A postbag objector objects to my post on Monday, saying that Warren Buffett explicitly disagrees with quarterly P/L swings on derivative positions, given that they are based on B-S valuations. See e.g. his 2010 newsletter p.21. Given that Buffett is taking in billions of premium without collateral, which he can then invest however he likes, why should this strategy be equivalent to taking a long position, even on no-collateral terms, when the latter would have produced nothing up front?
The objection has no substance. The claim is that some option P/L, like the put position, is imaginary or unreal, i.e. should not be marked to market in the conventional way. So ask whether the P/L is imaginary for some options only, or for all options?
(1) If for all options, then the P/L on a long call position is imaginary too. Then suppose that I am short a put struck at 100, and long a call at the same strike. This is what options traders call a synthetic. You can easily see that the payoff at expiry is identical to a long position in the underlying. I.e. if you bought when the market was 100 and the market is at 101 at expiry, you make 1 on the call position, and lose nothing on the put. If the market is at 99 at expiry, then you lose 1 on the put, and make nothing on the call. So the position has an identical value to buying the underlying at 100. But then it would be quite wrong to ignore the P/L swings on the synthetic, given that you should not ignore them on the real asset.
(2) If the P/L is imaginary on some options only, then consider our synthetic long position, and suppose the P/L is imaginary on the put, but real on the call. Then the P/L of the call position is identical to that of the synthetic, which in turn is identical to the P/L of a long position in the asset. This implies that the call option always has 100% delta, which is absurd.
(OK I haven’t explained why it is absurd, but I think enough is enough).