What exactly is the nature of swaption vol data?

Question

Newbie here. I'm curious about underlying nature of the swaptions data on Bloomberg.

Define an arbitrary tenor and expiry swaption time series as

$X$(vol type, currency, reference curve) = [ $x_{t_1}$, $x_{t_2}$,...... $x_{t_n}$]

where

$X$ = source (bank, vendor, bloomberg, etc)

Input parameters is a combination of:

1. whether the quote is normal vol or black(log) vol
2. The currency in question
3. What curve does it reference (OIS vs IBOR)

Example: BGN(Black, $, OIS) = [ (32 @ 3/4/1990), ...... , (78 @ 4/3/2018) ]

This would be a time series of Bloomberg quotes of black vol USD Swaption w.r.t to OIS in some time range

I know that swaptions were quoted in terms of black vol wrt to IBOR early on before switching to normal vol at some point and then reference OIS instead after 2009.

Thus, from today to 1-3 years back, the real data (the one that should be backtested) would be BGN(Normal, $, OIS). Or just stick to 3 years worth of data?

Today, does it make sense to also include the other combinations of inputs?

Are there specific dates to which (inputs) are the dominant convention? etc: 1990-2008 for black vol LIBOR then 2008-2009 for Black vol OIS, then 2009-2018 for Normal/OIS? Then if I'm working with data from 1990- 2008 , one should use Black vol + IBOR and so on? This combination would be the "real" vol data that one should work with?

How are data points outside the dominant inputs generated? (1990 Normal Vol+OIS)

What are some differences in backtesting the vol conventions?

For a even more confusing twist, should I even use a dealer's data (I feel the 4-5 year range is decent, but 2008ish where banks start changing their calculations may mess up the results)?

Thanks!

Answer 1

The following is approximately correct: use Normalized vol for the period that it is available, and prior to that , use Black vol * (the at-the-money forward rate). The forward rate is available on Bloomberg for several years of history. The Libor versus OIS choice of discount rate can be more or less ignored- it only changes the implied vol by a percent or two, since Libor-OIS spreads were fairly tight prior to 2008.