Oil with Rates arbitrage

Discussion in 'Financial Futures' started by rokka, Apr 27, 2013.

  1. rokka



    I'm looking to do some arb on Crude with 10Y Note futures.
    They have been moving in a lockstep from the beginning of 2012. Regime has been established and probably continues for some time.

    You can see it on the chart attached, it's inverted ZN with CL starting from beginning of 2012

    Rough estimate, based on notional and contract margin requirements, shows ratio between them is something like 1/5, so for a current trade it would be "short 5ZN, short 1CL".
    Oil is 10$ rich to interest rates, thus short CL.

    It could be richer though sometime, based on this model - up to 20$

    The question I have is how to calibrate model more precisely, it terms of bps (DV01)?

  2. rokka


    That's funny
  3. 1) The correlation will never be as "tight" as you want it to be. :cool:
    2) You can easy be "raped" when each leg of the trade goes against you, especially the crude oil side. :eek: :( :mad:
  4. rokka


    At some point regime will change, no doubt.
    Btw, on intraday basis (5-8 days intervals) its better correlated (the spread is 2-5$).
  5. before you get carried away with the trade, you should google the topic of stationarity and put some statistical measures together that can give you an idea of your likelihood of success entering the spread at various levels.

    The tools exist to tell your your probability of success (and even expected payoffs) with historic data that tends to be mean reverting .... [ assuming its not by chance, which it really isn't -- every correlation trader sees these relationships and plays this game... ]
    but ... short ZN = long ES = long NQ = long CL = long USDJPY = short ZB = long GC/SI (conditional certain regimes) = etc etc ...

    So you might find better luck analyzing different pairs.

    that said, correlations work until they don't, then they do again... often regime shifts will eat all of the $$$ you make previously on a stable tight reversion between a pair. Well such is true with most trading so that shouldn't stop you.
  6. rokka


    That's correct, there are different pairs well correlated with rates (or fx futures with the equity indexes) or what not.

    I don't think it worth doing a full stat research on pairs correlation, because historical performance doesn't guarantee anything.

    But what I'm pretty sure about, after trading futures for a couple of years, is that one can loose capital much, much more quickly by playing one product going instead of a pair(s). So I'd like to put some definitions around bps, hedge ratios and so on.
  7. bone

    bone ET Sponsor

    The two year correlation daily-close-on-close is - 52.7%. The five year correlation is + 7.7%. So, I reject your original premise for this arbitrage.

    Additionally, the CME has no SPAN margin credits calculated for this combination, which tells me that the exchange is not willing to provide hedge valuation credits for this as well. So, you will get no credit for it in terms of margin.

    Also, I don't see what you're seeing using live data. The Crude Oil ( electronic ) is the blue bar and the Ten Year Note ( electronic ) is the black bar:


    You are, however, quite correct about most traders being more consistent with spread trades generally speaking. They "behave" better in terms of market action and model better typically speaking. But you should be looking at highly correlated products. You can get seriously hurt without more thought and knowledge and background.
  8. rokka


    bone, thanks

    I'm using iqfeed data (the chart was from TOS). Just tested quickly in Excel: from Jan 2012 TNX and /CL have correlation 0.72 (I compared daily highs).

    But it shouldn't be too always, otherwise there is nothing to trade. Also I was looking into arb for daily periods, but there are a lot of opportunities intraday as well, for this particular pair.

    And the last note - crude and rates might not be the best choice, but there are plenty of others: es with zn, wheat with corn, gold with rates, 6e and 6a (the last one has probably correlation=0.99 for the last 2 weeks) - list goes on and on.

    It's interesting that CME doesn't provide margin reduction for this pair, crude should be sensitive to inflation "by design".
  9. bone

    bone ET Sponsor

    ES and ZN have a 2 year correlation of +46%. Corn vs. Wheat is + 90%. Gold vs. ZN is 35% over a 2 year period.

    Corn vs. Wheat and that currency pair are the only statistically viable pairs you have mentioned. The others are simply not tradeable. The point is especially apparent if you look at the order books in real time side-by-side. You will get sodomized. These are not arbitrage trades. Henry Hub NG Futures versus Transco Z-6 swaps are an arbitrage. Comex Gold versus LME Gold is an arbitrage. Owning two grain storage silos on the Illinois River and trading that deliverable against the CBOT futures is an arbitrage. You are asking for unsolicited and unlubricated prison sex when you speak of intraday opportunities and CL vs. ZN or GC vs. ZN. Seriously. Look at the intraday order flows. Look at the exchange SPAN margin credits. Look at the statistical cointegration.
    #10     Apr 29, 2013