As a general strategy, whether you buy or sell the straddle of a stock the day before earnings to buy/sell it the day after is just gambling. You are praying for a move larger than the straddle value or conversely a smaller move if you are short. Unless you have a crystal ball there is no real way of knowing either of these two facts will materialise.
During the last earnings round for giggles I posted the proposal that we sell the AMZN, AAPL, GOOGL straddle with AAPL at a 10 times multiple. I think the 3 way straddle I proposed at the time was selling at 150$ - of course there were no takers. As it turns out and despite a massive move by AMZN you would have made a huge profit on that sold straddle (100$ or so). When NFLX was going to announce earnings, I was long an ATM call which I sold before earnings. The stock moved a lot but the volatility collapse was even greater so that whoever bought my long position had a 30% loser on his hands despite being directionally right.
Statistically the volatility collapse overcomes the stock move the vast majority of times. The problem is that when it doesn't, the losses can wipe out your profits of dozens of other trades. Presuming you can find mispriced options is also a gamble (at best) or plain condescension towards the hundreds thousands professionals that look for such mini-differences for a living. Straddles are very expensive and you lose two ways when you hold them through earnings - one side will lose from the directional move (or absence thereof) and the both sides will lose from the volatility collapse. Furthermore trading costs add up when you choose stocks with cheaper options and risk is higher when you have stocks with very high volatility to start with. The simple observation that the priced in move (as suggested in the linked article) is lower than the average of the four previous moves makes no sense at all. For one the author averages both up and down moves and he looks at percentages rather than standard deviations which would normalise the post earnings move.
What can sometimes be done is a detecting a pattern - I noted for one stock for example that - presuming normal results - it moved alternately possibly optimism Q1, disappointment in the middle, renewed hope in Q3 and dashed hopes on Q4. The key diffentiator is was - in my opinion - that in Q1 and Q3 the company gave outlooks which it didnt do in Q2 and Q4 - the management was always conservative. So there was a higher likelyhood of a better than expected result in Q2/Q4 announcements than there was in Q1/Q3. Such patterns are repeatable but can of course change the moment management changes the way it makes announcements.
In my experience the only really reliable strategy is that volatility increases as the announcement date approaches. If you are not unlucky in your timing - and this is a big IF - you can make money by seeing the option price increase simply by getting close to the earnings date. Strategy is then to sell before earnings - this can work with straddles or long positions and its worth carefully checking individual stocks to see what happened in the past and why. Its still hit and miss 1 in 3 of these guesses is wrong and the losses are greater than the winners but overall at 2 wins to 1 loss you can get ahead if you know to cut your losses double quick.
