HomeVideos

What Can Quant Trading Strategies Teach Us About Markets?

Now Playing

What Can Quant Trading Strategies Teach Us About Markets?

Transcript

707 segments

0:07

Welcome to Merryn Talks Money, the podcast in which people who know the

0:10

markets explain the market. I am Merryn Somerset Webb and this week

0:14

I am speaking with Simon Judes, who's the CEO of Winton.

0:18

Now Winton is a quantitative investment management firm headquartered in London,

0:22

but with offices around the world, the firm manages around 6 billion worth of

0:26

cash, mainly for institutional clients. But I think increasingly with an idea of

0:31

having ordinary investors and wealth managers buy their products as well.

0:34

So we will talk about that along the way.

0:36

Simon, Welcome to Merryn Talks Money. Thank you very much for having me.

0:40

Right. Let's start by trying to explain what on

0:42

earth it is that you do. We hear a lot about quantitative

0:45

investing. We hear about CTA.

0:46

We get all this kind of, you know, model systems, etc., etc..

0:49

What is it that you actually do? Well, let me talk about quant investing

0:55

in a more general time, and then I'll come on to CTAs in particular, which is

0:58

one type of quant investing. So the way to think about it is imagine

1:01

you start off with a more traditional start of investment where perhaps you

1:04

have analysts thinking about a company and whether they should buy shares in

1:08

the company, whether they think the company is likely to do well.

1:11

That would be something that a lot of effort goes into, a lot of discussion.

1:15

And if you want to make that decision not about one company, but about ten

1:17

companies, you need to have ten times as much discussion.

1:20

And that leads you to scale those organisations quite heavily and focus on

1:25

having a small number of positions that you have a high conviction.

1:29

And that's how traditional investment works.

1:31

And quantum investment is based on a different idea where you say, well,

1:34

maybe we don't have to have that high conviction in every single investment if

1:39

we can have a large number of them. So it's a little bit more akin to how an

1:42

insurance company might might treat its products.

1:44

It doesn't have to make money on every single policy it sells as long as it

1:48

arranges things so that on average it does.

1:50

So the idea is, well, what if, instead of doing all this huge amount of

1:54

analysis on each company that you decide to buy, Merryn, you instead come up with

1:58

a rule that is perhaps much simpler. And it's something that's so simple that

2:03

you can automate, you can get a computer to do it.

2:05

So suppose that you look at some valuation metrics or something else

2:08

about how the company's performed recently and you decide based on that

2:12

you will have a position of a certain size in the company, and that's

2:15

something that you can get a computer to do for you, and you can do that with a

2:18

much, much larger portfolio. Now that might sound a little bit

2:23

speculative, like, how could you know that following a rule like that would be

2:26

a good decision for you in the long run? And the reason that you can get

2:29

confidence in that is because you can do something that you can't do in the more

2:32

traditional style of investment. What you can do is a back test where you

2:36

take that rule and you test how it would have performed over a much longer time

2:40

horizon. And that is what gives you confidence in

2:44

quantum testing that this is going to be a good rule for you to follow.

2:47

And if you think about what that then looks like, it means that when you are

2:50

allocating money as a quantum investor, you're not allocating or you're not

2:54

really thinking directly about each individual investment.

2:57

You're thinking about allocating to these rules these ideas or algorithms

3:00

that are themselves going to decide which companies they buy and which

3:03

companies they sell. So that's the general idea of quantum

3:06

investing. It's that the difference from more

3:08

traditional styles of investments, and that's how you can get confidence by by

3:12

doing these historical studies which show you how the rule would have

3:14

performed. How does that work with with long term

3:18

idea that every time an anomaly or a gap in the market like that is found, it's

3:23

almost immediately traded away? That's a great question.

3:27

You definitely see with many types of rules like that that they work for a

3:31

while and then you can even see in the spark test that they gradually start to

3:36

work less and less well. And that almost certainly is always

3:38

because many more people are doing it. So what can you do about that?

3:42

Well, the first thing you can do is that there are some ideas which don't have

3:47

thought that fall off in. And in a way, perhaps they're not based

3:51

on discovering a simple trick or arbitrage, which once somebody else has

3:54

discovered it, it will just stop working.

3:57

They might be based on more general behavioural biases that people have, and

4:00

that's really where CTAs and momentum in general can have set in this way of

4:05

thinking about it. So there's two, two types of rules that

4:08

we might follow here. One would be purely numerical

4:11

statistical and the other might be behavioural.

4:14

Well, even the behavioral ones, we would still measure through data.

4:17

They would still be measured through the way that either prices or other sorts of

4:21

data we might use are behaving. So there's something coded as systematic

4:25

algorithms that take in data transformer in some other way and then come up with

4:29

a rule for how you trade. So I think those two things might end up

4:33

being the same. But the second way I, I was going to say

4:36

that you can combat that people call it alpha decay, where the idea stops

4:40

working is by doing a lot of research and continuing to find new ideas.

4:44

And that's a big component of our activity as well.

4:47

Okay, so there are still new ideas to find.

4:50

Absolutely, because there's new data all the time as well.

4:53

All right. Can you give us an example of a of a new

4:55

rule? Well, that's why you might be

4:58

infringing. Okay.

4:59

You don't want to do that because it's proprietary stuff.

5:01

And this is one of the problems I think, that retail investors often have with

5:04

this type of fund is that they're told the theory, but it's not possible for

5:08

them to be told the practicalities because that's your model, your

5:11

proprietary data. So why don't we go back a bit?

5:14

Maybe you can give us an example of a rule that you used that once worked and

5:20

doesn't anymore. That way we're not infringing on your

5:22

corporate knowledge. Well, I think we can do that because in

5:25

the CTA space, actually, which is a particular type of quantum investment,

5:28

the basic idea of how we come up with the trades is quite well known and it's

5:35

something we can talk about very openly. So cities are a particular subsets of

5:40

quant hedge funds which apply that same philosophy that I was talking about

5:43

earlier to macro markets, primarily through trading futures, so that they're

5:48

trading equities, that they're trading bonds, they're trading currencies and

5:51

trading commodities. And predominantly the way they're doing

5:54

that is by looking at recent price momentum.

5:56

So the things that have gone up over some recent periods, you have a long

6:00

position and if they've gone down, then you have a short position and that that

6:04

is the algorithm. It really is looking mania, recent price

6:07

behaviour. Okay, So you've got a fund that trades

6:11

in futures in all the asset classes based on the direction prices are

6:16

already going with the idea that you can spot where the turn will come.

6:20

No, no, we can't really spot one tone from that.

6:23

That's very hard. It's based on the idea that if the price

6:25

has been going in that direction, that it's more likely than not to continue

6:29

going in that direction. Okay.

6:30

So you've got to have an awful lot of positions for that general rule to work.

6:34

Yes. You need, you know, more than 100 ready

6:38

to get the benefit of that diversification, because like I said, in

6:43

general with quantum dots and you don't have very much conviction in any one

6:46

position. So I'm slightly suspecting at some point

6:48

you'll ask me what gold is going to do next, for example.

6:51

Not yet. Not yet.

6:53

It's coming. I don't know when when the question

6:55

comes, I don't know. But we believe that there's a greater

6:59

than 50% chance that it will continue to go up, which is not saying very much.

7:03

But as far as the large portfolio, it's still a very valuable thing to do.

7:06

Hmm. Okay.

7:07

Before we move in to CTA, what does it stand for, by the way?

7:10

It stands for a commodity trading adviser, which is it's a regulatory term

7:14

from the the seventies in the US. So it doesn't have very much it doesn't

7:19

really the words don't really mean anything anymore, that the acronym means

7:22

something. But there was no longer mean anything.

7:24

Correct. Okay.

7:25

Let's let's go back to the to the the, the pre CTA conversation about the

7:31

different rules and valuations. I am usually really, really keen to nail

7:35

down a little bit on the kind of thing that you mean when we when we talk about

7:39

all these different worlds, all these different diversification, different

7:42

models except break up for that for the purpose of the retail investor who is

7:46

not, not used to this kind of fund and is used to more of traditional equity

7:52

funds where we're talking about holding growth and this is what growth means and

7:55

we're doing value and this is value means and within greater reasonable

7:59

price. And that's what this means, emerging

8:00

tech and those but this means we're mining.

8:02

That's but this means except for these this stuff is easy to get a handle on.

8:06

But what you do is not so a neat little example of one type of strategies less

8:13

models, less valuation metric, etc. that works or has worked for you.

8:17

It would be super handy. Absolutely.

8:19

I think value is a good one to focus on. So if you think about traditional value

8:23

investors, Warren Buffett, Benjamin Graham, this kind of thing, there's I

8:26

take issue with that while Warren Buffett and value investing come back to

8:29

that maybe well, think of somebody who's doing a lot of effort into assessing a

8:33

company and deciding if the current market price represents good value.

8:37

So maybe if that's the way that now describes every single investor there

8:41

is, because no one buys something which they don't think is good value.

8:43

Even a growth investor, if they're paying 70 times for something we'll

8:46

think is good value because they expect it to go up, right?

8:49

Absolutely. Same thing, same same.

8:51

So, yeah, okay. A lot of people would put themselves

8:53

into that that way of describing things. And now imagine instead of putting a

8:56

huge amount of effort into each company and looking at what the management of

8:59

said into looking all over the balance sheet and looking at the drivers of the

9:03

company's profits and so on and looking at what the price should be.

9:06

Imagine that all you do is you look at the price earnings ratio of every

9:10

company that you can find and you just rank them and you just look at the ones

9:14

that have lower than average and the ones that have higher than average price

9:18

earnings ratios and you just decide to go long.

9:20

The ones that have lower than average P ratios are short, the ones that have

9:24

higher than average. That is an example of an algorithmic

9:27

rule, which in and of itself is not very sophisticated or clever, obviously,

9:34

because it doesn't involve any interesting analysis of what those

9:37

companies are doing. It just takes one piece of data per

9:39

company and updates at roughly once a quarter, depending on how often these

9:42

numbers are updated. It would rebalance once a quarter

9:45

generally. Yes, exactly.

9:46

So this is an idea that many people have traded for decades.

9:51

It doesn't work out well anymore as well as it used to.

9:55

And many people have adopted more sophisticated tweaks on it.

9:59

You know, perhaps you don't just rank every single company.

10:01

Maybe you look at the ranking of a company within a sector.

10:05

Maybe instead of looking at the trailing earnings, you look at analyst forecasts

10:08

of earnings, maybe you try to forecast what the next earnings announcement will

10:12

be yourself. You know, there are lots of ways to make

10:14

the idea still conservative and systematic, but more sophisticated than

10:17

that original base idea. But ultimately, what they're all doing

10:20

is something that is much less in-depth than what a traditional investor would

10:24

do to assess whether a company is good value or not.

10:27

And on the flip side is that you can do it very easily for a very large number

10:31

of companies and you can back test it through history and you can see what

10:35

following this rule would have done. Okay, so you can do all that and you

10:39

have maybe a hundred different strategies running in a fund.

10:43

You said earlier it varies. We have different products.

10:47

So some of the CTA products, for example, are really based around the

10:50

properties of of momentum, of trampoline on macro markets.

10:53

And there are good reasons for that because that has a specific function for

10:56

investor portfolios, which we can talk about from other products.

11:00

We have combined a much larger number of different strategies where we're really

11:03

trying to benefit from the diversification you get from a large

11:05

number of different ideas. This sounds to me like the

11:08

kind of thing that I would be remarkably helpful for.

11:12

It certainly is helpful for us. It's it's not something that's extremely

11:17

new in a sense. It's the the latest example of a fairly

11:21

transformative technology or set of techniques.

11:25

Probably wouldn't surprise you given that we have a huge technology component

11:28

to our business and the people we employ are people with backgrounds in software,

11:31

engineering and sciences and so on, that this has always been something very

11:35

important for us. And before it was machine learning and

11:38

it was all kinds of data and it was big data and it was cloud technology.

11:42

And it's very important for us to keep up with the way that all of these

11:46

ideas develop. And certainly the recent developments in

11:49

large language models have been very helpful for us.

11:51

Does it make everything move faster, faster, faster?

11:53

So the edge that you might get from each new idea is removed much, much more

11:58

quickly? I, I think well, to some extent.

12:02

I mean, there's a general truth in investing that nothing, you know, ever

12:06

totally transforms the business because at the end of the day, if everyone

12:10

discovers something, then its value reduces because ultimately there's a

12:16

certain number of dollars to be made out of, out of a given idea.

12:20

And the more people that do it, the more they have to share it.

12:23

But that doesn't mean that you can just not, you know, follow it.

12:25

You still have to do these things. All right.

12:27

Let's move on to what I know you really want to talk about, which is the CTS

12:30

stuff and the trend following following. Explain how that works.

12:33

Yeah, absolutely. So this is a, quote, strategy that

12:35

trades the macro markets primarily through futures.

12:38

And it has a particular role in investment portfolios, which has always

12:45

been there. But it's become increasingly important,

12:47

really since 2022. There have been other moments as well

12:51

where investors have become particular interested in cities.

12:53

The financial crisis was was one particularly where equities obviously

12:57

fund lots and lots of other asset classes that did poorly and people start

13:01

to look around for what things have done well.

13:02

And at that time the two things that had done well were Bonds and CTAs.

13:08

And so there was a big surge of investment into both of them after the

13:11

financial crisis. And then in 2022, the remarkable thing

13:14

that that happened, which is a headache not just for the retail investors that

13:19

you mentioned, but actually for all investors, is that bonds and equities

13:23

fell together for four year.

13:26

And the reason this is a problem is because a lot of allocation philosophies

13:31

have relied on the idea that bonds could provide diversification to equities and

13:35

that that wasn't a crazy idea. You'd had a 20 year period leading up to

13:39

that where bonds and equities were negatively correlated and not every

13:42

time, but pretty much every time stocks had gone down.

13:45

Bonds had helped, as they did in the financial crisis.

13:47

The traditional 60/40 portfolio worked for everybody for a very long time and

13:51

then suddenly kind of stopped working. Exactly.

13:54

And in fact, it didn't have to be 60/40. You know, any reasonable combination of

13:56

bonds and equities would have performed extremely well over that time period.

14:00

And then that diversification disappeared.

14:03

Really starting 2020, the correlation flipped to being positive, which meant

14:08

that the potential for the both to go down at the same time was there, even if

14:11

it hadn't actually happened yet. But then in 2022 that potential was

14:13

realized and they both did go down together and that really drove home that

14:18

you can't just rely on bonds to be your diversified equities.

14:21

And so people started again looking at asset strategies, which had done very

14:24

well that year. So why is it that CTA strategies provide

14:27

this diversification? There are basically two reasons.

14:30

The first reason is that among the futures markets that they trade, there

14:35

are a variety of things which are very uncorrelated exposures and things which

14:38

are quite difficult for people to to benefit from otherwise.

14:41

So an example of that would be, for example, the big rise in the price of

14:44

cocoa, which took place in late 2023, in early 2024, and then subsequently the

14:49

big drop in the price of cocoa. This happened in the last few months.

14:53

That's a trend that you can benefit from by trading the futures markets, but it's

14:56

very difficult to get that exposure through stocks, for example.

15:00

So that's one reason. The other reason that you can benefit a

15:03

lot from sea change in the portfolio is that because they trade futures markets

15:06

and futures markets have this amazing property that it's just as easy to be

15:10

short as it is to be long. You can benefit from short positions in

15:14

the major asset classes. It's just as easy to maintain a position

15:17

that short equities or short bonds that is to be likes is long bonds.

15:20

And so trivially in a year like 2022. The strategy is short in those asset

15:25

classes and it does well. So that was a good year for CTA.

15:29

That was a period when you're the assets that you hold kind of peaked, right?

15:33

So that there's Winton as a whole manages less money now than it did at

15:38

its peak. What's driven that is that because

15:40

people looked at the successes in times like that and then go, well, since then

15:43

long equity has been mobile. It's very easy to make money and why do

15:46

you need to mess around the edges? Oh, Peak was at a different time from

15:50

that era. It wasn't.

15:51

Why was it that early on? Yeah, why in 2016.

15:54

But I think there's there's a peak of a single manager which has has more to do

15:58

with what's going on in that organisation.

16:01

But the peak of the industry is a different question.

16:04

So in terms of how much people are allocated to CTAs, it's a little more

16:07

difficult to tell. Because some cities are public like we

16:12

are. Others are offered through through

16:14

products that perhaps banks offer. And it is not publicly known how much is

16:18

being traded there. But what we're certainly seeing is that

16:22

the interest relates strongly to the performance of the strategy, as you'd

16:25

expect. Okay.

16:27

And the performance of the strategy long term.

16:30

Well, look, long term, it's been a great addition and diversify so you can take

16:34

any combination you like of bonds and equities.

16:36

And it's been beneficial to add an allocation to the cards as well over the

16:40

long term. Beneficial like how beneficial in the

16:43

sense that performance has improved and particularly the drawdowns that you get

16:49

when you have these big moments like 2008 or 2022 are mitigated to some

16:53

extent. But let's say, for example, you're a

16:55

retail investor and you make 5% of your portfolio CTA.

17:00

How would that, how would that have helped your performance over, say, a

17:03

decade long period? Well, you know, the exact numbers vary

17:07

depending on I know I know the exact numbers vary.

17:08

But but, you know, we need to we need to know that it would make a positive

17:12

difference of more than a couple of basis points here, that it does it makes

17:16

a positive difference. And you can measure in a couple of ways.

17:17

So you can say, well, there's an annualized number which might go up by

17:20

1% or something depending on exactly the weighting you give to it.

17:24

But then what is also important is it's not just that sort of average, but the

17:29

fact that historically, at least the seats have performed well in the curves,

17:32

which haven't been good for for equities.

17:34

And so it's added the profits at particular moments when all the things

17:39

were doing badly, which isn't as easily quantified by just adjusting the annual

17:44

return number. But it's a very important component of

17:46

why people are allocating to it. Yeah.

17:48

Can you tell from. Of course you can.

17:50

But looking back, how much of your performance comes from being long asset

17:55

classes and how much comes from being short an asset class?

17:58

Is there any any differential there? Are you equally good at both or do they

18:03

both equally fit into performance? You get good performance from both.

18:07

So I'll give you some examples. So

18:11

if we look at 2008 at the financial crisis, that that's a really good

18:15

example to play through. So that year a good portion of profits

18:18

did come from long positions in bonds. Bonds did very well, but about a third

18:22

of profits came from being short in equities.

18:25

So you saw profits on both sides. Now you can pick other examples again on

18:30

both sides. In 2014, that was a very good year for

18:33

for CTAs. Why was it a good year?

18:35

Well, there were two enormous trends. One was the downward trend in oil prices

18:39

that was driven by the shale boom. And that is a great example of a trend

18:44

that is very, very easy for us to latch on to.

18:47

Being short oil is as easy. Like I said, it's been long oil, but

18:51

it's not so easy to do in other contexts, so situated very well being

18:55

short oil. And that year, the other thing that was

18:57

going on was there was this tremendous upward trend in fixed income,

19:01

particularly in bonds. And it was

19:06

quite surprising actually at the time because yields on bonds were getting

19:10

towards zero and in fact they ended up going below zero, which is something

19:15

that people generally didn't expect could happen.

19:17

And these two examples actually reveal something quite unintuitive about, about

19:22

momentum, about the idea of following trends in these markets, because often

19:25

people think, well, if you're following a trend, you must be getting into some

19:28

very crowded trades and you must be sort of following a herd.

19:31

And just the general received wisdom. And in fact these examples show you that

19:35

very often you can be doing precisely the opposite.

19:37

And the way we know we're doing the opposite is that we can look at what

19:40

fundamental analysts are saying about what the market is going to do at each

19:44

moment. And we find there are these moments

19:47

which are very profitable where the trend disagrees with what the

19:50

fundamental analysts are saying. So if you look at what people were

19:52

saying about where oil would be through 2014, people were not focused on on the

19:56

shale, but they were focused on the apparently unquenchable demand from

20:00

emerging markets, the policies which might lead to the reduction in supply.

20:04

You had people talk about peak oil supply, for example, at that time, and

20:07

every analyst was predicting that oil would increase in value.

20:10

And similarly with that trend in bonds, everyone was saying, well, bonds cannot

20:15

possibly go up from here because why on earth would anyone give their money to

20:18

the German government to get less of it back in ten years time?

20:22

Yeah, we might still ask that question actually why they did that, but here we

20:26

are. Yes, but you might.

20:28

You might. But the great thing about about trend

20:31

following and about these algorithms in general is that they don't have to know.

20:33

They don't have to have a theory about why the markets do what they do.

20:37

They have a rule that that, you know, this market's been moving in this

20:40

direction. Maybe it will continue.

20:42

And they have evidence that following that in the past has been a good thing

20:45

to do. Okay.

20:46

So let's move on to talking about what trends are interesting at the moment,

20:51

because, of course, that's what everyone really wants to know, is if, as you say

20:55

there, the market is moving in one direction, there is momentum that before

20:58

it has become a consensus view. What should we be looking at at the

21:01

moment? Well, you know, the the things are not

21:04

necessarily things that are unknown.

21:07

So the big trends that we've seen recently have been in precious metals to

21:12

some extent in base metals, in equities, obviously equities have

21:17

been going up in general. There's been good short trends in some

21:22

places in the US natural gas, for example, and particularly in cocoa, the

21:27

example I mentioned earlier with cocoa. Well, it's interesting you mention that

21:32

because it's obviously a fairly niche market and particularly when we trade a

21:38

few hundred markets. Why am I mentioning this one?

21:41

The interesting thing is that the strategy itself is very dynamic.

21:45

So when there's a big trend, it will take a big position and when there's not

21:48

much going on, they won't take a huge position.

21:50

So what that means is that even though you're trading perhaps 200 different

21:54

things, it's not like building a lonely portfolio of stocks.

21:57

So you have to just always maintain this collection of the.

21:59

positions. The strategy is behaving very

22:02

dynamically. So at any given moment there'll be a

22:04

much smaller number of things which are really contributing and it will be based

22:08

on what those markets have done recently.

22:10

So I'm talking about cocoa because it's had this astonishing rise and then

22:14

equally astonishing fall recently. And that's why it's making a big impact

22:18

on the portfolio and not it didn't do for perhaps 20 years prior to that.

22:22

So if you give us any points in the last 20 years, I wouldn't have been talking

22:24

about you mentioned it. This is quite interesting.

22:27

So when you when you talk about one of your funds, for example, having

22:31

having all these different positions and things being traded, etc., the returns

22:36

over, say, an annual period will probably only come from a couple of

22:39

those positions. They'll only be a few that are of

22:43

reasonable size. Well, it's only a smaller number than

22:45

the 200. You're right.

22:47

It's typically a few big things that are that are happening every year.

22:51

And that's up to you, right? So cocoa.

22:54

Yes. Gold.

22:55

Cocoa, gold. Yeah.

22:57

The sun, the base metals as well. Copper.

22:59

Yeah. Aluminium, uranium,

23:03

less less uranium because it is a less liquid market for us to trade And that,

23:08

that the again as well there's been a big market for us over

23:11

the last few years. Uh huh.

23:13

And would you expect the yen to turn of conversation about the yen.

23:18

Absolutely. And again, something because again, you

23:20

don't know, right, that that has to be the answer.

23:22

I don't know. Now, although it is another great

23:24

example of the fact that the algorithm, you know, doesn't have a theory.

23:28

It's trying to follow as a massive strength, because obviously what we saw

23:32

the last couple of years was everyone predicting that the yen is going to

23:35

rally and they're not doing it or doing it for a bit and then resuming its

23:39

slump. And you can tell so many good stories

23:41

about wood, right. How it should rally will rally interest

23:45

rates going up, inflation, all the political change in Japan, etc., etc..

23:48

And it never happened. Exactly.

23:51

And in fact, you can see these kinds of episodes.

23:53

This is an interesting point. I think that often these kinds of

23:57

episodes are almost necessary for a trend to appear, because if you think

24:00

about what a trend is, it's a situation where a market moves a long way, but

24:05

that slowly, right? If a market moves along way very, very

24:08

quickly, then, you know, it's kind of 50/50 if you're on the right side of that

24:12

or not, for that to be a meaningful trend that we can benefit from, it has

24:16

to do it slowly. And that's a slightly surprising thing

24:19

that it could have happened, because what you hear about the way markets have

24:22

developed is the information is incorporated into them and have a faster

24:25

rate. And so you might think that if something

24:27

is going to happen to a market, well, it is going to be very, very quick.

24:32

But there are ways nevertheless, that things can happen slowly.

24:35

And one of the ways is if the market kind of has to fight against the

24:39

narrative that is saying that the thing that is going to happen actually can't

24:43

happen. And that's exactly what was going on

24:46

with the yen, right? Because all of the narrative was saying

24:49

the yen has to rally. And the fact that for whatever reason,

24:56

which I'm not going to be able to tell you for whatever reason, the market, in

25:00

fact, was really pushing in the other direction the fact that that had to

25:04

fight against what everyone thought was supposed to happen is part of the reason

25:07

that it got drawn out into into a long term trend and was therefore something

25:11

that we were able to benefit from. Interesting.

25:14

Are there any other things that you're looking at the moment that are moving

25:16

remarkably slowly? Well, look, gold is is a is a good

25:22

example that slow well but it's been over a couple of years but it's slow and

25:27

then very fast. That's right.

25:29

And the dollar obviously the dollar has started to weaken over a relatively

25:34

drawn out time period. Silver, again, very slow and then

25:38

very, very fast, which is in a way that is ideal behavior, actually.

25:42

There's another element of the algorithm which I haven't talked about before, but

25:46

what we do is when we have a strong signal that determines that we want a

25:50

certain amount of volatility from from that position.

25:53

So when you got the situation like with silver, I think it is silver, but gold

25:57

to where it starts, small builds and then accelerates.

26:03

That means that we start with building in a long position and then as it

26:08

accelerates, while the volatility of the market goes up and actually we don't

26:13

need to own so much silver anymore in order to get the amount of volatility we

26:17

want. So actually, as that market is

26:19

accelerating upwards, we are selling again.

26:22

Another slightly surprising thing that you wouldn't necessarily expect from

26:25

somebody who's following a trend, but that's typically what's happening as

26:28

others markets accelerate and in effect, you are locking in the profits that you

26:33

make. As you go.

26:36

As you go. Yeah.

26:37

Okay. There's a sort of a view that pretty

26:40

much all investing these days is momentum investing or trend following

26:43

investing, because everyone, everyone in a passive investment is effectively a

26:47

momentum investor. Just on the long side.

26:54

In the sense of passive learning to investment.

26:58

You know, if you've been invested, for example, until relatively recently and a

27:02

global equity ETF of any kind, you've effectively simply been following

27:08

American and tech momentum for years. That's it in our value investing in

27:12

growth, investing, but anything in just a momentum investor.

27:16

Yeah, there's an element of truth to that.

27:18

I think it's slightly difficult to compare not only investments with with

27:24

long short investments because even though some of the ideas might be, might

27:28

be similar, the actual trading that you end up doing and the nature of the

27:30

portfolio you end up holding is very different.

27:33

Now, obviously it's just just this idea that most people have one side of your

27:37

portfolio already. To a degree, the equity part.

27:42

Well, yeah, they have. Yeah.

27:44

One half perhaps of the of the equity portfolio

27:47

in some sense. That's right.

27:49

And the point being that it's kind of dangerous in that most, most people are

27:52

passively investing in a global ETF, are not aware that they have a one sided

27:59

momentum strategy. Yeah.

28:02

Look, it's an interesting point of view. You can think of any index, if you like,

28:08

as a type of of trading rule, as a type of algorithm, because it's it's

28:13

effectively saying if it goes in the index, then you buy it.

28:16

If it exits the index, then you sell it and other times you maintain it in these

28:20

proportions. And it's interesting because obviously

28:23

that is another rule that you can you can access over time.

28:25

So you can't, if you like, think of it as a as a type of quant strategy.

28:30

And we often do try to think of it that way.

28:34

I don't think I have anything useful to tell you really about whether that's

28:37

dangerous or not, but that perhaps is a deeper question and I'm prepared to

28:41

address it.

28:44

Well, we've been talking remember I asked you earlier about about the extent

28:48

to which it would have enhanced the performance of a portfolio.

28:50

Luckily, you have PR people, Simon, and they've sent me something saying

28:55

went to research, found that a 10% allocation to trend following would have

28:58

improved the returns of an equity bond portfolio and 87% of ten year period

29:03

since 1972, increasing the portfolio's outperformance over cash to 4.8% from

29:09

4.1% on average. And this return improvement nearly

29:13

doubled and the bottom decile of ten year periods for the equity bond

29:17

portfolio, highlighting diversifying properties and portfolio resilience.

29:23

Do they get fantastic? I congratulate them.

29:26

Well, I think you did the work. They just wrote it down.

29:31

So in that sense, would you say that a 10% allocation would be a reasonable

29:36

amount for a retail investor? So let's say we've still got a long

29:39

equity exposure, we've got some bond exposure, maybe we've finally, finally

29:43

got the message and we've got somewhere between 5 to 10% in gold.

29:47

Maybe, maybe we've got 3% in Bitcoin. Not recommending that, by the way, but

29:51

maybe we do 1%. Should we also have 10% in in a trend

29:56

following in a CTI strategy? Does that make sense to you?

29:59

Because certainly makes sense that the right level of allocation depends on,

30:04

you know, what your preferences are as an investor and you if you allocate to

30:08

things other than equities and bonds, then obviously you're departing from

30:13

that benchmark and you introduce some risk of of outperforming or

30:17

underperforming that benchmark. And different people have different

30:19

tolerances for that. There's another way of doing it as well

30:22

where you get to maintain the exposure to equities, which we offer through a

30:29

sort of an alpha type structure. This is something that has been really

30:33

always on the radar of institutional investors, and we offer it now for for

30:38

retail investors as well. And you use this product.

30:41

Okay. And I know I said you might have a

30:45

couple of percentage points of your portfolio in Bitcoin, but is crypto one

30:50

of the asset classes that comes in to your strategy?

30:53

We do trade crypto in that way. If you think about what we're looking

30:56

for, particularly with that kind of momentum strategy, we're looking for

31:00

markets which might have trends and which are liquid enough that we can

31:04

trade them. Obviously, crypto satisfies both of

31:06

those things. The only question would be is the is

31:10

there a way to trade them which is operationally safe?

31:14

And there is actually because there are futures on some of the crypto assets,

31:19

there are futures trading on CME or on Bitcoin and Ether and a couple of others

31:22

as well. And the structure of those is basically

31:26

the same as most of the other futures that we trade.

31:29

So it's it's fairly straightforward to add them in, to maintain system and to

31:34

benefit from them, which we have done this year.

31:36

Okay. And which direction.

31:38

Well this, this year is sure to be benefited but at other times it's been

31:42

more. Yeah.

31:42

And would you be short now. Well look again you shouldn't take any

31:47

of this as insightful advice about was absolutely not we know that this is

31:52

purely about momentum and trends, but we still want to know if that's right.

31:56

Well, really, you're just asking me, have they gone up or down recently?

31:59

No, I'm asking you, where are you expecting them to go now?

32:01

How does that trend look? Well down.

32:04

Yeah. Like I said, the trend is just what has

32:06

happened recently, which is which is down.

32:08

But the chance that that means it's going to go further down is is only

32:12

marginally above 50%. Okay.

32:15

And what makes it difficult for you? What are the risks in this strategy?

32:19

What? What?

32:20

What makes it go wrong? Look, there are two kinds of situations

32:23

where momentum investing is not going to do well.

32:25

So one is where there aren't any trends. Markets just move sideways.

32:30

The other is where there might be a trend, but then there's a sudden

32:33

reversal. So we've had good examples of that

32:36

recently. So last year, for example, there were

32:38

big reversals in March and April. The tariff announcements in particular

32:42

caused equities that have been going up.

32:45

They then started going down. The dollar had been strengthening, then

32:48

it started weakening. You know, a lot of commodities have been

32:50

going up. Then they went down.

32:51

So those kinds of situations are the situations where the strategy won't do

32:56

well. In the case where it does well is where

32:58

markets move steadily and consistently in a particular direction, or at least

33:02

enough markets do that to outweigh the ones that are behaving in a more

33:06

negative way. Yeah.

33:07

So what are your expectations for the rest of this year?

33:11

Well, I know you can tell me some things will go up, some things will go down,

33:14

but do you expect that type of volatility that will make it a very

33:16

heavy. Look, we have evidence.

33:22

You know, when we look historically, we see that there are there are some years

33:25

which are good, some years which are less good.

33:28

And we can't predict if we could time when Tremfya is going to work.

33:32

We would just build it into our strategy that it would take more risk in those

33:35

periods and less risk at other times. What we found in terms of allocating to

33:41

trend following is that it works best if you don't try and time it, if you just

33:45

regard it as a as a permanent allocation and live with it.

33:49

For example, after that period in April, April, May, June were difficult months

33:56

for the strategy for the reasons that I said.

33:58

And if you had been thinking about at that point, you get pretty depressed and

34:01

you think, Well, that's just hasn't worked.

34:04

When is it when is it going to start to work?

34:06

And then what happened was, over the course of the rest of the year, we saw

34:10

several big trends emerge. So obviously, the precious metals rally

34:13

continued. Equities went up relatively steadily.

34:17

We saw some of the base metals do well as well, and there wasn't a continuation

34:23

of that negative whipsawing behaviour. That was really a prominent feature

34:26

earlier on in the year. So that meant that overall the strategy

34:30

did very well, but there was nothing that you could have pointed to in July

34:34

that said, Oh, now is the moment, now is the moment.

34:36

It's definitely going to work. Just as there was nothing you could have

34:38

really pointed to in January, which really said, Now that is definitely not

34:41

going to work. So unfortunately, that that's the way

34:45

things are and I don't have any better answer to that to be able to predict

34:49

that this is a kind of trust strategy, isn't it?

34:52

Well, not really, because this is why you can look back at the track records

34:56

and at the end we can look back at the back test.

35:00

And it's those things that we are we are looking at.

35:02

It's that the more it's a more discretionary type of investor who's

35:06

making one decision, you know, today. And it's totally different from the

35:10

things they looked at last year and the year before.

35:12

They're the people who are saying, trust us because they don't have the system

35:16

back to us to look at. Yeah.

35:18

So let me ask you something. One thing that comes up a lot on this

35:20

podcast is ESG strategies and stewardship around investing.

35:24

And if you're investing like this in a way that is entirely non company

35:28

specific, you can't really have an ESG overlay of any kind, can you?

35:34

Not in the traditional sense, though. There are various things that you can

35:37

you can do. There are, for example, ESG versions of

35:40

the traditional indices and their futures on on those indices.

35:44

So we trade those where there's sufficient liquidity to to enable it.

35:49

We obviously can't create a liquidity where it doesn't exist.

35:53

So to some extent reliant on the market makers and other people to take an

35:56

interest in those things. But we are we're well positioned to do

35:59

it if if they become popular. Okay.

36:02

And tell us briefly about the the appropriate vehicle that you have for

36:06

retail investors. That's the Winton Trend Fund, right?

36:09

The UCITS? Yeah, Well, we have we have UCITS

36:13

vehicles for all of our major strategies.

36:15

So we do have the pension trust fund. In addition, we have another Winton CTA

36:22

fund which includes non-trade elements as well.

36:25

So that's the enhanced fund that I hold and I'll come back.

36:32

There's the first one that I mentioned was the Winton Armor Diversified Fund.

36:36

The Winton Trained Enhanced Fund is the fund I mentioned that has strong

36:40

following, but then also maintains the exposure to global equities.

36:43

So for example, if you wanted to add this into your portfolio that you didn't

36:47

want to disinvest from from your equity

36:51

portfolio, this is a way to do it because it maintains 100% exposure to

36:54

equities. And then the last UCITS fund, I should

36:59

mention, is the is our multi street hedge fund.

37:04

Which is also that these are all global funds, right?

37:07

Yes. They were looking at everything

37:09

everywhere all the time. That's correct.

37:12

It's a very busy 200 employees. You've got the.

37:15

They are very busy. But remember that they're not

37:18

individually they're not doing any individual analysis analysis.

37:21

They they're looking at the algorithms. And the algorithms are mostly things

37:25

that can be applied relatively broadly. Mm hmm.

37:27

Mm hmm. And do you think that over the next

37:29

decade, these jobs will still exist? Because an awful lot of it does sound as

37:34

though it's hugely automated already. Will it get more so will there be 200

37:39

people required to run 6 billion in a decade?

37:44

Using this kind of strategy? It's a good question.

37:47

Can we just put everything? I don't think you'll be able to find

37:50

quite everything. It's.

37:53

Yeah, I struggle to make predictions about this.

37:56

You know, it's interesting to look at. I mean, I'm going to get a prediction

37:59

out of you before we get to the end of this podcast, if it kills me.

38:01

So make one on this. Oh I'm sorry, Merryn

38:06

Oh, look, when you look at how particularly technology jobs have

38:09

changed, what you find, it's not necessarily that you need more.

38:13

A few people you find great variation in, in the skill sets that are required.

38:19

And that's that really is driven by technological change.

38:22

Often that's change in the type of software that we use that, you know, the

38:27

development of Python as a programming language that's much more broad

38:30

accessible than than previous programming languages were.

38:33

The changes in the type of databases that we use, the, you know, going back a

38:38

decade or two now, the introduction of virtualisation of PCs, which, you know,

38:45

drove a lot of the reduction in PC sales and and the

38:50

types of expertise that were then required to maintain

38:55

the computing resources also changed dramatically because you suddenly are

38:58

not dealing with a computer that's sitting on your desktop anymore, but

39:02

it's sitting somewhere in a remote warehouse.

39:04

And then the development from that to using cloud resources,

39:08

all of these drive changes in the type of expertise that are required.

39:13

And I'm sure that that will be true with with air as well.

39:16

But we are not. There are some types of business where

39:20

there's just a fixed thing that you're trying to do, and if you can do it more

39:23

efficiently, then you just have fewer people do that.

39:26

And that is not really what our business is because our goal is a bit more open

39:31

ended. We want to drive the best investment

39:33

returns that we can and if we can do the tasks that we're currently doing more

39:38

efficiently with fewer people, then what we'll do is we will keep the same number

39:42

of people and we'll do more. Okay, fair enough.

39:46

What do you think you'll get your kids to study at university?

39:49

When I ask this question now, there's a kind of clear division between between

39:53

people who still think it should be stem all the way and those who think, well,

39:58

we're moving into an age when empathy is more important than anything else, and

40:02

therefore, you know, sod chemistry do philosophy.

40:05

Where do you stand on that? Not sure.

40:09

It's really up to me. I my, my, my children have have their

40:12

own nothing. Nothing will be up to up to you.

40:14

But the way you try and feed it to them and make them think is their own idea.

40:18

What do you think that'll do? Well, my oldest son's interests at the

40:21

moment are split between maths on the one hand and classics and history on the

40:26

other. I think combo isn't that they are and I

40:31

actually quite like that, that combination.

40:33

I think there's, there's great value in both.

40:34

Okay. He was desperately trying to get

40:36

something definitive out for you about the future and we're coming up with do

40:39

both. That was the path I took.

40:41

So I studied physics and philosophy, and I think I've I've definitely benefited

40:45

from having both. Yeah.

40:47

You were on holiday last week. What were you reading?

40:51

I was reading a couple of the detective fiction novel by Louise Penny.

40:58

I think if you've come across those those I've read them all.

41:01

I reckon I've just discovered them. So this is I think I was on my fourth

41:05

one and I'm really enjoying it now. They get a little doubtful to the end.

41:10

There's an awful lot of them. I've now shifted towards a series of

41:13

murder mysteries, suspenses in Dublin by Tana French, which I recommend very

41:18

highly. Oh, thank you.

41:19

I'll give that a try. Thank you very much.

41:22

Simon, thank you so much.

Interactive Summary

Merryn Somerset Webb interviews Simon Judes, CEO of Winton, a quantitative investment management firm. Simon explains quantitative investing as a method using automated, data-driven rules and algorithms, often back-tested, to manage a large, diversified portfolio, in contrast to traditional high-conviction investing. He details CTA (Commodity Trading Advisor) strategies, a type of quant investing that trades macro markets via futures based primarily on price momentum. Judes highlights CTA's crucial role in portfolio diversification, especially after 2022 when bonds and equities showed positive correlation, by providing exposure to uncorrelated assets and the ability to profit from short positions. He discusses how these algorithms identify and follow trends (e.g., cocoa, gold, base metals) without relying on market narratives or fundamental theories. Simon also touches upon Winton's use of AI and large language models, their retail investor products, and their engagement with crypto futures trading, while acknowledging risks such as sideways markets or sudden trend reversals.

Suggested questions

7 ready-made prompts