what do you mean by "time decay"? If you're trading options then it's the linear EV of the option you care about (which is not the linear EV of the stock, of course); "time decay" is built into calculating the linear EV of the option.
Conversation
And swaps *are* just linear instruments, so what I said holds, you just care about the linear EV of underlying (marked to how long you're planning to hold, plus interest rates you'll have to pay, etc.)
1
Right, but interest rate volatility factors into your expectation operator and potentially does as a constraint
Worst-case volatility corrections != average case, and for some markets you do one and not the other
1
why does interest rate *vol* matter, and not just EV of interest rate?
2
Anyway all of these feed into "linear EV of the instrument you're trading/the trade you're doing"
1
But the constraints you engender in practice (liquidity constraints, observed correlations / seasonality) turn “linear” into “linear with constraints” which can be... very different!
1
I’m not arguing for logarithmic measurements, but instead a simple observation that you often end up sublinear when you add in practical constraints that are you realized
1
strongly disagree with your claim in practice -- nonlinear factors rear their head once in a while but 99% of the time you're basically just dealing with linear factors
2
Maybe, but it’s not that uncommon
I’ve been blown away in spot markets by crazy buying near the close because some ETF was behind schedule before a rebalance and then went crazy on options expiry day causing a wild gamma squeeze
I think this happens more than 1% of the time
1
getting somewhat far from the original claim that linear EV of what you're trading is most of what matters (agree there are sometimes nonlinear effects on other things that contribute to the linear EV of the thing you're trading)
But also I think this is still not most trades
Yeah my only claim:
a “pure” linear objective function is never really true, you often add lots of constraints and features based on prior events — and these tend to be concave corrections vs. convex
(Weaker than log wealth claims!)
1
agreed, though I think linear is usually pretty close, and also in the end exactly what matters (though with some usually not that important nonlinear behavior from other factors feeding in)
1