4 ms·

> For instance, if you paper traded a delta neutral book simulating a market maker, you’d understand how options pinning works (it’s not manipulation, it’s a si

by loxias 5y ago

> For instance, if you paper traded a delta neutral book simulating a market maker, you’d understand how options pinning works (it’s not manipulation, it’s a side effect). Great comment. With added apologies for hijacking, is there any resource you'd recommend with nuggets like the above? Is there any better way to learn than slowly paper trading and testing different things over long time spans? I'm interested in vol trading and dispersion strategies, but the lack of easily available historical data makes it difficult to run experiments.

TuringNYC 5y ago
I've seen entire careers and departments devoted to collecting this data for the purposes of backtesting strats. You could start collecting it starting today, but you wouldnt have a huge back-test and it wouldnt reflect different regimes. It would regardless start giving you insights. I'd guess the best ways to learn it are either doing it and learning bits over time, or joining a hedge fund where you get it all as common infra and learn there. I'm also curious if anyone has ideas here. I havent seen a non-internal historical dataset.
- qeternity 5y ago
  > I've seen entire careers and departments devoted to collecting this data for the purposes of backtesting strats. Very true. The best quant shops don’t have the smartest people, they have the most data and the best systems. Everyone has smart people. Not everyone has great infra.
qeternity 5y ago
There is no better way than doing. It’s not slow. These things will hit you like a freight train. You either get vol or you don’t. For me, the first month of running my own options book taught me more than all my derivs/vol classes in school combined. You can read about convexity until you’re blue in the face, but you won’t really get it until you watch the interaction of every leg in your book and the resulting risk/pnl. Once you see that, it’s like ascending to a higher plane.
- loxias 5y ago
  Much appreciated! Talking out loud, as you obviously know all this: > You either get vol or you don’t. I think I get it, on a conceptual level -- option prices include the market's expectation of future underlying variance, so if you subtract/hedge-out everything else, you can trade that fluctuating expectation. With dispersion, you're taking advantage of index options and those of the components eventually converging, while being mindful of the changing correlation of components. > you won’t really get it until you watch the interaction of every leg in your book and the resulting risk/pnl. Once you see that, it’s like ascending to a higher plane. You're spot on. I can tell I'm missing something "clicking" at a more intuitive level in my head, something that feels close, because my book almost does what I expect, but.. there's always some "error" in the pnl at the end of the week. So I must be missing an "aha" moment. > There is no better way than doing. It’s not slow. My "real job" skills (C/C++, numerical programming, systems engineering) I only learned by doing, I guess no reason to expect this would be any different! Though, I can easily write and test code on my laptop with no consequences if it's wrong. I feel like I'd learn fast if I could set up a trade experiment, click a button, and then see how it fared at the end of the week, immediately, instead of waiting the week. I've got dozens of questions I think would be answered quickly that way. But you're right, the path to learn more is clear, I'm gonna start paper trading those experiments. Thanks for your comment, it's really encouraging. I can't wait for Monday.
  - qeternity 5y ago
    > my book almost does what I expect, but.. there's always some "error" in the pnl at the end of the week. So I must be missing an "aha" moment Depends what that thing is, but most likely you're not. If your pnl/risk expectations have the correct direction, and rough magnitude, I would say you're doing pretty well. What sort of strat/structures are you wearing at the moment?

by loxias 5y ago

TuringNYC 5y ago

I've seen entire careers and departments devoted to collecting this data for the purposes of backtesting strats. You could start collecting it starting today, but you wouldnt have a huge back-test and it wouldnt reflect different regimes. It would regardless start giving you insights. I'd guess the best ways to learn it are either doing it and learning bits over time, or joining a hedge fund where you get it all as common infra and learn there. I'm also curious if anyone has ideas here. I havent seen a non-internal historical dataset.

qeternity 5y ago
> I've seen entire careers and departments devoted to collecting this data for the purposes of backtesting strats. Very true. The best quant shops don’t have the smartest people, they have the most data and the best systems. Everyone has smart people. Not everyone has great infra.

qeternity 5y ago

There is no better way than doing. It’s not slow. These things will hit you like a freight train. You either get vol or you don’t. For me, the first month of running my own options book taught me more than all my derivs/vol classes in school combined. You can read about convexity until you’re blue in the face, but you won’t really get it until you watch the interaction of every leg in your book and the resulting risk/pnl. Once you see that, it’s like ascending to a higher plane.

loxias 5y ago
Much appreciated! Talking out loud, as you obviously know all this: > You either get vol or you don’t. I think I get it, on a conceptual level -- option prices include the market's expectation of future underlying variance, so if you subtract/hedge-out everything else, you can trade that fluctuating expectation. With dispersion, you're taking advantage of index options and those of the components eventually converging, while being mindful of the changing correlation of components. > you won’t really get it until you watch the interaction of every leg in your book and the resulting risk/pnl. Once you see that, it’s like ascending to a higher plane. You're spot on. I can tell I'm missing something "clicking" at a more intuitive level in my head, something that feels close, because my book almost does what I expect, but.. there's always some "error" in the pnl at the end of the week. So I must be missing an "aha" moment. > There is no better way than doing. It’s not slow. My "real job" skills (C/C++, numerical programming, systems engineering) I only learned by doing, I guess no reason to expect this would be any different! Though, I can easily write and test code on my laptop with no consequences if it's wrong. I feel like I'd learn fast if I could set up a trade experiment, click a button, and then see how it fared at the end of the week, immediately, instead of waiting the week. I've got dozens of questions I think would be answered quickly that way. But you're right, the path to learn more is clear, I'm gonna start paper trading those experiments. Thanks for your comment, it's really encouraging. I can't wait for Monday.
- qeternity 5y ago
  > my book almost does what I expect, but.. there's always some "error" in the pnl at the end of the week. So I must be missing an "aha" moment Depends what that thing is, but most likely you're not. If your pnl/risk expectations have the correct direction, and rough magnitude, I would say you're doing pretty well. What sort of strat/structures are you wearing at the moment?