[Music] welcome to the me favor show where the focus is on helping you grow and preserve your wealth join us as we discuss the craft of investing and uncover new and profitable ideas all to help you grow wealthier and wiser better investing Starts Here meth bber is the co-founder and chief investment officer at Cambria Investment Management due to Industry regulations he will not discuss any of Camry's funds on this podcast all opinions expressed by podcast participants are solely their own opinions and do not reflect the opinion of camber investment or its Affiliates for more information
visit camria investments.com 2024 may be coming to an end but I want to talk to you one more time about our friends at y charts if you're a financial adviser you know that time is one of your most valuable assets between prepping for client meetings creating proposals staying on top of market trends and managing portfolios your days are packed that's why the tools you rely on have to work seamlessly and why chart's quick extract does exactly that quick extract is the ultimate timesaver for Port folio analysis it lets you upload client portfolios in nearly any
format PDFs spreadsheets screenshots that's amazing so you can skip the hours of manual data entry and dive straight into analysis in just seconds quick extract sets you up to focus on what matters most high impact tasks building client relationships and driving AUM growth if you're ready to see how quick extract can transform your practice click on the link in the show notes to learn more and get 20% off your initial y charts professional subscription when you start your free y charts trial if you're a new customer where else are you going to find information about
shareholder yield go sign up today and tell them that Meb sent you welcome welcome everybody another awesome episode today Christmas markets are out we got our guest Matias hanau he's researcher on robo's Quant equity research team his areas of expertise include International Factor Premia stock selection research today we're going to talk all about factors in the setup for one of my favorites value to heyas welcome to the show thank you map thank you for the invitation it's a pleasure being here we've probably featured more Robo pieces and pieces of yours in the best investment writing
in the idea Farm over the past year or two than just about anybody so kudos to the amazing work you guys have put out I think I said at one point that your piece about the top 10 charts about value the report shorts of my death have been greatly exaggerated was probably my favorite chart book of the Year you guys did this in January and like the investment gods they love to laugh at all of us and of course and I think the S&P as we're recording this is who knows up 25% again this year
walk us through this post so what are you guys talking about here and kind of maybe talk about some of the charts in your head we'll interlace them in the YouTube video for listeners who are doing this on the podcast as well as in the show notes talk to us what's the setup this post started several years ago in 2020 when we were really in the mid of the value crisis the Quant winter when a lot of people said like value investing is that it doesn't work anymore just because the realized returns have been really
really bad and then we decided to write piece on it yeah explaining how we see a value that it's not only book to Market but the time series that French has on his website but also that maybe there are other metrics that you can look at Value investing and also some more risk management like saying making industry neutrality for for Value investing these kind of things and we started writing it down in the paper with David Blitz where we said resurrecting the value premium and actually suddenly more or less put it on SSN when this
value Factor had really its low in 2020 and then had the first graph shows this long-term performance starting in 86 through the beginning of 24 and really we posted this graph at the low and afterwards we see a nice resurrection of the value factor out of sample and this is just like the long-term evidence we also see that there's a major draw down not only in 2000 in the buildup of the.com bubble but also over this period 2018 2020 and so when you guys say value what does that mean for you you talk about things
like the enhanced value Factor yes you're right it's not uh about just one single Dimension we look at several dimensions of valuation so we always have fundamental value divided by the price and the first one is actually book to price but not the simple book to price as farmer and French have it but we also take some of the Investments or the the costs like investments in inanga capitalize them so it leads them to higher book values for firms that are research intensive but we also mix it blended with signals such as ibita to e
that's quite popular in the private Equity world or cash flow to price yeah looking at different financial statement next to the balance sheet income statement and also something that we call Net payout yield that is dividend yield plus net share ISS so similar what you call I would say a shareholder yield walk us through what it's saying right now or what it's been saying for the past few years so you've had obviously everyone knows values worked for a really long time you had this bad period potentially binging out at the craziness at the end of
2020 2021 start but people look around and they're like yeah but me the S&P is up 25% this year 30% Whatever It Is by the time this publishes you guys just don't get it what's it been saying the last couple years and what does that mean going forward well value stocks by this definition are cheaper than Crow stocks but there's some variation over time and what you see in the second chart of this post was yeah that the valuation of value stocks when you looks at something like the forward PE ratio yeah it was around
8 to 10 on average and for the expensive stocks based on this definition they were also having valuations of forward pees of 20 around on average but there were episodes in the time where really this number went up to 30 40 50 and that that we have seen during this Quant winter and this explains why value stocks were underperforming because because the expensive stocks were getting even more expensive and when you divide these two times here you get something like the value spread and we saw that the value spread was like two three times higher
than it used to be And to clarify one thing this is like the difference between the average value stock and average growth stocks so even like last year for instance and also partly this year it is that the average value stock was doing similar or better than the average Crow stock but yeah there was the big tech stocks that are really dominating the S&P like Nvidia Amazon Tesla these kind of stocks that they are not in the typical value portfolio but they were outperforming the rest of the S&P 500 and therefore maybe the typical value
strategy not having in this stock were also underperforming S&P but it was not that value itself as a factor was not working it was more that like size as effector wasn't working and you guys are analyzing this globally correct because like a lot of times I mean I'm looking at figure three which is the valuation spread for value and so often when you think back to the US charts you think hey you have the Mount Everest is 1999 2000 and then maybe 2020 2021 might be aen kaga or McKinley one of these other mountains but
in this case the valuation spread is even wider During the period give us a walkr on that is that largely due to foreign and emerging or what's influence there no I think difference is not that it is like in the US less pronounced than in the other markets I think one of the enhancement what we have for this enhanced value factor is that it's also region sector neutral so we don't compare the utility stock in Japan with a tech stock in us so we tried to compare Apple to Apples and I think what we have
seen during the com bubble was that certain sectors were also outperforming other sectors like Telecom media technology so there were more sect effects but what was different in 2018 to 2020 was that also within each of the sector the more expensive ones were outperforming the cheaper ones until November 2020 and then we saw a strong comeback of the value stocks also across or within each sector and it was less a sector allocation thing but more really a selection within sectors I think this explains this difference we can run the same chart without these region seor
neutralities then I would expect that these two peaks would be more at the same height a lot of people say great we know this you could have told me this last couple years what does this mean going forward the fourth chart is value Returns versus value spreads and that looks like a nice Quant would say a nice regression you got a nice r squared on this where do we kind of stand today and what does that mean going forward so what you show here is actually that the realized value returns that are on the y-
AIS mainly explains by the changes in the value spread yeah so half of the variation is explained by it and when you look at the intercept of these lines where the dotted line crosses y-axis so this is about 10% and this 10% is even higher than the realized value return over the whole sample period and this can be explained that the starting valuation for Value was more normal than it's still now so the spread came back recently over the last 3 years but it's still higher than an average and if you think that the value
spread will normalize so converge to the historical average then I think this would be further Tailwind for the value Factor but even if you say there are reasons that these valuations structurally stay at this higher levels the difference then I think it's still the structural value premium coming from let's say the migration of cheap stocks going into value portfolios when they become more expensive dropping out the carry component the higher dividend yield these kind of things that still this part you should expect to stay high but it can also be of course if the valuation
spread widens again that we have uh we years for Value coming in the next years listeners I I always liken this you know you can look at this on markets in general too where you have these very long regimes where you may have say the whole entire US Stock Market think back to 2009 looking at something like the cape ratio was trading in the low teens and today in the high 30s and this multiple expansion there's a huge Tailwind of performance additive to what's just going on with the rest of the market and the opposite
is true as well if you go back to 1999 and we're at 44 and then ending in 2009 Lowes that's a huge headwind and so it's the same thing with factors right where you get these value spreads can be a Tailwind or a headwind as well so what's the same going forward so it's hard to predict where the valuation spread is going but I think it is more helpful for understanding what is happening because back then in 2020 people were really asking is there a reason why it stopped working it could be that expensive companies
had then really higher realized growth than less expensive companies so there was a reason that this value Prem was gone away but actually as spread widen something like fundamental performance was not really explaining it or let's say that over the last 20 years value was getting crowded so that value stocks are less cheap than they used to be and Crow stocks less expensive so actually this is the opposite what we see so this is more for us helping to stick to value in this case not giving up because we saw the explanation where it was
coming from I think it's really hard to say where the spread is going I would say the a rational assumption would be that the value of spread stays where it is currently and then you would expect like a premium that is close to the historic average a lot of people always say what about interest rates interest rates interest rates value does better when blank what' you guys find what do you think I think this was really a very popular explanation in 2020 and 21 when a lot of people said yeah all these growth stocks they
have longer duration because a lot of their future earnings yeah will be 10 20 years in the future and these value stocks they have now the high dividend yield so they have a ler duration and therefore grow stocks profit when rates are lower so they are getting this counted with a lower rate and we actually saw this pattern really during 2018 2020 so when the treasure yield was going up value stocks were doing better but when treasure yields were going down grow stocks were outperforming but when you really look at the long-term evidence this is
like all the blue dots like this all the evidence from 86 to 2017 there is hardly any relationship between changes in interest rates and value returns and also in the period after 202020 you you see little relationship so there was some correlation during this time I think maybe it was also a bit self-fulfilling at a certain moment but I think there's no long-term structural relationship between interest rates and value returns and if people have an opinion in which direction interest rate should go they shouldn't play this via value versus grow but they should play this
via other instruments that are directly playing on this move yeah I think that that's a take that is probably a lot of people might disagree with I like it I think it's accurate I think it's one of those traditional examples of sounds good when you describe how interest rates might affect certain types of companies but I think it gets a lot more complicated once you start to look at the data too and then you have different regimes as well and then you have different not just interest rates but relation to inflation at various points in
the cycle a lot harder I think than than most people make it simplistic one of the things that everyone is listening is probably say yeah yeah yeah but again growth growth gross growth your chart 7 I think is a pretty and it kind of dovet tales with chart nine it's all been about the mag 7 type of companies the big companies the growth companies what have you guys found with sort of expectations I mean looking at some of these companies today paler and doing the old malbon thought experiment where you work backwards from the evaluation
and try to come up with what sort of growth they need just to reach these sort of multiples what did you guys find with this kind of gross stock concept well there's always a reason why certain stocks are more expensive than the others and usually they have more better prospects a better Outlook also had maybe a better growth historically and what this shows is the past and expected growth for value stocks that are the blue lines and grow stocks that are in red lines and the solid lines they show the past realized growth in re
and the dash lines they showed the forward-looking growth expectation from analyst and you see that at the moment you form these value versus growth portfolios the grow stocks had like a 4 3% higher growth historically and also expected versus the market and value stocks had lower expected and realized path growth but the longer you hold these stocks so we keep the sample fixed and just wait one two 3 four years and you see that over time these expectations converge as also the realized growth rates converge so if you compare the solid line at month 60
so this is the real SCE turns or growth over the last five years with the expectation so this is the dash line you typically see that the realized growth is lower than the initially expected growth and I think this is the more the key reason for the value premium and not interest rates and ctivity or these kind of things within this wrapper everyone talking about value there's also this reality that us is scamed foreign and emerging as well and so looking at this Emerging Markets versus World talk to us about figure 10 what this means
the Blue Line shows the relative performance of the msci world versus the msci Emerging Markets so if the line goes up Emerging Markets are outperforming if the line goes down the developed markets are outperforming what you see this relative performance is volatile starting at the 2000 emerging markets were outperforming until 20 10 but for the last 10 to 15 years the MSC world so developed markets were outperforming mainly due to very strong us markets and you can say they have been realizing higher growth they have higher earnings growth what is partly true but also the
valuation component plays a big rol when you look at the Orange Line so this is like the value spread for emerging versus developed markets and you see that these two lines really move nearly parallel and I think if you look at oneyear changes in the two things then the valuation changes explain about 75% of the relative performance so this indicates the big part of the relative performance is not explained by fundamentals but it's more explained by valuation changes and I think these are less sustainable in the long run than fundamental differences so if you expect
of course that developed markets will stay similar expensive compared to emerging market markets then maybe we don't see any relative out performance of Emerging Markets versus developed markets or if you even think that they even become more expensive then merging markets are supposed maybe to underperform further but if you think that these valuation to frenches convert more towards the historical Norms then we would expect that Emerging Markets actually are outperforming developed markets and I think early this year we have seen this partly but with the last weeks with the elections in the US I think
we have bit developed markets again outperforming Emerging Markets we just had this election in the United States and you had this massive value jump on the week days after what do you say to people when you try to say hey has this inflected or what do you think about the predictability of how this might play out do you have any good suggestions for the the listeners out there I can only guess here but I think one Catalyst be now the market is really optimistic to all things related to AI I think they they see it
as a growth opportunity saving costs having higher revenues but it also means that companies do a lot of AI they also have invest a lot so if this perception at certain point changes that people more focus of the challenges than opportunities of AI that might be something where a turning point for the big tech stocks for instance and I think it's also that maybe we see some more really fundamental growth earnings growth in markets outside the US that could then also trigger comeb back of Emerging Markets Europe versus DS for instance we often try to
describe to listeners it's not always a binary hey value versus growth type of decisions there's usually a lot of overlapping Vin diagrams so for example if you look at something like emerging or foreign they may have more value opportunities or something that I want to lead into now is size people would say well small caps versus large cap and there happens to be I think probably a decent amount of overlap with what we're talking about now you wrote recently at the end of June you're like the value spread between small large caps levels not seen
in over 20 years offering a multi-decade opportunity for investors so let's talk about that I think if you look at it I think DM versus em and large versus small cap show a lot of similar patterns I think there's relative performance there are Cycles I think there have been Cycles where small caps have been outperforming large caps I think if the long run we see outperformance of small caps we are a bit critical on size as a standalone Factor because if you control for the higher risk then yeah size itself it's not a factor Standalone
does not mean that you shouldn't have small caps in your portfolio think my starting point is that you should hold the market portfolio and an allocation of 10 to 15% dep depends a bit on how you define small cap should be like in your strategic ass allocation and then there are maybe technical opportunities that you can say because of the current valuation I want to overweight small caps and I think this is what we pointing to in this Insight that you just mentioned because of the underperformance of the last 10 to 15 years of small
CS versus large caps we see a big part of of of that is driven by valuation changes and this might be an opportunity that uh small caps might come back for the next 10 year and then is of course as you said it's not binary it also depends a bit how to do small caps I think small caps you have a really big breath you have a lot of dispersion in stocks you have a lot of unprofitable stocks but you also have several profitable stocks and with good quality low valuation low risk and actually there
the playing field is much bigger than large space so factors also working in large capap space but typically what you see is that factor premium are higher in the smoke up space it's interesting because you look at sort of known truths that have been in our industry for decades and so when I came up I feel like most people talked about small caps almost like a totally separate Factor where you saw the long-term ibbitson chart or maybe it was in stocks for the long run I don't know and it was bills bonds stocks and then
small cap and small cap was like outperformed everything and if you were a risky young investor you wanted to put a lot of money in small caps but I feel like the general Vibe and understanding is that it's not really a unique factor in the sense that it's like Alpha you can magically get but rather as you mentioned it's somewhat of a risk allocation and people don't think that small caps magically outperform I feel like that's now the common take on it talk so I feel like this is a good lead into to the factor
zoo I think a struggle with so many investors is they go to the grocery store and it's just a limitless amount of cereals you can buy it's like factors it's like where do I even begin can I just choose between Cheerios and grape nuts and not only that it's harder because the store also says you can mix Cheerios and grape nuts and Froot Loops so it's not just one factor it's 100 and so you guys had a paper called the factor Zoo maybe tell us top level abstract kind of what this paper was about and
then we can kind of dive in on some of the topics too yeah you're absolutely right uh you can think about as crossies shop maybe it's more a candy shop where you can find a lot of sweet stuff and in the academic lure there are like more than 150 maybe 200 300 reported factors or signals that claim to explain the cross seure of stock returns and I think this is in a big contrast to like academic Factor models on the other side that have a handful of factors like the former French 5 Vector model or
other models so the question was really is it 150 is it five or is it something in between what we did in this paper is We examined how many factors it really needs to explain this Sue of 150 factors that we had in the data set and turned out that 15 factors were enough to explain all of the available Alpha of this Factor so so it's something between the five factors of the common AC academic asset pricing models were not enough but it's also not 150 independent factors because there was a lot of overlap several
momentum factors several value factors several factors measuring earnings quality these kind of things and typically then you pick only one of two out of these categories imagine how many times you've been asked this question but I guarantee listeners are thinking in their head what were the best ones I understand that it is total data mining but I still want to know what were the best ones the best one turned out to be cash-based operating profitability and this is like an enhanced proability factor of the farma French five Factor model so it looks at Topline profitability
so not button line like Roe but more something close to cross profits and then translates it into a cash based version so you get all kindly the earnings quality AC crueles Factor Blended in and so this is an factor that is quite persistent so easy to train you don't have a lot of turnover and the nice thing it nicely diversifies with value because typically these stocks with higher probability are also more expensive and it's not too related to other factors like momentum so it's really another dimension of stock returns and then I think the next
factors on the list were like also fundamental factors related to acrs earnings quality but also valuation and momentum and what was very interesting there was that residual momentum was picked up as one of the first momentum factors that is a factor that was discovered by my colleagues in a paper in 2013 how should investors think about putting the factor to zoo together I think I would start really asking yourself what are my beliefs in which type of Concepts do I believe is it valuation is it momentum do I think that there is added in like
saying analyzing bance sheet and income statements and then maybe you have your first categories that you want to start with say value momentum quality something like shortterm depending a bit on your trading possibility if you have higher turnover you can also blend in more short-term factors like short-term momentum short-term reversal or seasonalities and then maybe you pick one definition that you find most reasonable maybe you find academic evidence and then you have a running model and then it's a bit about launch and iterate then maybe you start fine-tuning you do some research can I enhance
my value definition can I broaden it a bit maybe you don't want to rely on one single definition but you want to have two or three signals measuring that concept like valuation doing the same on momentum and quality and so on depends a bit how much capacity you have in your team fine-tuning these Factor definitions enhancing them or finding new categories that add autal Alpha to your process this is for instance also what we see in this paper is that when you start the analysis in the 80s when only with these factors that were published
up to this point then you were choosing other factors than the 15 that we find in the end and then you see credly over time when adding more and more factors that newly published factors then come in and older factors drop out yeah some old ones also stay in but this is also like the evolution in the factors so you can think about that and this is also what we try at riko we continuously try to enhance our models researching can we take out unrewarded risk can we find new Alpha sources this is what we're
doing if you look at a lot of the American multiactor Funds or you pick a stock that screens well across a lot of the traditional Chicago PhD metrics and you type that stock into any of the online sites I used to say Yahoo finance but anyone why charts probably my favorite now and so you type in some of these stocks and then it's just a laundry list of the same large Quant shops in the US it'll be lsv or aqr Vanguard depending on who's doing the the factor things and so really you either need to
come up with factors that are somewhat Timeless or new or different meaning hey don't be like Matias and publish keep it to yourself you know maybe it's a super secret factor but we're this new age which I think is a good segue into kind of this next topic on what you guys have been writing about a lot which is a little more modern talking about machine learning talking about AI how to incorporate sort of new processes into what it is you're doing so maybe walk us through what you guys have been writing about here with
machine learning AI is also a big topic for us of course and you can maybe say II is a big topic and parts of it is like machine learning more statistical learning other things are like natural language processing where you analyze unstructured data such as text and translate it into numbers that you can use them later for your Quant models but one Avenue that we were going was machine learning and I think this is more an enhancement to like the previous process where you combine factors in a linear way when you do machine learning you
can also start with a very simple all s model you say Okay instead of choosing the factor weights by hand I let the data speak and my alls model says okay this weight goes to this value signal this weight goes to this momentum signal and so on so a bit more data driven but the more complex models can also capture things like nonari and interactions in the data and are there a way to enhance a bit the performance of your models because you can also identify larities and interactions manually and just PLU and blade them
to the model but I think the model has much more freedom to discover these interactions and nonlinearities and I think we have been doing writing some papers in this area we find that yeah also these machine learning models are not immune to like performance 2K so you see that in the US a lot of the traditional factors had a low performance after publication after 2004 where there is maybe more liquidity in the market interestingly the same factors don't show this Decay internationally so this is another reason why you maybe should think about investing internationally because
Factor Decay seems to be the Lesser concerned than in the US but also we have been looking critically at machine learning strategies because there are a lot of design choices in these papers that have to be made and there's not like a clear consensus what you can do so for instance people some of the papers do rolling Windows versus the expanding Windows when they fitting the models what do we explain is the return of a stock over the risk free rate or should it be better the return across over the market if you're interested in
relative returns I think this might make more sense to do like a pre feature selection these kind of things and what we found out in this fresh paper is that when you perate over all these design choices you get thousand models that are all feeded with the same data evaluated in the same way and what was quite astonishing was that the worst models over like 50 year 50 years out of sample period have a return average return of 0% nearly and the best models have returns close to 2% per month so this is all before
transaction costs but I think this shows an immens of dispersion in returns and that people should focus at design choices explaining their choices why they do it and have a clear economic reason why they do that I think people probably perked up when you said 2% assuming you were going to say per year and then you said per month and so when you see the output of some of the how often do you have to mentally or physically constrain what's happening because you're like oh hold on this is crazy output or this is actually really
interesting wow we just found this brilliant new insight that no one knows about so let's do it how do you kind of balance that sort of real world application with what it's spitting out I think this 2% is a really high number but this is really the best out of 1,000 machine learning models so would have fre Iz this return before transaction cost if you would have picked out of these 1,000 design choices particular this best one and this I think is highly unlikely and I think you shouldn't then also when showing to your clients
or Prospect like a back test you shouldn't pick then this one single best back test but maybe say okay we know this played out very well in the past but maybe it might be a biased we just diversify over different design choic and this gives maybe a more robust and or more realistic expectation for the future outperformance of the strategy but on the other hand you also have to think about it okay which design choices really make sense if I have a model that has to estimate a lot of parameters because it can capture more
complex relationships non larities inter actions then you need a lot of parameters that you want to estimate and then it might make more sense to have an expanding window than a rolling window and we see in this paper that some of these design choices really give a significance structurally higher return and others make maybe less sense and so how do you balance that as far as like if you're going to implement it where you're like letting the model run wild it's like I think to the early days of looking at whether it's machine learning or
AI applications where people look at and we're like these moves like it's not even something someone would consider and then once you start to go through it you say okay well actually this does make sense or on the flip side in our world there's so much with perception and I think a classic example I love to talk talk about we say you can't look too different and so we talk about something like even Tren falling and you run a traditional Trend falling simulation through the various optimizers almost always it kicks up a large allocation to
Tren falling yet no one is willing to do that allocation right they say oh no no no we got to constrain the model that's crazy so how do you think about actually implementation do you have some strategies where you're like let's just let this run hog while or do you guys kind of take the output and say okay let's put our PhD sort of real world Street smarts on top of this and figure out something that's actually pable that someone might want to invest in so I think there's really a strong research protocol kicks in
I think we we start feeding the model with features where we think they economically make sense and then also in the terms of design choices before testing all available options so we did this in this research papers with two students from Technical University of Munich but this is not how we did it in at rubiko so really pick design choices initially we really think this makes sense from an economic perspective and then we run the models back test them but then usually we blend the different models Ensemble them on one side we see that ensembling
is like diversification and machine learning that is the free lunch but also we won't just pick the best model historically because we are unsure if this model is also the best model out of sample it's similar like we won't take one momentum definition that is maybe the best one historically for this sample and just pick then we also have their several signal measuring it similar with value and this same philosophy also holds for this more Advanced Techniques fun so before we skip off from machine learning any final thoughts on this topic about what you guys
are working on or what can we expect to see coming from you guys in the coming months so well Machine learning is one area as I said before this is more an enhancement of these more linear ways of combining factors for me I think we can maybe find more aoral information unstructured data it might be text Data where we can with the help of NLP get some new information out of it it can also even be the next level from text maybe Voice or maybe videos like seeing the faces of the CEOs and CFO and
earnings calls how they react to a question of an analyst for instance so this could be something very interesting to look at and also kind of alternative data so there's so much alternative data around we're looking at at various sources it's always a bit a challenge because some hot data you don't have the long history to back test and then you have maybe a really a clear economic idea why this should work and what we usually also really like is that we have a lot of control how to combine the data we don't like if
a lot of signals are already pre-processed because we don't know what was like optimized under the Hood from this data provider so we like to have the raw data and do it then ourself the the final signal as we wind down 2024 give us a preview what else is on your brain what else are you thinking about are we just going to be doing this annual value update for forever anything else that Matias is working on that he can give us the preview to for the future so the thing is I like to doing this
value updates but there's a lot of more than value that we do at Ron and what I do so but it was very popular by by the people so probably I will do some update because just people want to see it but we not just value we are also momentum quality and a lot of other propriety signals that we developed over the last 20 years but probably at some point I will do an update again momentum you're speaking to my heart we're almost like at the back end of the podcast and momentum finally comes up
one of my favorite metrics are you guys still on board with that one is that still in the quiver yes so we have several ways how we measure momentum it can be the pure price momentum that was working really well the last years but you always have also the situations like in 2009 or in the 1930s where it often happens after bare Market when the market turns that you have these momentum crashes of course these ones you want really to avoid and we found some techniques to limit these crashes giving up only a small part
of the premium and if you think about momentum as a concept then maybe you think there is credal information spillovers then you may also want to look into the other sources where you can play momentum being it news price momentum from other markets that you can exploit so it's just more than a pure price momentum strategy I like it one of our favorites forever I think all right so give us a preview what else you working on what's on your brain as we close out the year the continuation of what we have done over the
last years I think Rika was known really as a classical Quan shop with value momentum quality but over the last years we have been really going into like machine learning NLP alternative data and I think we want to go into this direction in more we spent a lot of time investing our infrastructure building pipelines that we want to leverage and now it's much more efficient analyzing these potential signals we also thinking about more contexture models do we want to have the same models for all stocks or do we want to have some more Regional or
industry models so this is one way we want to go into more fast models so we have already a lot of short-term signals but can we even go go faster and then maybe also not for the existing products but also having products that are faster than our existing ones these are direction that we're looking but also enhancing the existing classical factors like value momentum quality what's something that you would say that most of us would disagree with doesn't have to be me because I'm crazy but you sit down a group of your friends at the
conference room table what's something that you would say say that most your professional friends would not agree with you on I think we started the podcast with like how to define value and I think a lot of people in the Quant investing World maybe also fundamental investors they say like yeah there's various ways to Define value but they probably say book to price one of the oldest one is among the weakest ones and I would say that is not true I think when you look at the Standalone performance book to priz is maybe weaker than
te like e to EV or shareholder yield but when you look at the marginal contribution in a multiactor model where you also have quality so items like signals like proability or momentum in your model then I think that book to price actually gives you the highest marginal value because it's most negatively correlated to the other factors so I think you shouldn't stick to the most basic book to price definition you can also enhance it by incorporating in Tanga bles into the signal but I think book to price is really in a multiactor context a really
good signal what's been your most memorable investment the best investment or for me I think was doing the PHD so when I was a master student I thought doing I was doing a several internships I thought after my master thesis I will go somewhere in Industry maybe Insurance banking as a management wasn't that sure which direction but then writing my master Theses I analyzed the former French factors for the German market so this was my first contact with uh Factor investing or asset pricing and I really like to talk topic and of course then during
the PHD broadened it went into international data us data but I think doing the PHD really helped me to think about problems formulate them thinking about how can I empirically test them looking at the thoughts of others reading their papers and then really writing down my own thoughts and and this really helped to structure my thinking to grow in my professional career I think without doing this investment in the PHD I think my career wouldn't be as it is now marus thanks so much for joining us today thanks for having me map podcast listeners will
post show notes to today's conversation at mefa.org podcast if you love the show if you hate it shoot us feedback at the mebf show.com we love to read the reviews please review us on iTunes and subscribe the show anywhere good podcasts are found thanks for listening friends and good investing [Applause]
