Trading Overpriced and Underpriced Options

We often read books were options 'guru' and ex-market makers teach that we should look for underpriced options and buy them and overpriced
options and sell them. I am assuming they are looking at the implied
volatility and comparing it with historical volatility. 

As an ex-market
maker I was wondering if you felt there was an edge in this that could consistently be exploited, or would this be difficult  for a
retail trader with limited tools etc.

Dave

***

Thanks Dave.  This is one of those really good questions that somehow gets overlooked.

Any time you can gain an edge, that's to your advantage.  And yes, most
people who refer to trading over-and under-priced options are referring
to making implied volatility comparisons. Yes, MM can take advantage,
but I seldom try as a retail investor.

It's going to seem as if I have a negative comment about everything,
but the truth is that edge is difficult to define, and the bottom line
is that I make little effort to take advantage.  One philosophy of trading is that if you have no edge, you have no reason to anticipate earning any money from the trade.  I accept that idea as reasonable, but the difficulty is knowing when you have an edge vs. when you think you have an edge.

The best hedge is to find overvalued/undervalued
options in a
single security, and produce a spread.  However, that once simple idea
has been overshadowed by the simple fact that option volatility is known
to be
skewed and that lower strikes appear to be overvalued – when
they are
not.  In the early days, we sold naked OTM puts.  But with a better
understanding of option skew and the explosiveness of OTM puts on a
market decline, those who survived no longer consider selling a bunch of OTM puts as trading with edge.  [Selling naked puts as a precursor to acquiring stock for a portfolio is okay – when the risk is understood]

So yes, you can compare current IV with historical. However, the
time period for 'historical' is open to question.  Should we care how
the stock traded over the past few years, or only the past month or
two?  That's a difficult question to answer.  The very nature of some stocks
changes over time.  Big growth stocks become stable companies and grow
at a slower rate.  Surely we recognize that IV is going to be lower
than it used to be for those stocks.  We never know how stocks will behave in the future, and with which historical period to make comparisons.  That's just one of the difficulties encountered when comparing current IV levels with those of the past. I don't know which historical data to use.

I'm taking a long time to say that yes, take an edge when you can get
it, but it's next to impossible to know whether you truly have an
edge.  Those playing in the market right now have priced options in
such a manner that tells you that you would be wrong to bet against
current IV.  I once ignored current IV, depending on
historical measures instead.  Years ago I became convinced that this was
the wrong approach and that current IV represents the best estimate for future volatility.

Pre-earnings and pre-FDA announcements are special situations, because the very life of the company may be at
stake.  Thus, huge price changes are possible upon release of the news.  It's may be tempting to refer to these options overpriced and sell a limited number to gain an edge, but risk of loss is real and the best approach is to control risk by paying careful attention to a worst cast scenario.

But, day in and day out, current IV is a decent measure of fair
pricing.  Again, if we are in an unusual environment, such as 2008 when
IV was steadily rising, we all knew there was going to be a great
time to sell volatility.  The huge problem was: when is that time? 
Sell too early and risk a disaster.  Wait, and you may miss a big
opportunity.

As a market maker, we get to make a lot of trades and can built a
substantial pile of edge over time.  Buying near or at the bid price is far more efficient that buying at or near the ask price.  That's the best way to get edge. Our task was to hedge various risk (most often vega, but it can be gamma
as well) factors, giving up as little edge as possible.

As a retail trader, that's a difficult task.  Yet, you can sell premium
when prices feel overvalued (higher IV) and buy premium when you feel
options are cheap.  However, you will have vega risk.  If IV does not
return to 'normal' levels you may not be happy with the position. 
That's the problem. 

The big trader – market maker or otherwise – can
easily and relatively inexpensively build a  position that reduces much of the
much risk.  As retail, if we own a position, it takes too
many additional trades to get neutral everything.  So I just accept the
fact that I cannot do that.  It's not limited tools – it's limited
capital and limited margin room.  If you want to trade a 20-lot spread,
you don't really want to be trading another 20 contracts (or more) just
to hedge all risk.  By being unable to buy bids and/or sell offers,
your trades would not be made at prices that are good enough to justify
the effort to completely hedge.  The main reason to buy edge, is to
lock in that edge and collect the profits when the options revert to fair value. 

As retail customers, we seldom lock it in.  We wait for
the edge to work (become cash) – and then collect if we are correct, or
pay if we are not.  The professional adds edge and then hedges risk and
then adds more edge.

When conditions are favorable, or when I perceive more edge (higher IV), I
can (and do) increase position size.  Not by too much.  That's being
greedy and taking too much risk.  But I can add 10 to 20% more
contracts. 

Similarly if IV is low and I don't like the idea of buying
premium (and I don't), I must change my style when selling
premium.  I accomplish that by selling options with less vega.  Alternative: reduce position
size.  That's one way to capture a little extra edge.

766



, , ,

10 Responses to Trading Overpriced and Underpriced Options

  1. Marty 08/18/2010 at 1:07 PM #

    Great question and response! As far as comparing historical to implied volatility, wouldn’t it be wise to use a historic window of the same length as the remaining life of the option/spread in question? If I was looking at a 3 month IC for example, I could compare the IV to a rolling 3 month historic window over the past several years.
    I’ve done this before when I was trying to optimize the inner width of an IC, and I’m not sure I agree that market IV is automatically the best estimate of future-realized volatility. Perfoming this exercise today with SPY on a front month IC with 1 point wings (I don’t actually trade the front month, but it’s the easiest example), I could theoretically get something like $0.55 in premium with short strikes that my historical analysis says will only be touched 20% of the time and will expire OTM 98% of the time!
    But as you said, price is only one factor in most spreads, including the IC, so how do I know that IV won’t move against me? I haven’t found any good way to pick the IV top or bottom…in my opinion that’s probably harder than timing stock prices. The only solution I know of would be to consistently open ICs at the same premium (so adjusting the width of the inner strikes in sync with IV) every week or so, and ONLY when IV is above a level that my historical analysis says is well into the tails of past real-life movements over identical timeframes.

  2. Mark Wolfinger 08/18/2010 at 2:14 PM #

    Hi Marty,
    1)I don’t see any benefit to looking at rolling three-month periods to generate a number we can use to represent historical volatility (HV).
    My perspective is that I don’t know the future, so when using the past, my choice of which dates to use is not necessarily going to have any relationship with future volatility. Plus, I don’t know how long I plan to hold the trade, but it will not be as long as three months.
    All that gibberish above means: Choose any historical that seems reasonable to you. I have no idea which data gives the best approximation of future volatility.
    2) The market makers and those participating in the trading do consider IV to be the best current estimate. I used to shrug my shoulders and say ‘who cares what they think?’ The point is: There is no reason why my guess should be better than theirs. A lot of people are trading on current IV. Or, the option prices may have a certain skew for a reason that I don’t see.
    I choose not to fight those real time market prices. Sure, I’ll trade some options, but I will not bet the ranch that I am right and they are wrong (although I did do that many years ago).
    3) Is collecting 55 cents really a realistic price? Your statistics ant not anywhere near reasonable – and that’s the danger when fictionalizing them. Marty – as a rule of thumb, chances of touching are approximately double the chances of expiring ITM.
    You earn $55 98 times out of 100 and lose no more than $45 only twice out of 100. Sign me up. I’ll do some of this trade – and I’ll do it with no adjustments.
    4) You never know that IV will not move against you. You take that risk any time you own a position that is not vega neutral.
    Yes, timing the entry and doing some trades weekly is one way to play. You lose out on some theta, but save some money when the market moves.
    Yes, collecting the same premium is also a viable strategy. Just know: the premium comes about as a result of IV and thus, you are taking IV into consideration when choosing strikes.
    Regards

  3. Marty 08/19/2010 at 8:23 AM #

    “Is collecting 55 cents really a realistic price? Your statistics ant not anywhere near reasonable – and that’s the danger when fictionalizing them. Marty – as a rule of thumb, chances of touching are approximately double the chances of expiring ITM.”
    That was a real price yesterday afternoon, nothing was fictionalized, I actually ran the model in my example for a 1 month SPY IC. And I’m afraid I have to disagree with your rule of thumb in this case…if you analyze a rolling window like the one I mentioned (which you dismissed out of hand for some reason?), the historical touch and expire probabilities form a log-normal distribution. The farther out on the tail of this distribution you go (bigger UL price movements), the less that rule of thumb applies.
    “The market makers and those participating in the trading do consider IV to be the best current estimate.”
    As you pointed out previously though, market makers shouldn’t care what current IV is as long as their vega risk is appropriately hedged…they’re in it for the spread. As for others participating in trading, they tend to exhibit herd mentality just like any other market. In fact, take a look at the VIX overlaid with the inverse prices of SPX. Their movements are almost in perfect lock-step. IV is an emotional measurement, it’s a very effective fudge that makes it possible to model option prices in a world full of irrational, fearful or greedy human beings.
    So to trust the accuracy of an emotional gauge in predicting future events is not smart in my opinion. Would you say that equity or options markets were pricing volatility risk correctly in Aug/Sept ’08? I hope no one would. Options WERE incredibly cheap that month and ridiculously overpriced the following months. All due to emotion. I do agree with you that it’s near impossible to pick the exact top or bottom, hence my strategy for “dollar cost averaging” when IV seems excessively high. But as far as I’m concerned, overpriced and underpriced options absolutely do exist and to think that the market place has the best estimate of true risk or volatility is an idea I can’t find much support for in the historical data.

  4. Mark Wolfinger 08/19/2010 at 9:08 AM #

    Marty:
    1) If you could get 55 cents for a one dollar spread, and if the chances of expiring worthless was 98%, why didn’t you load up on this trade? Or did you?
    2) I am guilty of dismissing it out of hand – because I was unable to see the benefit. I have never used a rolling window – and I am blind to the benefits. How does the method for calculating HV affect the probability of touching?
    My ‘probability’ calculator just uses a ‘volatility number.’
    3)Agree in Aug/Sep IV was far less than future realized volatility turned out to be. Thus, options were incredibly cheap. But how would anyone know that they were cheap at the time? I know I thought IV was high.
    I don’t know what percentage of traders recognized that the options were cheap and bought them (to make money, not as protection). It was clear that IV was less than recent realized volatility, but the real question for volatility traders was: Will that high realized vol continue? It was a decent warning not to sell vega, but was it a clear signal to buy? Not to me.
    4) The way I see it, it’s not a matter of whether current IV turns out to be a very good (or poor) estimate of future events. I don’t know how to find a better estimate.
    For a long time, I made trades by betting that IV was a poor estimate of future volatility. But in those early days (late 70s/early 80s) no one used different IV levels for each strike (in other words, skew was either unknown or ignored – at least I was totally unaware of it). I used historical volatility to price options and was told again and again – by other market makers – that it was foolish and that arguing with the ‘market’ was a bad idea.
    I made money fading IV. But eventually that became less effective. I know IV is not going to be that accurate. However, I have no better estimate. Sure I sell extra vega when I feel IV is high, and less when I believe it’s low. But I no longer aggressively trade as if I ‘know’ that my opinion on IV is better that the market’s view.
    5) Marty: Do you fade IV? Do you buy premium when you think IV is too low and sell when you believe it’s too high?
    Regards,

  5. Marty 08/19/2010 at 1:32 PM #

    Hi Mark,
    Thanks for the response.
    I think the advantage with the rolling window comes from having hundreds of different volatilities to build a large distribution. From the shape of that distribution I get the HV skewness value (which a lot of calculators can use) to go along with the mean value. My theory is that a lot of market participants are over-adjusting their skewness input after big events because they’re afraid of getting burned by another fat-tail move, or maybe they’re picking a value based on their subjective feelings, earnings, Fed meetings, etc.
    Or it’s not even a conscious decision for them to change the value itself, but they change some other assumption that feeds into it. To me that’s all emotional and can’t be any better than plugging in whatever HV skew the UL price data itself suggests. I’m trying specifically to keep my assumptions to a minimum so I don’t have much to tweek (or screw up!).
    I agree that nobody knows future volatility, and Aug/Sept 08 was the most obvious example. A pro trader on the floor would disagree with me saying they’re acting emotionally…they’d say the skew numbers they put in are based on years of experience interpretting the data. The data was there in Aug/Sept 08 and in hindsight I think the crash makes sense, but IV missed it completely. Nearly everybody missed it, I don’t think it was possible to predict the exact timing of that event using either method.
    So…I guess my question is if neither IV or HV can see the future, why choose one that’s almost guaranteed to have a large emotional component? At least raw data is completely objective as long as you never change the calculations.
    So from all that rambling you can probably guess my answer to #5! Yes, I do like to fade. I definitely have a contrarian nature, but also to me a method free from emotion just seems to have a big edge over chock full of it.

  6. Marty 08/19/2010 at 1:44 PM #

    Also, to answer your other question in #2:
    I’m compiling two distributions, one for probability of touching, the other for probability of expiring. The curves are both skewed heavily to the left, but the shapes are different. So on the left the 2:1 rule of thumb works well, but further out on the tail it becomes 5:1 or more in certain places. I haven’t seen a probability calculator that takes separate touch and expiration skewness values.

  7. Mark Wolfinger 08/19/2010 at 3:03 PM #

    Thanks Marty,
    Agree with your conclusions. If you can generate good statistics based on fact and not emotion that is beneficial.
    I admire the work you put into this, but I am happier playing in a less complex arena.
    You are using emotionless data that has been collected over numerous time slots, to find a historical skew. So do you fade IV, or do you fade skew?
    BTW: I’m curious why you did not sell that 98% probability spread when the max gain is $55 and the max loss is $45?

  8. Mark Wolfinger 08/19/2010 at 3:04 PM #

    Neither have I. Probability calculators are fairly simple, but useful devices.

  9. Marty 08/20/2010 at 7:40 AM #

    Literally I’d be fading IV when/if the price doesn’t match the amount of risk my model shows. Since my calculations are based around skew, you could say I’m fading that as well. It’s my skew vs whatever skew the market would have to use to generate the current prices.
    In reality skew may not have been a deliberate factor in the other market participants’ decisions, a lot of people who made trades probably either didn’t know or didn’t care about that variable. But since their prices allow me to back into a skew value, I would play it as if they’re all using the same model I am. I think the key is to be consistent in making that assumption.
    I didn’t make the trade in my example. I’m still double/triple checking this model before I put any money behind it. There might be some error in the calcs or flaw in this logic that watching and waiting will show me. Right now I’m automating the historical data collection, then I’ll start making paper trades with it and keeping a journal of observations.

  10. Mark Wolfinger 08/20/2010 at 8:17 AM #

    Marty,
    Glad you have the patience to do this research properly.
    Trading skew for edge” That feels like a more efficient way to trade. I hope this works for you.
    Regards,