Greg Maddox loves detail.
I’m calling him for this interview as I’m in New York, he’s in San Francisco and Citywire wouldn’t cough up for a Learjet on this occasion.
His first question is ‘What are you using to record this?’ He wants to know everything about my device. The brand, the model, how it works with the phone and what type of sound file it creates. The minutiae fascinate him.
Why is he so interested? Well, it turns out he has plans to record his own manager interviews and send clips out to advisors.
While he concedes his idea faces compliance issues and no small amount of manager resistance, it is one of the many ways he would like to use technology to enhance his team’s due diligence process and the speed at which it communicates its work to advisors and investors. He has already introduced machine learning to help monitor manager performance and has developed a proprietary risk and performance analysis tool.
Given these developments and his proximity to Silicon Valley, you would be forgiven for thinking that Maddox is obsessed with tech, but he insists his interest is purely professional.
‘I didn’t have a bunch of computers torn apart as a kid. I’m a late adopter,’ he says. ‘I still carry a Blackberry.
‘I think about [tech] a lot for our business but technology is more just a means to an end for our business to improve the probability of a better investment outcome for our clients.’
Maddox is head of global manager research at the Wells Fargo Investment Institute, which also houses a global investment strategy unit, a global alternative investments division and a global portfolio management team.
Maddox leads a team of 45, which is responsible for putting together a list of around 1,000 strategies to be used by various business lines within Wells Fargo. These include the broker-dealer Wells Fargo Advisors, Wells Fargo Private Bank, the family office Abbott Downing and the pension service Institutional Retirement Trust.
The manager research team is structured by asset class rather than by investment vehicle, so analysts evaluate strategies in a variety of formats, including mutual funds, exchange-traded funds, separately managed accounts, unified managed accounts, closed-end funds and hedge funds.
The size and structure of the team enables it to be based across the US and beyond. As well as San Francisco, the team has offices in New York, Charlotte, St Louis, Denver, Raleigh, London and Hong Kong.
Any of those would be a decent place to work, but Maddox acknowledges he has been dealt a good hand with his west coast location.
‘I’m not sure how I lucked into this,’ he says. ‘I grew up in Ohio… but I count my blessings every day coming into work. My commute is on a ferry boat. I get on every day with something to read and a cup of coffee and have a 10-minute walk on the other side. It’s not a bad way to go to work.’
Maddox’s attention to detail is evident in the team’s two research frameworks, which it uses to assess strategies before adding them to its influential list.
The first is called M-REF, which stands for manager research evolution framework. This is a quantitative and qualitative process that evaluates a strategy across 13 categories using a 95 sub-factor model.
The second is called Risk Optics. This evaluates managers across 12 categories, using 50 sub-factors. As the name suggests, it is focused on the risk that a fund presents to investors, specifically the risk of permanently impaired capital. It looks at liquidity, stress testing and lock-ups, among other criteria.
Once again, technology is central to the process. Risk Optics includes a proprietary tool that scores a strategy based on a series of questions that analysts must answer. The analyst is not told the score he or she has given, in an effort to ensure they cannot steer the overall ranking of a manager based on personal bias.
Maddox says the tool aims to find problems before clients invest. ‘By the time an event shows up in the standard deviation for the manager, it’s already too late,’ he says. ‘So it’s about trying to rewind the tape, if you will, to before an event occurs and ask what’s the likelihood of certain risks occurring in this strategy, with this team, with this firm, which could lead to a permanent impairment of capital.’
He says the focus on risk has increased since the financial crisis, along with an improvement in the tools available to measure and understand it.
‘Before the financial crisis the price of admission used to be “We know our positions better than anybody else.” Well, that’s a good starting point, but today with all the factor attribution and risk exposure tools that we have, that's not best-in-class anymore,’ he says.
‘Knowing their interactions, their correlations, the currency effects and exposures, the factor overlaps – there are a lot more insights for portfolio managers [today], we find, that improve their odds of beating their benchmark and their peers net of fees.’
While the frameworks can sound complex, Maddox says their aim is to answer relatively simple questions:
1) What market inefficiencies is the manager trying to exploit?
2) What is unique about the manager’s process that lets them exploit the inefficiency?
3) Is the inefficiency a temporary or permanent condition?
4) If this inefficiency moves around or changes in different regimes, what does the manager try to do about it?
5) How does the manager think about, measure, monitor and account for risk?
The human touch
Managers hoping to make it onto Maddox’s influential list do not get there by simply passing through the algorithms. Both frameworks are qualitative too and the team visits all managers at some point during the due diligence process and then at least once a year while they are on the platform.
Wells is a long-term investor, but a strategy can be dropped from the list for a number of reasons, including personnel change, underperformance, or a change in business control or prospectus. The team can recommend restricting flows to funds that it is concerned about. It can also recommend a sell or a sunset rating, with the latter giving clients two tax years over which to straddle a sale.
The team has not been afraid to issue sell ratings to seemingly successful funds, even if it means upsetting advisors.
‘We have a number of examples of very popular products reaping huge inflows that we have stepped away from,’ Maddox says. ‘In one case the manager touted his three-step process, which included a quant screen, fundamental analysis and a momentum component. This was a very popular brokerage product and we took a lot of heat for selling it.
‘But when we went in and looked at their process, it really was essentially only momentum that was sustaining them. The quant process was naive, there was literally no fundamental process in the strategy and I think folks chased a lot of performance.
‘We measured the two years before and after [we sold]. In the two years after we stepped away and recommended another strategy, that [new] strategy outperformed it by almost 700 bps a year.’
In another sense though, Maddox looks to go out of his way to please the brokerage, or any other of his lines of business. This is why he wants to record manager conversations and it's the reason behind his team’s foray into machine learning.
‘[The aim is] to help our analysts handle some more of the communication, commentary and data flow that comes in. That process has been underway for the better part of this year,’ he says.
‘We are feeding portfolio attribution data into the machine now and [have] trained it to interpret that and it is writing commentary. [It is] giving speaking points to the field on how a manager performed, what the key drivers were, the detractors and contributors.’
He says the speed at which such information is disseminated determines how useful it can be to advisors and their clients.
‘There is a decaying shelf-life for that information,’ he says. ‘The sooner we can get that out to our folks and arm them with that, the better off we are. Nine times out of 10 we want to let our clients know we are on top of this for them and we have systems for watching this stuff for them.’
Maddox says the biggest danger for clients who do not have timely information is that they panic and sell when the best bet is to hold firm.
‘They should stick to their plans and do nothing. When firms fumble getting information to their front-line and can’t persuasively reassure their clients that there is a system in place watching over their assets, then clients get nervous. Nervous clients end up making irrational choices that sometimes lead to them abandoning their plan at just the wrong time. Machine learning is helping us make more clients able to sleep at night.’
And this from a man who still uses a Blackberry. But then again, as he says, ‘it is the best for email.’