Macro RV: A Primer

This post is a broad response to the question posted here: How is RV actually traded?

In a word, macro RV rests at the intersection of art and science; of discretion and statistics.

RV trading for macro products is akin to value investing for equities: you are trying to identify cheap (rich) assets (stocks) and take a position in them accordingly. The difference between these two methods rests in the requisite skills required to identify "value" in macro products rather than equities. Macro RV "value" is identified in within a specific currency paradigm (for instance, 2s being cheap in the 1s2s5s USD swap or treasury curve fly), or across currencies (GBP 10y is trading cheap to USD 10y). This holds for linear instruments (bonds, futures, swaps, asset swaps, cross currency swaps) and nonlinear instruments (bond future options, interest rate future options, and swaptions).

RV opportunities are uncovered through statistical methods, and trades are typically placed when the overarching flow or macro landscape indicate that a catalyst exists to support mean reversion from a stretched level, or even a continuation of trend. Thus, you need strong statistical chops to determine if something is actually rich or cheap in one of the aforementioned paradigms, and you need a strong understanding of market fundamentals and the overall macro environment to determine whether a catalyst exists for this trade to have positive expectancy.

To provide some nuance to the statement above, you may also just be searching for the best structure to express a specific position. If you want to be long USD rates, i.e., if you expect USD rates to fall, should you be long 2y, 5y, 10y, or 30y treasuries? swaps? or futures? Should you be long duration in an outright fashion, or via curve or fly? Thus, RV is a mechanism to glean additional Sharpe and mitigate portfolio volatility; you might have an incorrect call on duration (rates may head higher), but the construction of your long position may fair better if put on in one sector of the curve versus another, or if it was expressed via curvature.

RV is not eliminated once you say "done" on a trade. Rather, the Bid/Offer you cross to put on the desired risk eats into your positive expectancy. To reiterate; the the RV opportunity exists because there is some flow that has put the market into that level of pricing. A great example of this is 10y rate vol pricing during MBS convexity events; 10y vol tends to outperform irrespective of the level of rate and certain measures of curvature, but this vol can continue to become more stretched. It is therefore key to determine when the convexity event is nearing its end, less you run the chance of being stopped out of a speculative term structure trade within the vol surface. Alternatively, you can capitalize upon the relative richness of 10y vol by constructing conditional curve trades (if you have a view on the curve), and obtain some financing for a position you would gladly put on in linear space at prevailing levels. This plays into the dynamics I alluded to above; RV is about intelligent trade construction and intelligent portfolio construction.

There are most certainly traders that specialize in the RV domain. But one trader's RV strategy may differ from another's. 

The requisite requirements for any rates RV trader is a strong understanding of DV01 (this is a significantly denser subject than you may realize -- convexity is not free), PCA, the Ornstein–Uhlenbeck process, the treasury bond basis (i.e. the futures/bonds basis), volatility surface, skew, and consequently SABR (which pertains to the last two topics). 

yieldcurvemonkey your 2y 1s2s15s fly popped because this is a PCA flattener. You should compare paying (receiving) 2s in this fly to other curves/flys (which you appear to have done!); is this superior to a basic 1y 2s10s curve on a rolling sharpe basis, conditional on being paid/received the 2s? These are the questions RV traders must ask themselves.