Vega problems

ballsofgold said:
However, this phenomena in equity indices don't apply to say a market like lme copper or gold.
More...

I was a MM and one of my symbols back then was GLD. I made most of my money with ratio spreads and selling 6 and 9 month options and buying front money when it got low. The back months moved very slowly but each small move, moved prices a fair amount. I found that these did act like equity options except the back months were MUCH higher. like 27 vs 17. This was 2008-2009.
 
Rob, right so in Gold (Im not sure about GLD, perse)

being long front end and short back end is also being short black swann, tail, event risk and this is how ones blows up on very big moves

Let's agree that each time series are path dependent options, as if to say the behavior of one option does not impact the other. if you agree on this then we can move onto the next point.....

take an option calculator and check the vega on an atm option for fronth month vs back end. You should see that the further out you go, the HIGHER the vega exposure

So what some traders do, thinking they are hedged is put on a ratio trade as implied vols in the front move slightly faster as you described. They will make money until one day event risk causes this relationship to break down as on a big move, a lack of liquidity in the back end causes pain on your vega exposure on far dated options. so you win 99 times and that 1 time you lose, you drop a big percentage in pnl

You are better of selling rallies in front end and buy dips in back end to leg into the time spread, especially since over a longer period (1, 3, 6 months) a big move can recover theta losses on far dated options

notwithstanding the above points, in any scenario gold doesn't behave like equities which was the real point. gold can go up and implied vol can still go higher if there is a technical breakout. when equities go higher, IV comes in..do you agree?
 
Yes, with commodities, each month can move independently vs stocks that don't, however gold futures only trade with a cost of carry, so they are tied.

I don't agree that with a large move in gold futures that the ABS of the back month will move more front months.

I do agree that the skew for gold and many futures is inverted to equities and the fear comes from a shortage. As stocks move up, most IV tend to drop while when gold spikes, OTM calls jump.

With regard to the question at hand, I still believe that back months tend to have smoother movements
ballsofgold said:
being long front end and short back end is also being short black swann, tail, event risk and this is how ones blows up on very big moves
More...

I don't agree. I had some very good days short my calendar with 30 to 50 point moves in the future with my GLD positions. GC and GLD options track each other.
.
 
Rob, if you've made $ with that strategy, well done! But i still feel you are missing point. Allow me to illustrate.

On the option calculator, you will see the vega on the 9m is 3x higher at .35$ vs .12$ in the 1m. So while the while Implied volatilility moves are smaller in the back end, for every 1% you get in the back end you are risk 3x as much in $ terms. I think you are failing to see the risk with negative convexity. As i said, 99x you may win, but in big moves bc of the lack of participation in the back end, you will lose on your vega position on the spread.

If you managed to make money, then I can tell you it is an aberration. Take a look at the spread between 1m and 9m, between 2008 and 2009, during the financial crisis we had big vol moves in gold and you can see as things started to settle down, the spread went from +19% to -10%...so if you still made money after the crisis, then you are one lucky SOB.
calculator.JPG
2008-2009.JPG



This year you can see that the trend is that the 1m is still under-performing 9m...it is statically here for you in black and white...

this year.JPG
 
Last edited:
I agree that the VEGA on long dated options are MUCH higher. I just expect that that value will change slower than the front month. Keep in mind that I had an opinion back then:

-That back months were way too high from excessive demand from hedge funds buying OTM calls everyday
-That the OTM call skew so too high, vs ATM (E.G ATM 3 months out might be 18 but 10 GLD points-100 GC points-might be 22 to 25) This is common in futures like sugar #11 where you might have a shortage of product. Not sure we ever had that in gold.
-That front month vol was too low.
-Front 3 months would range from 16 to 21 and back months were 25 to 27, and actual vol was 17 to 19. I was a buying around 17.5 ish.

I would not do that today. The prices are much different now.

Bob
 
Well in fairness to your point, if you are market making your job is to provide liquidity and you make a bid/ask in which your edge will be execution and timing to risk manage these positions. So it is not uncommon for market makers to have contrarian positions acting as a liquidity provider to the client flows.

But i will say this...calculating fair value on back end vol based on daily breakeven is not the right way.....the funds buy long dated and out of the money bc, as i previously mentioned, the longer term moves will tend to cover theta. So if you loose 3 weeks of theta in the back end but then get a $75 move over the next 3 weeks, you will have covered your time value cost. They are taking a longer term view. That kind of thinking will get a trader into spread positions that move against them without knowing, as per the threadstarter not understanding why his positions are moving against him

So let's say you take, daily breakeven

($1250 x 12% implied vol)/ (square root of 252 days = 15.87) = $ 9.5 daily breakeven

A fund comes in and pays, let's hypothetically say, 17% in back end vol which settled at 16^ yesterday..and you as the market maker are like WTF this is overvalued

They push the ABS from 4% to 5% between the 1m and 9m...and you are taking the other side.
at 9.5$ daily breakeven you want to sell back end.

But what happens if you change the time parameters:


($1250 x 17% implied vol)/ (square root of 12 months = 3.46) = $61.34 MONTHLY breakeven, so is back end vol overvalued or not if there is event risk on the horizon, especially when back end has higher vega exposure, also reflected in the bid/ask?
 
sle said:
The idea is the short-dated implied volatility moves around more then long-dated, right? Which means you need to be dividing by the square root of time to get root-time-weighted vega. So, a cursory look (also, I am flooring the front at 1m to expiry, otherwise root time vega tends to blow up):
May: Vega: -0.14 rtVega: -0.49
Jun: Vega: 0.71 rtVega: 2.07
Sep: Vega: -0.82 rtVega: -1.35
Net Vega: -0.25 Net rtVega: 0.23
So you are net long volatility exposure


Dude, he's exposed to FTSE vol - their budget is permanently fucked anyway :)
More...

At the risk of my thread getting totally side tracked. Is anyone able to help with how these rtVega values were derived please? I can't get the same numbers. I assume the poster is only using the data i posted in my spreadsheet, however, i can't replicate his numbers.

Any help would be gratefully received.
 
BallsofGold and SLE excellent posts! I need some clarification though... 1st question.. kindly define ABS?
2nd question- I am confused with the daily breakeven value? When I looked at it it sounded like "gamma rent" ? - which should involve theta.. so kindly explain the "$9.5 breakeven" vs the $61 monthly breakeven...?

lastly, I am also in the dark about the rttime vega... I got June vol at .71 with RTV at .71*sqrt 43 = 4.65
 
I don't know how SLE calculated the rtvega..tbh.....I can't figure out his numbers or I am doing them wrong....I have experience but would not consider myself an expert options trader as there are limitations on my math. Let's wait for SLE to respond.

I thought the calculation for rtvega when comparing different time series was vega * (square root of DTE) but when you do it on the portfolio level you need to do it slightly differently

Base Time = 1 month...this is arbitrary..lets use 30 days

So vega * (Base Time/(square root of DTE of option times series )

-.14 * (30 / (Sqrt(15) ) = 1.08
.71 * (30 / (Sqrt(43) ) = 3.24
.82 * (30 / (Sqrt(134)) = -2.12

Again, I apologise if I have done this wrong.

ABS, if you look at the bloomberg chart is the absolute spread between tenors...but you need to take convexity and as SLE said, rtvega, into consideration. In fixed income terms, it is basically like bond duration. This is what leads to traders putting on ratio trades based on their respective, path dependent behaviour of each time series

on the daily break even yes, it is just calculating what move you need in the underlying to cover time decay...
 
You must log in or register to reply here.
Back
Top