How far in time do we need to go out to get a decent price?
One thing I have noticed with the example trades: The trade gets cheaper the further the two Butterflies are apart.
They also get cheaper the longer you buy before expiration. So what is the ideal time to enter a butterfly trade? Here is the trade value of an ATM butterfly over 6 months before expiration:
The trade value really takes off in the last month. Until about 6-8 weeks before expiration the change is very slow. I would put the trade on 6-8 weeks before expiration. While the stock has more time to move out of the profit zone, the cheaper trade might make a huge difference to the final % gain (if there is a gain ...), also adjustments should become easier with a lower initial debit.
How far should the distance between butterfly body and wings be?
The above graphs were all for butterflies with a 2$ difference between butterfly body and wings. What about 3$, 4$, or even 1$? What changes is the width of the breakeven zone and the achievable maximum % gain.
For comparison purposes these are modeled trades, 8 weeks to expiration, with the underlying being right in the middle between the two butterfly bodies and all options having an implied volatility of 18% (calls) and 20% (puts).
The shape of the risk graph is the "Middle strikes 3$ apart" from above - the one that does not dip below zero in the middle.
2$ difference between body and wings
35-37-39 puts and 38-40-42 calls
net debit: 94$
max profit 113%
profit zone 5.12$ wide
3$ difference between body and wings
33-36-39 puts and 37-40-43 calls
net debit: 189$
max profit 59%
profit zone 6.22$ wide
4$ difference between body and wings
32-36-40 puts and 37-41-45 calls
net debit: 288$
max profit 39%
profit zone 7.24$ wide
It is of course difficult to compare these numbers. In my opinion the much lower possible return is not worth the wider profit zone. If I am spot-on with my opinion that the underlying will not trend, I want to be paid handsomely. Also the flat region in the middle of the risk graph, between the two peaks, becomes very wide with the bigger difference in strike prices. In the last example (4$ difference between body and wings) this flat zone is three dollars wide, with a profit of only 12$ - barely enough to pay the trade's commissions.
Something's missing, no? What about a 1$ difference in strikes? In this case the butterflies should not overlap, because one would just end up with a Condor risk graph. But what about a 36-37-38 put butterfly and a 38-39-40 call butterfly with QQQQ at 38? The risk graph drops below zero in the middle (at 38):
The stats are the following:
net debit: 26$
max profit 285%
profit zone 3.53$ wide (ignoring the dip below zero in the middle)
Now I doubt that in reality one could get in that cheap. However I have put on butterflies with 1$ difference in strikes four weeks before expiration for a debit of 20-25$, so eight weeks before expiration 15$ may be possible. That would be a debit of 30$ plus at least 8$ commission for 8 legs, so in reality the max profit and the profit zone are a lot smaller. For sure in real life this trade is a different animal than in theoretical testing...
Which kind of underlying do we use?Butterflies become cheaper with higher IV. High IV has the same effect as going further out in time. Implied volatility is quite low right now for QQQQ, but the options are the most liquid and have 1$ strike price intervals. Also, if you put on a double butterfly on a very high volatility stock, it might be cheap, but the risk that the stock just flies away or crashes down is a lot higher. It is all a trade-off.
Many people prefer indexes for these non-directional strategies, since they have a lot less event risk.
QQQQ´s implied volatility has dropped so much (as of October 2005) that it has become difficult to enter butterflies at reasonable prices. One underlying that has drawn attention recently is the oil services exchange traded fund OIH. Its implied volatility is between 30% and 35% at the moment.
Personally I still prefer QQQQ for its options' liquidity and the many strike prices available.
What do the greeks tell us?
First of all the risk graph over time. The last day is 3 days before expiration: 
Now the charts for the greeks: delta, gamma, theta and vega of this position over time, with the same settings as the risk graph above. I chose the last date 3 days before expiration because the greeks go too wild in the last days to show the progress over time in the same graph.
About 10 days before expiration gamma turns positive at some underlying prices. This basically reflects the fact that at those prices there would be a loss, or not very much profit, at expiration - the underlying needs to move for the position to become profitable.
This graph shows nicely how theta is normally opposite to gamma. Positive gamma means the position profits from movement, positive theta means it profits from non-movement. How nice would it be to have positive gamma and theta at the same time ...
Since there are no time spreads involved, at expiration volatility is irrelevant. A few days before expiration, though, we have positive vega at the long strikes and negative vega at the short strikes. In other words, higher volatility helps if the long options are at-the-money and hurts us if the short options are at-the-money.
Is this a "good" trade? Can I make money with double butterflies?
Many experienced option traders share the opinion that even the fanciest option spreads are in the end a zero sum game. The 'holy grail' type of trade that automatically makes money by just putting it on again and again does not exist. Same for the double butterfly. Although, for example, 2005 has been a good year for non-directional strategies on indices, in the long run putting on a large number of double butterflies (month after month for a long time) should yield a return of not more than the risk free rate on the invested debit - minus commissions and bid-ask spread.
Edge
Option traders need some kind of 'edge' to profit over the long run. Easier said than done ... but here we go, three areas where one can possibly find an edge over the other market participants:
#1 Superior underlying analysis: Unlike the market makers, a trader can decide when and whether to put on a trade. Rather than mechanically entering double butterflies every single month, a trader can try to avoid times with strong trends and only enter the trade when analysis shows that the market is likely to stay range-bound.
#2 Superior money management: Many traders eventually lose all their capital because trade too big. The key is position sizing. An edge can be found by practicing better money management than other traders.
#3 Superior trade and risk management: Adjustments during the life of the position if it becomes necessary.
Many books and articles have been written about #1 and #2, and these subjects go beyond the scope of this article.
Information about trade adjustments is harder to find, so let's take a look at some ideas.
What kind of adjustments are possible/thinkable?
Adjustment trades need to either lock in profits early, widen the profit zone or remove risk from the position, which almost always comes at the cost of giving up some of the maximum potential profit.
Adjustments that increase risk are not desirable. Rolling down a losing trade and doubling its size and risk "for a credit" is not a valid adjustment, unless it was part of the original game plan to initially enter a half-sized position and double it later if necessary.
There is a bit of a myth out there that adjustments can do magic things with a trade. There is no adjustment that magically transforms a losing trade into a winning trade. That said, there are ways to (a) lock in some profit early and (b) increase the likelihood of at least some profit of a trade that is not going optimally.
For these examples I will use a double butterfly with overlapping wing strikes, for example a 36-38-40 put butterfly together with a 39-41-43 call butterfly, put on for a total debit of 90$.
Currently, with implied volatility ultra-low, one has to either skillfully leg into this trade or go out 8-10 weeks before expiration to be able to enter for this debit.
The maximum profit of this trade is 110$ (everything without accounting for commissions).
Scenario 1:
In this scenario QQQQ stays range bound between the short strikes. Since the risk graph does not dip below zero in the middle, non-movement does not really worry us. The crucial moment approaches about 2 weeks before expiration. Let's say QQQQ has gone above 40. We put in a limit order to sell the call butterfly for a credit of 95. The trade would get filled if during its daily fluctuations QQQQ got somewhere close to 41, the call butterfly's body strike. This adjustment covers our initial debit and makes this a risk free trade, with a higher maximum achievable profit. Here is the expiration risk graph after the adjustment:
We would only do this adjustment if we had a strong opinion that this is the end of the move and QQQQ will head back south. The problem is that we end up with practically nothing if we are wrong. To hedge this forecast a bit, we could at the same time sell a 41/42 bear call spread for 35$. That way we have a bit more profit should QQQQ just stay where it is, with an increased maximum profit, and a maximum loss still lower than with the initial trade:
Scenario 2:
Using the same trade as above, the assumption is that we find QQQQ between the overlapping wing strikes (39 and 40) 1-2 weeks before expiration. As the greek charts above have shown, gamma turns positive 1-2 weeks before expiration. Between 39 and 40 we have no loss, but also practically no profit. The idea is to do very careful gamma scalping in the last week of the trade. If we are bold we could start 2 weeks before expiration: although gamma is not yet high enough, we would be anticipating the higher gamma a week later. One could for example go long about 20 shares per double fly at 39 and sell them again at 39.40, or go short at around 40 and cover at 39.60. Very careful, because the stock could very well go towards one of the short strikes of the double butterfly and we don't want to wipe out all possible gain in this case with an adverse stock position. Ideally the gamma scalping would lift up the risk graph trade by trade.
(Note: This gamma scalping works even better for double butterflies that don't overlap, there is higher positive gamma and it comes earlier.)
Scenario 3:
Something similar to scenario 2 can be done if QQQQ is at one of the outer strikes, 36 or 43 in the last 2-3 weeks before expiration. For example going short 20 shares at 36: If the market continues down - and away from any possible profit of the double butterfly, the short 20 shares will recoup at least some of the loss. If the market wiggles around 33 we can trade back and forth the 20 shares for little profits. The price we pay for this opportunity is a lower return should the market start rallying without ever looking back.
Scenario 4:
QQQQ flies away or plummets down beyond the outer strikes early in the life of the trade. I don't think that in this case there is an adjustment that does not add risk. In this scenario it comes down to money management: Does your position size and trade plan allow for the occasional total loss? If yes, the best action could be no action. After such a big move QQQQ is not unlikely to reverse a bit and move back into the double butterfly's profit zone. If no, simply closing out and recouping some of the original debit is always a possibility. With a trade that can frequently return 50% to 100%, the occasional ~50% loss should be part of the trade plan.
Butterflies
