structured products

nikol · Post by **nikol** » Sat May 16, 2020 8:04 pm

All losses on divs are due to someone did not account for withholding tax (operational risk) or

some country changed that postfactum (legal risk)

sigma · Post by **sigma** » Sat May 16, 2020 9:03 pm

Not only, there is always a dividend risk

For a hypothetical example, say you short $100 strike call on a stock with $110 spot and $10 dividend next day. The delta today is about 50%. If the dividend is canceled by the company by the end-of-day , all other things being the same, including the stock price, the call price will increase by at least $5. This is a gap risk due to dividend risk, that you cannot hedge with delta.

kloc · Post by **kloc** » Sun May 17, 2020 6:12 am

Few points:

(1) These were Asian desks - the underlying wasn't SX5E. Imagine the cost of hedging KOSPI dividends.

(2) Imagine the cost of *dynamically* hedging any index dividends.

(3) Imagine the cost of dynamically hedging dividends on *single* names on HKEX or KRW.

nikol · Post by **nikol** » Sun May 17, 2020 8:43 am

@sigma

True. I'm coming from the old "respected" delta one, so cannot imagine someone didnt take this into account. In those times we had separate reporting with stress 25% steps of dividends. So, these losses were considered.

nikol · Post by **nikol** » Sat Jun 06, 2020 10:10 am

[duplicate]

nikol · Post by **nikol** » Sat Jun 06, 2020 10:11 am

Seems that I have lost intuition.

My pricer returns autocallable coupon price as:

Price(guaranteed coupon) Price(memory OFF)

Am I wrong?

Strange · Post by **Strange** » Sat Jun 06, 2020 2:23 pm

It's both and will depend on the exact parameters of the note. The memory feature essentially accumulates the coupons from dates when coupon conditions have not been met and pays them out when coupon conditions are met. So it becomes a tradeoff between the value of the accumulated coupons and the value of optionality due to met conditions.

nikol · Post by **nikol** » Sat Jun 06, 2020 4:24 pm

@Strange

Spot(S), AutoBarrier (AB), CouponBarrier (CB), Strike (K), ProtectionBarrier (PB), LocalVol (LV) are fixed for both memory ON and OFF versions.

Assume also European, hence payoff at maturity is everywhere the same.

Therefore, I expect (CFs are discounted of course)

P(memory=ON) ~ coupon_rate * Num_of_coupons * Prob(CB < S < AB) + Payoff

condition is a bit more complicated, but ok

P(memory=OFF) ~ coupon_rate * Num_of_coupons * (1-Prob(kick out by CB)) + Payoff

Roughly (1-Prob(kick out by CB)) < Prob(CB < S < AB) , because we add probability to come back.

Guaranteed is when CouponBarrier = 0.

Autocall kick-out probability is the same in all cases.

thanks for reply.

day1pnl · Post by **day1pnl** » Sat Jul 04, 2020 8:41 pm

High barriers to entry. The dealers who are there have been for 15years and been optimising each year. As kloc said, scrambling over a client base consistently shrinking over time. They cant get the entire client eco system needed to run Their greeks (correlation swaps) out even if you have a “pricing model”... which is why it is still banks running that business. but maybe im wrong.. maybe there will always be a market for ‘hard exotics’. But hard exotics are going out fashion and light exotics (no path dependencies, ‘light customisation’) is the future. The fees are not as fat as for hard exo and which is why I Dont see how below mentioned shop Can Do vertical integration into running those trading risks.

Edit: imo, more viable a bank simply buys a promising distributor / intermediary to get deeper in the supply chain. But maybe doing a bit of running your own risk is what puts you “on the map” to be bought.