A Revised Earnings Strategy: About five years ago, we did a lot of work on a model which was designed to successfully predict the direction of a stock price change after an earnings announcement. The model was primarily based on the assumption that investor expectations played a greater role in how the stock moved after earnings than the actual results themselves. If expectations were too high, the stock would fall regardless of how good those results might be. We tried to get a handle on how high expectations were, based on several variables, including whisper numbers published by WhisperNumbers.com and how much the stock price rose or fell in the week or two before the announcement.
At one time, we successfully predicted the direction of the post-earnings change on 8 consecutive earnings plays (usually selling vertical credit spreads which we hoped would expire worthless), but then we ended up being wrong on a couple of consecutive ones, and we discontinued these plays. We decided that the whisper numbers we were dealing with were not particularly reliable. It was not clear how these numbers were compiled by WhisperNumbers.com. Of course, they said it was a proprietary algorithm, but they admitted to polling selected analysts while inviting the public to cast votes on their website. The composite estimates of analysts is supposed to be what defines expectations, so we wonder about these analysts who were asked to reveal their “true” estimates. If their opinions were indeed different, why wouldn’t they be telling their clients what they really believed rather than favoring a private company with their innermost thoughts. And asking individual investors what they were expecting, if they actually counted those votes, should be a contrary variable instead, since individual investors are notoriously wrong most of the time. Bottom line, we did not have confidence in those whisper numbers.
Now we want to try again, with a couple of changes. Instead of relying on whisper numbers, we will check what the stock price does in the days leading up to the announcement as an indication of the level of expectations, and also look at the historical record of earnings announcements for each company to see if we can identify any consistent patterns. We may also take a look at recent hedge fund or insider trading activity if there is any. But for our first attempt, we will constrain ourselves to stock price movements leading up to the announcement date.
Equally important, we will use a different options strategy than we applied in the past. This time around, we will use diagonal spreads which we sell at a credit (or extremely low debit in some instances). The short side will be in the weekly series that expires on the Friday after the announcement, and the long side will be in the longest-out weekly series (usually six weeks later). We will place both a put and call credit spread, usually in the afternoon of the day before earnings are released. One of the two spreads will be guaranteed to make a profit (both the initial credit and whatever the residual five-week-out out-of-the-money option can be sold for).
The other spread may also be profitable, depending on the strikes that were selected. In the event that the stock moved so much that one of the spreads loses more money than the gaining spread makes, causing the combination of spreads not to be profitable at the end of expiration week, there will be about 5 more weeks that new premium (weeklys) might be sold against the remaining long option, hopefully enough so that a loss will be avoided. The spread that made a gain will usually be closed out on the expiration day of the short options.
If both spreads are successful, the entire play begins and ends in a single week. If the combination of spreads is not profitable, we will have to spend up to five weeks trying to recover from the too-high post-announcement stock price change by collecting new premium each week. If the strategy works out to be profitable in a single week over half the time (which I believe should be the case), and we can roll out of the others over five weeks or less, it certainly should be a profitable concept.
Let’s use Red Hat (RHT) as an example. They announce earnings after the close next week, on Monday, March 26. We are uncertain about the expectation level (see discussion below), so we will give ourselves a little wiggle room in both directions (and sell both 3/30/18 puts and calls which are slightly out of the money). If RHT is about where it is right now ($149 – aftermarket bid), here is what we would do:
BTO # RHT 04May18 139 puts (RHT180504P139)
STO # RHT 29Mar18 146 puts (RHT180329P146) for a credit of $.25 (buying a diagonal)
BTO # RHT 04May18 160 calls (RHT180504C160)
STO # RHT 29Mar18 152.5 calls (RHT180329C152.5) for a credit of $.35 (buying a diagonal)
If the stock ends up next Thursday (markets will be closed on Good Friday) anywhere between $146 and $152.50, both these spreads will end up being profitable (both from the credit collected plus the value of the residual May 4 options). If either of the short options expires in the money, we would need to buy it back and close out all the positions if a profit could be made, or roll over the in-the-money option to the next week and collect some additional premium.
Here is the risk profile graph for 5 of these spreads if we assume that IV of the May 4 options will fall by 10 after the announcement (from 34 to 24 – I don’t believe it will fall this much, however, so this graph might underestimate the potential gains):
The investment here is about $690 per set of two spreads, or about $3460 after commissions. The investment amount is greater than the amount truly at risk because there will be residual value in both of the May 4 options regardless of what the stock does after the announcement. The May 4 long options currently are marked at an average of $4.30, so they have a very long way to fall before either or both of them would not have significant value at the end of the day next Thursday. I plan to place the above spreads in my personal account on Monday afternoon (adjusting the strikes if there is a big move either way during the day) and close out all the positions next Thursday unless there is an unusually large move in the stock price. I will report back on how it went.
Here are the numbers for RHT for the past eight quarterly earnings announcements:
There are some interesting numbers in this table. The most significant message is that there is essentially no relationship between how well earnings stacked up against expectations and what the stock did after the announcement. For example, in the first five quarters, earnings exceeded expectations by a huge margin (averaging 60%), yet in 3 of the 5 quarters, the stock fell after the announcement (by an average of 8%).
It seems that the most reliable prediction of what the stock will do after the announcement is to check what the stock price did in the two weeks before the announcement. If the price rose, it is an indication that expectations are high, and this often resulted in a lower post earnings price regardless of how good the results were. The numbers in green in the table are lower than the stock price at the close just before the announcement, suggesting that expectations were high, or at least rising. You can easily see that green numbers prior to the pre-announcement close exist in every one of the four quarters when the stock fell after the announcement.
Selecting Companies to Include in the New Earnings Strategy: As a starting list of potential earnings play candidates, we might use all the IBD Top 50 companies which have weekly options available. These companies should have tailwinds of some sort, so there might be a bias toward higher prices after the announcement (especially if we find that the stock price has fallen in the two weeks prior to the announcement).
Here are the 18 companies meeting those requirements and their actual or expected next earnings date: