Hello, everyone. Welcome to our breakout session and, in this presentation, Marija and I would like to discuss our research analysis on the momentum effects in the US stock market. We will start by covering a recent research paper that I co-author with David Turkington in which we analyse data of S&P 500 of stocks in the past 25 years and apply a coherent framework to evaluate various sources of momentum. In the second half of our presentation Marija will share her thoughts on the momentum strategy and how it links to the market outlook. As a reminder, please feel free to submit your questions at any time and I will leave some time for Q and A at the end. So, momentum has been observed for individual stocks and for aggregations of them. However, findings on momentum and reversal effects are often disparate. Differences in methodology and data have made it challenging to reconcile the various documented relationships. The central objective of our research is to attribute stock return predictability to a variety of distinct momentum and reversal components within a single, coherent framework. We analysed stocks in the S&P 500 universe between 1995 and 2020. We choose this large cap and highly liquid market to remove concerns that pricing anomalies are mere artefacts of market frictions. This also gives us more confidence that the relationships we identify would have been practical. We ran pooled panel regressions on individual stock returns. Our dependent variable is the total return of each stock for the following month normalised as a cross-sectional percent rank and centred to have zero mean. This normalisation mitigates any due influence of potential outliers as well as heteroscedasticity in errors. Since there are months in which stocks mostly rise or mostly fall, the cross-sectional ranking keeps our focus on relative winners and losers. We have, in total, 18 explanatory variables in the regression. There are two-time horizons. Twelve minus one month and one month lag returns and for each time horizon we consider four categories of price signals. First, we include the returns of each stock in excess of GICS Level 3 industry return and these would make up the individual stock level signals. Second, we include the returns of hierarchical industry portfolios according to GICS. These return signals are ensured in the nested fashion. So, we computer stocks, sector returning in excess of the broad market, its industry group return in excess of sector return and its industry return in excess of industry group return. Thirdly, we include size, value, investment and profitability factors as proposed by the 2015 Fama and French paper. Intuitively, it must be the case that if factor momentum occurs for groups of stocks, the stocks that compose the factor portfolio must inherit the performance trends of the composite factor. So, we create decile portfolios to compute their past returns and, lastly, for each stock, we include the returns of an index of peer stocks whose price behaviours are statistically similar. So, for each stock we measure the Mahalanobis distance of every other stock in our universe and so like the three nearest neighbours. The Mahalanobis distance accounts for the covariances of the variables and we use a factor of 24 monthly returns to describe each stock. Thus, two stocks will be closed if the spread between their respective past returns are small compared to the typical variance across all stocks and if the spread between their past return patterns are close to the typical pattern of returns that is observed across all stocks. This, statistically, defined factor
may identify dimensions of similarity beyond the traditional ones that we enumerated previously. In each case, we cross-sectionally rank the return variables to mitigate possible multi-co-linearity and enable the comparison of different effects, directly and in common units. To summarise the key results of the full sample regression, we can take a look at the t-stat that capture the sign and strength of each component. At the individual security level, the results suggest that 12-month momentum is subsumed by other forms of momentum but single stock, 1-month reversal is not. The influence of industry classifications and factors appear to differ. Industries and sectors exhibit strong reversal effects, whereas positive momentum appears to be dominated by factor cohorts. However, I would say it's premature to conclude the fact that factor momentum explains the industry momentum on this basis. Further investigation that segments of the sample, according to market volatility actually shows that industry momentum is a prevalent force and it is obscured in the full sample by occasional crashes. We realise that using the full sample for tests involves an implicit assumption that momentum effects are stable over time. There are good reasons to doubt that. As a strategy that benefits from
the continuation of price trends, one might expect stock market shocks to overwhelm the persistent trends. To investigate this issue, we segment our sample according to market volatility, measure it as a month's average VIX level. We run two panel regressions. One for the next month returns of stocks that coincide with low volatility which is a measure that's VIX below 25 and the other, when volatility is high and VIX is above 25. So, approximately 20 per cent of the sample would belong to the high volatility sub-periods. We
plant these two sets of regression t-stats. The low volatility results as showing in the blue bars are qualitatively similar to the unconditional results from the full sample. The high volatility results, as showing in the orange bars, however, shows a quite different picture. I will summarise some of the key observations for each variable category. For stock-specific effects, when we regress each stock's future return around the stock-specific lagged returns, we observe highly significant 12-month momentum and 1-month reversal. Much of the significance can be explained away by sector and factor momentum, when regressed simultaneously. When VIX is below 25, the 1-month reversal effects remain significant even after adding other variables in the regression. In fact, the stock-specific 1-month reversal has the largest t-statistics in magnitude in this sub-regression. The effects observed during low volatility periods become insignificant during highly volatile periods. In terms of industry and sector effects, during normal volatility periods, sectors, industry groups, and industries each exhibit significant and distinct sources of momentum at the 12-month horizon and reversal at the 1-month horizon. When market volatility is high, however, we see crashes of momentum effects from sector and industry
groups, excess return of sectors over market, for example, has a t-stat that is greater than 4. In the normal VIX subset, while its t-stat is below negative for when VIX is above 25. As for a factor of cohort effects, like sector momentum, 12-month momentum effects from size and value factors are highly significant when VIX is below 25. The regression results suggests that industry momentum and factor momentum actually coexist. Now, interestingly, unlike industry momentum, we do not see crashes of factor momentum, during highly-volatile periods. The statistical peer stock factor that we created, we believe is a potential source of momentum or reversal, because we focus on the most accessible and liquid stocks. It is probable that the stocks return patterns embed the market assessment of which stocks react to common shocks. From our regression, we observe meaningful residual 1-month reversal effects by closed peer stocks during low-volatility periods. To summarise, prior literature has documented robust, momentum and reversal effects. These findings, however, can be challenging to reconcile. We take a bottom-up approach and use panel regression on stock returns as a single vantage point, to evaluate a multitude of prior return effects in one, coherent framework. This approach facilities comparison because it allows for consistent assumptions. Time periods can also be consistent, as well as the relationships between components at the security level. We apply cross-sectional transformations and nested sets of excess returns to facilitate a clean
attribution. We find that the ability of past returns to predict future return stems not from one, but many sources. Stocks tend to follow the trailing 12-month returns of their industry and sector cohorts, but these trends are prone to crashes and are, therefore, easy to overlook in the full sample analysis. Factor cohorts also play an important role in predicting future returns and can coexist with sector effects. The statistical peer factor that we created, based on past returns, and the Mahalanobis distance showed significant 1-month reversal effects that is not explained by other variables. We took a pretty simple approach to identify the peer stocks and it already showed some promising results. We believe there are opportunities for more investigations and innovative ways to capture the meaningful residual momentum effects, beyond the well-known sources from momentum. With that, I'll hand over to Marija and we'll hear her thoughts on momentum and the market.

Thank you, Andrew, and thank you everyone for joining this presentation. That's been a great theoretical foundation for momentum stocks. I have to say, I work in a global markets and strategy team and the first I heard about presenting on momentum, my first thought was to say, 'Okay. I should talk about US stocks.' Indeed, I've been talking about it for the last five research conferences, so there is definitely a momentum. On the next slide, you can see, the reason I'm doing that, I mean, the US stocks have been going higher for, actually, longer than five years; for almost a decade. What I'm going to talk about in this presentation is my belief that this momentum strategy can continue. The reason for that is really twofold. The way I like to think about it, I first of all think that global stocks - and I'm going to demonstrate it to you - that global stocks are going to go higher next year. US is basically the best set of stocks you find in the universe. With that, let's jump to the next slide and what you can see is that it's really maybe like a step back to think why have stocks gone up so much this year? So, stocks are up a good 20 per cent as we are coming out of recession and there are really two reasons for that. On the left-hand side, you have strong earnings, on the right-hand side, you have very accommodative financial conditions and, really, it's a combination of those two actors allowed for very, very strong performance. What I'm going to argue, in this presentation, is that those two factors are actually, indeed, still strong and likely to continue into the next year supporting strong performance of global stocks. On the next slide, what we can see is earnings expectations. I mean, I do love this left-hand side chart. That shows the evolution of analysts' earnings expectations for next year and the year after. I have to
say, I've been doing equity research for about 20 years, almost, now and you rarely see a chart like that. Normally, what happens is that analysts start... I mean, we are all optimists. We all love equity stocks go up, so we start very optimistic. We have great expectations for the year. Then, as the year progresses, we kind of face reality and the expectations/earnings expectations come down to earth until we settle on something. Normally, those lines go from top left to bottom right and that's a reality. Every time you see earnings expectations beginning of the year, you probably subtract five/six/seven percent. Not what we saw, the last couple of years. We saw... As economists, as we exit from a COVID recession, analysts continue to underestimate the strengths of the recovery. What was really encouraging to me is really the right-hand side of this chart. Since the latest reporting season... Again, we had a great reporting season, the majority of companies exceeded expectations and guide it higher. We're seeing higher earnings expectations for the next couple of years. That's great. Earnings expectations are strong. What is really exciting to me is the right-hand side chart and apologies for going too much into the weeds. If you think about what we are observing right now in the economy, it's really hard to open a newspaper, turn the TV on without hearing about higher costs that companies are facing, higher energy prices, higher commodity prices, higher input costs, higher wages. All that, potentially, can hurt companies and what I'm really excited about with the right-hand side chart is the margins. Margins are still very, very high. What it tells to me is that economy is strong, there is enough underlying demand to support those margins, to support those earnings. If we jump on the next slide, what is kind of the justification
for that, left-hand side, global manufacturing PMI and the breadth of it. Looking back at this chart, we only had a few episodes. It was pretty much every single country was an expansion in territory, but the right-hand side really gives you all the answers. What we are seeing on the right-hand side is how rich are consumer and corporates right now. It's really unusual to exit the recession in much stronger financial conditions that come to recession... Think about financial crisis. A global financial crisis in 2008, it took us a decade
to recover the wealth we had. Look at the light-blue line on the right-hand side chart. Consumer wealth is proportional to net disposable income. Highest ever. All asset owners, people who own financial assets, people who own houses, all those prices have gone up. Those people feel richer. On the other hand, lower-income consumers, they've been very strongly supported by government programmes, by the checks done to consumers. Again, they have some money to spend. Obviously, all of us spent a lot less on services during the pandemic time when we had somewhat restricted mobility. Corporate, on the other hand, again, their balance sheets are very strong, as well. They had an unprecedent access to cheap financing. We know that credit issue has been at historically high
levels. Companies were able to borrow at a very, very low rate and at very long duration. Companies are feeling quite rich, as well. Companies... That tells me that the economy is strong. There is plenty of money to spend and that's why we're seeing stronger earnings. That's why we are seeing equity, more corporate margins remaining strong. Earning side is strong, so what about interest rates? Are we going to see financial condition tighten? If you look on the next slide, that's really the debate we are having in markets, right now. As I said, economy, as we're recovering, companies have money. What normally happens, at this time, is prices rising - exactly what we are seeing - and the normal reaction would be for central banks to come in and stop overheating and that tends to slow business cycles, hurt stocks. Is that something that's about to happen? I would argue not so much. If you look at the left-hand side, the left-hand side chart shows you our measurement of hawkish-dovish central banks. You can see orange bars; how it's changed since the summer. Pretty much every central bank, with a few exceptions, is getting more hawkish. Central banks are talking about, 'We need to raise rates,' and that's exactly what the market is pricing. If you look at the right-hand side chart and it's a slightly complicated chart, so I'll explain it. The lines here, the light-blue line is your dollar contract market expectation of interest rates which we've taken in the beginning of the summer. Darker-blue line, it's current. The orange area on the right-hand side axis is the difference between the two lines. Basically, what is fixed income market telling us? Fixed income market telling us that hiking cycle is starting sooner. Yes, central banks are getting more and more hawkish; makes sense, but it's also telling us that the hiking cycle will be a lot shallower. That, to my mind, is the most important driver for equity market. As equity investors, we're discounting future earnings at some interest rate/risk-free rate and it really doesn't matter when you start hiking cycle. What matters is how far it will go. Basically, right now, market is telling us that the hiking cycle will be very shallow. Great news for equity investors. Why hiking cycle will be shallow? The answer is really on the next page. It's because labour market still needs to heal. Labour market, it's an... I mean
it's very interesting. I find it fascinating that the new Fed mandate is to used monetary policy to improve outcomes in the labour market; not just get back to where we were and, actually, where we are now is at a worse position than we were... If you look at the left-hand side chart, participation/overall labour force participation is below pre-COVID levels and particularly low for women, for Black people, for minorities. Fed wants to use monetary policy to address those inequalities in the society and we are not even back to where we started. To me, that basically says that there is some way for... You have to be patient. More importantly, what you need to get a persistent inflation, something that our central bankers really need to push against, is wage spiral. Higher wages push higher prices and we get more wage inflation and so on, and so on. We are not there yet. Again, look at the light-blue line. That's real average hourly earnings; still falling, so we are not really in this very, very tight labour market that central bank needs to address very, very aggressively. Here we go. You have a situation where in the medium-term we continue to see economy being strong enough. That supports overall earnings, corporate earnings. On the other hand, central banks are not there to spoil this party. That a very strong foundation for global stocks to go up. This stock is about US stock. If you look at the next page, really, why US stocks not any other stock? That's my favourite chart. The people who follow our research have seen this in my presentation, probably for years. The light-blue line and the orange line are return on equity and you can see that US is so much profitable than any other market. Dark-blue area is a gap between the two. The very, very first chart of my presentation was that US stocks have started to outperform global
equity of about a decade ago. That very much forces gap open and continues to grow. To me, US stocks are the best, the most profitable stocks in the universe, so global stock is going higher. US stocks will go ever higher. The only pushback I get on this view is really on the next page, but aren't US stocks expensive? Yes, we know they're wonderful. We know, Marija, you love them but actually they are expensive. That's a very, very fair comment. I think if you take an absolute, yes, stocks are expensive... What I'd like to argue is
actually with the prevalence of so much very, very cheap financing - we have so much liquidity in the market - any asset, any risk asset is expensive. If you are in a world where you cannot invest and not worry about inflation, I don't think many people are, then, yes, US stocks look risky. If you are comparing US stocks to anything else, actually, they're really, really attractive. What I'm doing on this chart, I'm comparing yield on US stocks to bonds. If you include dividends and buybacks, it's almost double, so US stocks offering more. If you think about the arguments I've given you that earnings are growing, balance sheets are very, very solid, then you probably can argue that yield that US stocks are offering is not just safe, it's actually likely to increase, as well. To me, US stocks look as a fantastic opportunity to see the momentum growing. Really, in summary, on the next slide, yes, we have seen very, very strong momentum in US stocks and we very much expect it to continue. With that, I think I'll open up to questions. As I am a moderator of the section, I'll try to read them out. I'll give Andrew the very first question so I can catch a breath. The first question is: your regression results show that industry momentum effects reverse strongly during volatile periods, but factor momentum signals don't, so any idea why that could be the case?

Yes, sure. That's a great question. Yes, we actually thought about that and we discussed that in paper. Why does industry momentum reverse so strongly during volatile periods while a factor, I was saying, almost do not? I guess one possible reason is the relative stability of the industry composition compared to the revolving composition of size and value portfolios. Persistent momentum in the sector will inflate the value of stocks within it and since stocks don't change their sectors often, we'll continue to experience sector-driven momentum, until it is overvalued and eventually crashes. Persistent factor momentum, likely has a different impact on stocks because stocks that are included in the factor, eventually will rotate out to the extent that some stocks are temporarily
overvalued. They may correct more gradually. Therefore, we speculate that, perhaps, bubble-like behaviour in sectors would lead to more spectacular crashes than the similar bubble-like behaviour in factor cohorts, such as value and size factors.

This next question, I suspect that's probably for me. So, growth stocks have outperformed value for over a decade, until this year, so has momentum in growth stocks have been broken? And, yes, no, I do love this question. I've been advocating for growth stocks for a long time, so definitely momentum in there! I would argue that it's a little bit like to what Andrew was saying, so what is growth stocks? What's value stock? That keeps changing over time a little bit. To me, the key, key characteristic of growth stocks and what makes a growth stock is equity duration/how sensitive those stocks are to interest rates. With a prevalence of very, very low interest rates, the duration of growth stocks has been increasing over time. I mean, one of the arguments in my constructive view on equity markets in general, is that interest rates are not going very, very far, so our hiking cycle will be fairly shallow. That's an important driver that is there for growth stocks too. That's something we like. The other concern we have with value stocks is sector composition. The biggest component of value which was opposite to growth is financial stocks. For financial stocks, we're still very much concerned about their ability to make money in the current environment. Low-interest-rate environment is not great for them. What corporates have done very, very successfully in the last year was issuing a lot of credit. They haven't borrowed from banks, so bank lending is not as great. We're a bit concerned about that sector, really. Probably, we'll still maintain the preference for growth over value despite a bit of setbacks this year. Andrew, one more question for you and I think we are beginning to run out of time so you have the last one. The last question is, in your regressions, did you consider whether momentum effects mainly come from long positions in the winner stocks or from short positions in the loser stocks?

Yes, sure, again, it's a good question and I believe that this is the type of question practitioners will consider when they implement a strategy like momentum. Actually, in our paper, we did a further decomposition of momentum and reversal effects; by allowing for the possibility that those effects would differ when a signal comes from a past winner versus when it comes from a past loser. I guess, we did that by applying piecewise regression that separates each explanatory variable into two new variables. One is created by including all positive values - that is the top half of the ranks - to represent winners/past winners. We also set negative values to zero. Likewise, we created a second variable that represents only past losers. Based on the expanded regression setup, overall, the results show that momentum effects manifest disproportionately as negative return outcomes. This is likely to be driven by large, negative relationships that occurred for past winners during market crashes. Especially, when volatility is high, when we were in the subperiods where we observed this to go above 25. There, we see a lot of crashes from past winners. Yes, I think this is rather intuitive in a way that it is easier for the overweight positions to build up when the market is relatively quiet/when there is not much market volatility. This is likely to lead up to crowded overweight positions and we eventually see crashes due to some common market shocks. That will have an impact disproportionately for the past winners and past losers. 

Very quick question and I think I will probably suggest we'll take it offline, but somebody asked how to calculate the yield including buybacks and dividends. We do it on stock-by-stock level, but drop me a line and I'll walk you through the... I'll express it how to do it; either myself or your salesperson. I think, with that, we'll probably have to say, 'Thank you to everyone,' and suggest that you need to re-join the main stage for the next presentation. So, thank you very much.

Yes.