[MUSIC] Learning outcomes. After watching this video, you will understand the concept of post earnings announcement drift. You will also understand how to implement a trading strategy based on PEAD, post earnings announcement drift. [MUSIC] Finally, we are done with Pietroski. We have spent a lot of time. Don't think that every trading strategy we're going to spend three four hours, then it's not going to work out. The idea there was that you must be accustomed to this way of thinking, that's why you spend a lot of time as I've indicated before. Now we're going to quickly, all of the strategies from now, I'll quickly introduce the strategy. I'll tell you that the basics, the algorithm, main algorithm and how to implement it. So we're not going not spend as much time as we did with Pietroski. However, I strongly encourage you, by very, very strongly if you will, to read the paper fully. At least the algorithm part, the result part and then the abstract. At least this much you should read. Don't just start jumping into trading immediately after this module. Read those papers fully, do your own back testing, we will reveal some of results, but you do your own back testing. And then start trading. Now the strategy that I'm now going to talk about has remained a major challenge to market efficiency concept. Even the semi-strong one. So market efficiency if you recall is the notion that prices get immediately adjusted, sorry, every information gets immediately adjusted or incorporated into stock price. That is the notion. So this particular trading strategy which is known as price earnings announcement drift, PEAD. Is a major challenge to this notion of market efficiency. So what is this price earnings announcement drift? Ball and Brown in their famous 1968 Journal of Accounting research paper discovered this trading strategy or this idea. As I've told you before for them this is an idea. It's more like a concept. For people like you and me who'd like to trade, this is a implementable idea, a trading strategy. So what they found was that, suppose a company comes out with a good news. Now, recall if markets are semi-strong efficient then what should happen for such a company's stock price? Yes, you guessed it right, instantaneously the stock price should adjust, right? If the news is good then the stock price should go up. And if the news is bad, then instantaneously stock price should go down. But what Ball and Brown observed is that that is not the case. Even after you know the information, even after the information is revealed publicly, the stock price takes a lot of time to reach the new equilibrium level. In other words, prices get incorporated into stock prices slowly. At least, not as fast as strong from or semi-strong from a market efficiency will tell you. Now as a concept this is very interesting, but what would I do as a trader? This is a test for you. By now you should have thought about the trading status. Just pause for a while and think. If you know that publicly revealed information does not get reflected into stock prices instantaneously then what kind of trading strategy you will employ. Think about it. Suppose a company announces a result, and if the profit number is better than expected, then what will you do? Yes, even after announcement of the result, there is scope to buy and hold and still make money. That's what this concept is telling us. Similarly, if a company's announced negative results or not so good results, then it's good to make money after public revelation of information if you short the stock. Now the big question is, all this is fine as a concept, but how do you calculate this? How do you say that a particular income number is better or worse? See for a company we just expected to report say 100 million in loss. If the loss is 50 million it could be good news, not a bad news. And if you go and short that company you may lose money. Similarly, a company which is expected to make a lot of profit. If it makes less profit it may be bad news. Just because the profit number is positive or a profit number is negative you cannot just think that the drift is going to be in that direction. This is not like Pietroski mind you. In Pietroski it was very simple. All that you require was that whether a particular number is positive or is it increasing, that's all. Can we look at increasing? Can we say that like Pietroski did? Just recall delta array. What do you do in delta array? You calculate, you can pause and think about it. While I do a quick recap. What do you do in delta array? You calculate array for a particular year. You calculate array for the previous year, and see whether their debts increased. An increase is considered as positive. It's simple, straightforward, right? Can you do that here? Can you just say that if the last year profits was 100 the sale profit is 110? So then the positive sale price and hence the stock is going to go up. Can you do that? [COUGH] The answer is no. What is important is whether the stock price, sorry, whether the earning number is better than expected. If they expectation was that the stock prices, the earnings is going to grow by 20%. And sorry I keep, I'm saying stock price, it's not stock price, earnings. If the expectation was that earnings would go up by 20%, then if the earnings, in reality if earnings go up by 10%, then it's a disappointment. So then how do we trade using this knowledge? How do we come up with a effective trading strategy using price earnings announcement drift? So there are two ways you can do this. One, is a concept known as standardized unexpected earnings. It is also called SUE. Don't worry it sounds very technical and complicated, but actually it's not. Very simple and straightforward. You don't need to think that there is some crazy calculation you ought to do, none of this. All that you are to do is some addition, and some multiplication, and some division, that's all. Nothing crazy going on here. How do you calculate this standardized and expected earnings? So let's start from the last part of this unexpected earnings. Now, using this method, the way you calculated unexpected earnings is, first you calculate the average earnings per share in the past four years. Suppose you are doing, you are in 2016. So take the, EPS I'm sure you guys are familiar with. EPS is earnings per share, if not it's very simple. Just take the earnings after tax number, I showed you a pro forma balance sheet as well, take that earnings after tax number divide it by number of shares, you got EPS. That's all, nothing fancy about it. So take the CPS number for previous four years. Take the average of that. As you're learned, this is the expected earnings. Then, take the actual earning for the year. So now, we have actual earnings. And you have expected earnings the gap between them is unexpected earnings. Now the unexpected earnings could be positive or could be negative, right? Suppose the last four years average is say, 15 and the current number is 20 then you have a positive 5. On other hand, last average four years is 15, abd the current number is 10 you have minus 5 straight forward. Now using all you have learned so far just imagine, is this information sufficient? Are we missing something? Can you think of instances where this kind of classification could be misleading? Let me give you an example. Suppose a company's expected earnings is ten. And the actual earnings is say, 20. Again, how did I get the expected earnings? Average of last four years, that's straightforward. Now the unexpected number is what? Ten. Think of another company who's expected earnings is 100, earnings per share, now, why is it higher? It could be just because the company could have issued less number of shares and nothing else going on. And nothing to do with fundamentals. A company would have a $10 face value share, another one may have a $1 face value share. Right, so if the expected earnings per share is 100 for a particular company and the actual earning is 130. What is that unexpected earning for this company? 30. Unexpected earning for the previous company was how much? 10, now tell me whose performance is better than expected, significantly better? If I ask you to rank the two of them, the first one unexpected number is ten. The second one is 30. But you know that that's not the case. The second one has just gone up 30%, the first one has gone up 10%. Right, so we have to find a way of scaling this. You can't just take the difference. Now, one option could be, why don't we scale it by the level of EPS? In other words, we can scale it by say, 130 minus 100 divided by 100. Which will give you say 0.03 or 30%. And so for the first one we can scale it by 20 minus 10 by 10. Which will give you 100%. Now we know that the first one was done better than the second one. Agreed that this is more sensible than the approach that we did earlier. But we're still missing something. We are missing something very critical component of a stock. And our traders, variables that our traders should consider, very, very important. Can you pause for a while and guess what is that we are missing? We are missing variability? We are missing the standard deviation. What if the first company has very high variability? Very high standard deviation. That means deviation from the expected number is quite natural. It's actually a part of the expectation itself. A company with high variation can produce high or extremely low number which can be a one time affair just because of the fact that earnings are volatile there. Think of companies who's earnings streams are extremely volatile. Versus companies which are quite stable. So a company with a highly volatile earning stream, suppose in a year produces very high earnings. It does not mean that it has, it has perform significantly better than expectation. Remember this, I kept telling you, even during Pietroski I'm going to repeat it now. What are we after? We are looking at sustainable performance. Expect a sustainable performance in the future, right. So a high variability company coming out with a wider margin even after scaling for the initial level, this is not a sale price. So we have to somehow account for the variability. And the best way to do it is, consider the standard deviation. So, what you need to do is that take the actual number, subtract the expected number, and divide the same by the standard deviation of earnings. Now, how do you calculate the standard deviation? You already take on this last four years earnings numbers, right? Just calculate the standard deviation of these earnings numbers, you know the standard deviation anyway. It's very simple, those of you that don't know can just figure it out. So get the standard deviation number, and then take the actual number, subtract the expected number, divide it by the standard deviation. So what you get is called standardized unexpected earnings. Once you have the number what to do now? As in the case of Pietroski, how do you implement this?
