inflation hasn’t been good but there was a turn in trade balance figures last October, retail sales have started growing from Feb, IP readings improved this year, PMIs averaged 54 recently. EUR repricing from this week could bite EURNOK. Scandies strengthened on Friday despite EUR doing okay against JPY and USD, and overall week-on-week move doesn’t feel adequate for what happened in EURUSD and EURJPY. Targeting 9.05-9.10 now.
Long USDRUB from around 59 on 10/03/2017 with 60.40 target but moved to 57.20 by 10/04/2017. By 10/03/2017 RUB appreciated by more than 4.2% YTD despite front brent falling from 55 to 51. Local stock market was more than 11% down and reversed all post-Trump gains. Government wasn’t too happy with RUB getting stronger past 60. I disregarded the risk on the dollar side because last year this currency pair was numb on big dollar events like Fed meetings or GDP releases. Dollar fell against a range of currencies around the Fed meeting on 15/03/2017, and RUB decided to respond too. Oil then recovered to 55, so this was pointless to hold.
Short schatz against Euribor Dec17. I related this to French elections. The spread moved 13 bps after Fillon probe was opened in late Jan in addition to the 7 bps it had moved in 2016Q4. The spread had previously bounced off the same levels in Dec. EURCHF reaction was quite muted (and SNB data later revealed they were not selling CHF in Jan and Feb). I thought that Fillon would shake off the accusations (wrong) and Le Pen had little chances to win in the second round against either Macron or Fillon (ultimately right). Risk-reward seemed like 3:2 with prospect of clawing back the 20 bps versus losing 30 bps if I am wrong. Choice of 30 bps was halfway to spread levels that were seen in 2011 as Le Pen winning wouldn’t be as serious as situation was in 2011. Had quite conviction, so accepted a not so high risk-reward. The idea was not too bad but it was probably silly to assume whole 20 bps would be clawed back without seeing the election result. 45 Schatzes was also too big a position for my vol target. I should have waited for better entry level and done smaller size. In hindsight the position was closed prematurely as the spread recovered and would have made 5 bps (just ¼ of the 20 bps I had assumed) profit by March end, though again plunged in April (presumably on Melenchon rising in the polls but could be I am overestimating the impact of the elections).
Steeper CAD / flatter AUD. This has been a train wreck with meh idea, poor assumptions, poor attention to circumstances and operational screw ups. The idea itself was that 2s10 in AUD has steepened too much against 2s10 in CAD without reasons I could see (historically AUD has always been flatter than CAD). At the position on 21/02/2017 inception BoC communication seemed to indicate they were committed to keep rates low (and both were unlikely to move) and CAD economy was doing better. Combination of central bank restrained short rates and CAD economy doing better had suggested CAD steepener should have outperformed. However, the primary motivation was that historically CAD has been steeper than AUD. CAD 2s10 stood at 84 bps and AUD 2s10 at 90 bps, so +6 for the box. The box was range-bound between +11 and -1 until 12/06/2017 when Wilkins indicated they could remove some accommodation. This violated the underlying idea that central banks would remain inactive, and should have cut the position a day later when 10y sold off and offered an excellent opportunity to get out of trade. This would have saved 0.5% of portfolio.
The 2s10 box has been 13 bps up from 21/02/2017 until 30/06/2017 with AUD 2s10 falling by 6.5 bps and CAD 2s10 falling by 19.5 bps. PNL has been different because the actual trade was CGB futures, XT futures, BA Dec-17 and IR Dec-17, and it has been worse. This construction lost 24.5 bps versus 13 bps for 2s10. The CAD part flattened by 31.5 bps in the construction versus 19.5 bps in the 2s10. The differences were receiving 10 bps in BA Z7 roll-down, losing from 3rd BA cheapening versus 2y swaps by 5 bps (and prior to Wilkins speech they had overperformed the 2y by 5 bps, so 10 bps loss), Canadian government bonds richening to 10y swaps by 17 bps. -10+5+17 = 12 bps indeed. The AUD part flattened by 7 bps, so roughly the same as 2s10, with no change in 10y swap spread, paying 11 bps in rolldown, profiting from 3rd IR yield falling less than the 2y swap by the same 11 bps. When entering the trade I completely disregarded the spreads for CGBs versus swaps under assumption they would stay constant, which definitely was not a proper thing to do. The construction correl with 2s10 was 0.86 on CAD side and 0.74 on AUD side.
To make matters worse, initially calced the DV01 incorrectly for the CAD, and had -5 bond futures instead of -4 in the June contract. I only noticed the excess 1 lot short that when it was time to roll to Sep, and the bond was more than 4 CAD up. This trade also was not a good complement to long BA Z8.
The conclusion I am drawing from this is to avoid four-legged trades that are in fact tracking another structure and stick to intuitive simple things. I am closing this next week after the RBA meeting as I think they could flatten the curve by moving IR Dec17 yield higher.
Week 26.06.2017:
EDU1-EDU2: +1500 USD
Short EURNOK: -2125 USD
Short EURCZK: +1063 USD
Short GBP, short long gilt: -3912 USD, +2610 GBP
Long bund against short gilt: -6400 EUR, +5220 GBP
ER Z7-Z8-Z9: 0 EUR
Steeper Cad, flatter AUD: +4545 CAD, -8883 AUD
Long BA Z8: -5500 CAD
Total: -8165 USD
