macro paper trading

Discussion in 'Journals' started by macro_paper, Feb 13, 2017.

  1. I wish you all the best.
    Interesting stuff. I get what you're trying to do but you seem to be all over the place - fx, bond and stock etc.

    Just some comments here.
    First, you're not really analysing public info apart from just mentioning them.
    Second, for bond trading/investment, I'd look at their credit rating too. Italian bonds are currently junky .
    Last edited: Feb 19, 2017
    #11     Feb 19, 2017
  2. Credit ratings are the least useful input for the sort of trading the OP is doing.

    Furthermore, Italy's current sovereign rating is Baa2/BBB-/BBB+ (Moody's/S&P/Fitch). That's not junk.
    #12     Feb 20, 2017
  3. Fyi - fair enough. Technically, they are not junky because they stay just above the border.
    Yes, you're right. Credit rating is least important since he is doing paper trading.

    #13     Feb 20, 2017
  4. Thanks. All over the place? Definitely not in number of asset classes. Perhaps in number of instruments/countries, but diversification is the only free lunch.

    I don’t think fundamentals or ratings matter for Italy at the moment. For example, there was zero reaction to either further contraction in retail PMIs or a very nice industrial production print this month. There is some cushion before being downgraded to junk by all credit agencies, so it cannot happen overnight. It would definitely matter more for Portugal. In terms of politics, if rating agencies didn’t cut Italy to junk when BTPs were bunga-bunga’d in 2011, I don’t know if they ever will.

    Noticed that Australian curve has gotten really steep relative to Canadian curve (pic below LHS). Neither economy looks fundamentally stronger to me, so let’s try to fade it. The latest increase from 24/01/2017 in this box coincided with a heavy upswing in coal prices on Dalian exchange, +12.5% versus oil going sideways. Curious if it’s another bubble, false promises to cut capacity or real efforts to cut capacity. Level/change correlation doesn’t show up versus AUDCAD but chart suggests that AUDCAD sometimes leads the box, so based on this the box could go a little bit higher before going back down.


    Did it through:

    BAZ7 +15 @ 98.935 CAD

    IRZ7 -15 @ 98.11 AUD

    CNM7 -5 @ 136.49 CAD

    XTH7 +5 @ 97.195 AUD

    Schatz has been 7 bps down from position inception, not nice.
    #14     Feb 21, 2017
  5. actually my weekly discretionary econ data assessment indicator shows canada has been doing better lately but due to some statistical offices being nasty and coming up with arbitrary dates for econ releases instead of 3rd wednesday, 5th biz day and etc, I am finding it too troublesome to quantify it based on rules.. so might be biased assessment.
    #15     Feb 21, 2017
  6. There's a syndicated new 10y benchmark issue in Aussie (closing today), so this may have smth to do with it. Also a long-dated linker syndication coming up...
    #16     Feb 22, 2017
  7. I downloaded tender + syndicate data for AUD fixed coupon bonds between 2000 and 2017 and indexed coupon bonds between 2009 and 2017. I then filtered in issues with 8.5+ maturity and looked at performance from 10 business days up to the tender date for: 2s10 steepness for Australian swaps, 2s10 steepness in Australian swaps in excess over 2s10 in USD swaps, 10y AUD government bond spread versus swaps, 10y AUD swap spread over 10y US swap spread, AUD 10y swap versus USD 10y swap. That’s 449 tenders + syndicates.


    There was no result so I tried only with issues over 1b AUD (mostly syndicates). There were only 16 of them. The curve has on average flattened on those occasions.

    Bid/coverage ratio for nominal bonds has been at a relatively high level versus 2013-2017 and somewhat below average for indexed bonds, so it doesn’t look like sudden urge to stop buying their bonds.
    #17     Feb 23, 2017
  8. Butterball


    It's refreshing to see someone on this forum who does not target 200% annual (or monthly!) returns.
    #18     Feb 24, 2017
    samuel11 likes this.
  9. Weekly PNL:

    +1500 USD on EDU1-EDU2. I am expecting this to go up from bearish steepening, so far it’s been bullish steepening.

    -200 USD on short AUD/CAD. Nothing exciting here from 10/02/2017.

    +1978 USD on EUR/JPY. Trimmed this from full + e-mini to just e-mini as spot has moved from 120.45 on pos inception on 10/02/2017 to 118.45. 10y yields down roughly 5 bps in the US and 10 bps in EZ, stable in Japan in the same period.
    -4770 EUR on Italy-Germany spread narrowing. This widened 13 bps in cash and 9 bps in futures from previous Friday, which is in line with FTSEMIB performance, but happy to hold as it had widened too much previously too. OAT didn’t widen against the bund, though.

    -10 925 EUR on Schatz cheapening to EURIBOR via DUH7 and IZ7. Cash schatz yield is down from -81 to -95 bps from prev Friday, DUH7 yield from -70 to -84. IZ7 only moved 2 bps. Will close this if it doesn’t stop generating positive PNL by Thursday, 03/02/2017. May-16: no month end effect, June-16: 9 bps down, July-16: no, Aug-16: no, Sep-16: 4 bps down, Oct-16: no, Nov-16: 10 bps down toward month end but 2-3 back up in the last 2 days, Dec-16: 10 bps down toward month end but 4 bps up in last 2 days, Jan-17: 4 bps down. In Feb-17 it’s 23 bps down from month beginning, so this could give back a few bps on Mond/Tue as in November or December and then normalize in early March. Out of this if this doesn’t happen like that, but will try re-entry if there is a trend change.

    -2520 AUD on lower AUD short-term rates (via IRU8 long). Still like this very much. Eurodollar yield decreased further, wages data didn’t print higher yoy, yet 90d bill yield went higher. NAB biz survey was good though.. Lowe said he didn’t wanna cut so as not overstimulate lending as long as employment doesn’t disappoint too much.

    +0 GBP on higher probability of a BoE rate cut by September. GDP was good enough not to cut probably, CBI’s quite good, let’s see more data.

    +12 USD on betting on higher odds of a March Fed hike via short ZQJ7.

    +1742 AUD, -4800 CAD on CAD curve steepening relative to AUD curve. Hopefully will return to this soon.

    1.85% down for the week, 2.03% from topic inception. Portfolio risks seem imbalanced, e.g. risks in ZQJ7 and the leftovers of EURJPY (e-mini) are a fraction of some other positions.
    #19     Feb 24, 2017
  10. Returning to AUD flattening relative to CAD (ADSWAP10 – ADSWAP2 – CDSW10 + CDSW2) – the box hereafter, which has actually gone further up from 4 bps when I posted the graph (in post #14) to 11 bps. I alluded there that AUDCAD might be leading the box by looking at the graph (positive correlation, i.e. weaker CAD -> more AUD steepening), but checking correlation across years suggests AUDCAD is actually negatively correlated to the box (graph1 bottom). I haven’t found any use of this box-FX correlation, though. Another negative correlation is with the level of US rates, so higher US rates imply higher box (graph1 top). This makes sense because AUD rates are a higher beta on US rates than CAD rates are, so current increase in the box could be a lagged reaction to the post-Trump increase in US rates (graph2).

    The box is by construction related to 2y AUD-CAD differential, higher 2y differential -> lower box if 10 year yields stay in place. Graph3 shows that as RBA cut rates through 2016, the 2y differential decreased and Australian 2s10 flattened less than Canadian 2s10, so the box reached 0. However, as the 2y differential widened again, the box kept marching higher. This divergence is best pronounced in graph4.

    The interesting question if there is sound relationship between the 2y differential and the box. Plot graph between the 2 series for 2015-2017 period has rather poor R^2, graph 5.

    In the 2011-2014 period the box went higher as 2y differential was falling in 2011-2012 but the box didn’t react to a decline in the 2y differential in 2014, graph 6. Scatter plot shows better R^2 with 80 bps intercept for the box and -0.44 coefficient to 2y differential, graph 7. Residual then confirms that the box was has got 50 bps too low in 2014, graph 8. It’s plausible that current box increase is correcting the underreaction in 2014.

    The exercise is repeated for the 2008-2010 period to check if the results for 2011-2014 are robust, graphs 9-11. R^2 in scatter plot is a lot lower, but plotting residuals with either 2011-2014 coefficients or 2008-2010 residuals makes little difference, so could as well use 2011-2014 residuals here. The poor-fit model shows that the period began with the box -15 bps too low and ended with ~20 bps too high.

    Graphs 12-14 do the same for the 2004-2007 period. R^2 is better in this period, but coefficients are totally different from 2011-2014, higher intercept of 126 bps and higher coefficient on the 2y differential of -0.89. This makes marked difference when plotting residuals. The 2004-2007 period coefficients show that the box started period 15 bps too low and ended period 95 bps too high. The 2011-2014 coefficients would suggest the same direction but 110 bps too cheap in the beginning and 15 bps too cheap in the end.

    Graphs 15-17 are for the final 2000-2003 period. R^2 is good and visual fit is really good. However, coefficients are very different again. The intercept is only 40 bps and the 2y differential coefficient is -0.68. The residuals show that with 2000-2003 coefficients the box was 50 bps too high and ended the year fair, the 2011-2014 coefficients show that the box started 20 bps too cheap, ended 110 bps too cheap.

    Graph 18 plots 2y versus the box on the whole period, but no result. Graph19 plots residuals with the whole period coefficient versus using 2011-2014 coefficients. Results is very conflicting, it doesn’t even tell if the box is too low or too high. However, it’s very tempting to use some graph to confirm my original idea that AUD is too steep to CAD rather than dismiss this as garbage.

    Table1 then summarized what the regressions said about box being too low or too high in the start and the end of the period using coefficients from each period separately and using 2011-2015 coefficients (columns 6 and 7 show box levels). It’s hard to put much faith in this when 2004-2007 period coefficients say that the box was 94 bps too high, but 2008-2010 coefficients say it was 12 bps too low.

    Later today I’ll try to see if there is anything I can with those residuals or they are UTTERLY useless.

    Graph names are above a graph, not below as conventionally.

    Graph1. Box correlations with AUDCAD and US rates

    Graph2. Box versus 5y US rate

    Graph3. 2y differential, AUD 2s10, CAD 2s10


    Graph4 2y differential vs box, 2015-2017


    Graph5 Box on 2y differential, R2, 2015-2017

    Graph6, 2y differential vs box, 2011-2014


    Graph7, 2y differential vs box, R2, 2011-2014


    Graph8, 2y differential vs box, resids, 2011-2014


    Graph9 2y differential vs box, 2008-2010


    Graph10 2y differential vs box, R2, 2008-2010


    Graph11 2y differential vs box, resids, 2008-2010


    Graph12 2y differential vs box, 2004-2007


    Graph13 2y differential vs box, R2, 2004-2007


    Graph14 2y differential vs box, resids, 2004-2007


    Graph15 2y differential vs box, 2000-2003


    Graph16 2y differential vs box, R2, 2000-2003

    Graph17 2y differential vs box, resids, 2000-2003


    Graph18 2y differential vs box, R2, 2000-2017


    Graph19 2y differential vs box, resids, 2000-2017


    Last edited: Feb 25, 2017
    #20     Feb 25, 2017