“An invasion of armies can be resisted, but you cannot resist an idea whose time has come.” Victor Hugo
This article was prompted by watching a video of wealth managers discussing artificial intelligence (A.I.) finding its way into asset management. The lines of resistance were clear, weak and unfortunate. There was no one there to balance the discussion, but for the record, the horse has already bolted, and it is only a matter of the speed of take-up, with the early adopters winning the performance battle.
What follows is not a criticism of them personally, they are representative of a narrow church; thankfully, some have seen the reality dawning and have already acted, with more coming forward—clients’ fortunes matter.
Pearls of Wisdom
In essence, the starting gun in the video was an articulation that safety was a concern and the analogy of accountability of those building driverless cars, the risk of being prosecuted should someone die was emphasised. This was meant to suggest this may impede acceptability of A.I. into the industry due to accountabilities and liability. On a similar theme, one said the industry is slow to accept new ideas due to bad experiences of the past, and safety comes first. Another reflected and said people might be fearful of losing their jobs – meaning asset managers. Yet another, “people don’t trust computers,” asset managers are comfortable as they are, and investors need to speak to someone. By far the most significant comment, one said asset managers make a PROMISE of delivery and maintaining client’s conviction is essential. Finally, a key point, that a Balanced Fund may diverge into a Growth Fund.
Zero recognition of the real world around them. Let me explain.
Let’s examine the value of these views, but before I do, here is a short description of our SigmaPlus A.I. supported advisory service: It is long-only, plain vanilla highly liquid assets, the universe of assets, the risk profiles and constraints are set by the asset manager and they control the trades, so oversight and control is 100%. This is not the famous Medallion Fund with its complexity of 5,0000 shorts and 5,000 longs running simultaneously. SignmaPlus digests a large volume of the day’s data and forms a consensus on whether to continue to hold, sell or buy a given security – something that every asset manager does in his daily routine, except it is impossible for them to fully digest the high dimensional data. SigmaPlus portfolios are not high turnover, they are dependent on the strategy and the asset mix, and portfolio turnover is generally 0.5 to 1 per annum. Indeed, built into the system is a design to hold securities for as long as possible to minimise trading costs.
The link with Cars killing people, accountability and liability – is silly?
The system runs with a complete absence of human whims and behavioural biases, which is the cause of so much inefficiency and poor decision making. This is a big step in the right direction.
Stretching the idea of A.I. supported portfolio management to driverless cars and risking death or even fearing a regulatory punishment of the same order leaves me wondering about the minds controlling our assets. It is a meaningless analogy.
A.I. in asset management is not mystical, and the assets do not suddenly morph into gremlins, or start eating investors, they work with data to optimise outcomes – very successfully too.
I hope this kills the “High-Stakes-Coral” idea of deaths and prosecutions.
Slow To Take On New ideas = Safety
Most certainly the industry is slow to take on new ideas, but to suggest this is to avoid catastrophes is plain absurd, the most significant losses come when asset/fund managers get “smart” and diverge from their original plan, the risk profile changes and people lose money. Had they forgotten the recent example of the “Guru” Neil Woodford, there is no better example of a changing risk profile and the human ineptitude of those continuing to hold his fund seems lost on everyone. Asset managers adopting SigmaPlus have more time to spend digging into the asset universe they proposed, looking for exactly these changes. Illiquidity does not belong in the liquid space.
Fact, investors, in growing numbers, are not trusting their asset/fund managers; this is the reason the world is becoming Buy/Hold – another disaster en route? This sad fact is a reality, and it is a great shame because asset managers give tremendous value – just not always on the return performance scale.
Let’s look at how that Promise is fairing – Is this the delivery of the Promise?
527 Flexible Funds. Risk – Horizontal, Annualized Returns – Vertical Data 12 years to-date.
I took May 2008 as the starting point deliberately to include the 2008 downturn. I could have been more cruel and started at Oct 2007 – at the time of writing the FTSE 100 is now below this point. Asset Management is a global affair, but there is a human tendency to stay close to home.
Whatever is going on hasn’t worked in the past, whatever that Promise is, it isn’t translating into outcomes.
By Comparison the S&P Balanced Equity and Bond Index – Conservative. Ticker: SPBXCI gave an annualised return of approximately 7% over the period. It is designed to allocate a 25% weight in core equity (S&P 500 TR) and 75% weight in fixed income to provide a regularly rebalanced multi-asset measure for conservative risk/reward profiles.
WITH THESE RESULTS, ONE MUST STOP TO THINK, WHAT ON EARTH ARE THESE PEOPLE THINKING. THEIR MARKET IS SHRINKING DUE TO THE INDUSTRY FAILING TO MEET THE MOST BASIC OF CLIENT EXPECTATIONS – A LITTLE AGGRESSIVE, BUT “HEAD AND CLOUDS” COMES TO MIND.
Humans Need Human Communication
Absolutely, investors do need to speak with someone; humans need humans – no doubt. In fact, adopting A.I. support would give more time to support clients directly. More time to understand clients’ needs, their circumstances and ongoing queries as well as doing a host of technical tasks that A.I. is not presently doing. This “defence” only goes to show the attempt to justify the status quo is not just weak, but entirely out of the ballpark.
One volunteered that when A.I. supports Doctors to diagnose diseases, the results are significantly better than if the doctors made the decisions alone – yes. My god, many surgeries using lasers would be impossible without computer aid – would you let the shaky hand of the surgeon laser your eyes to improve your eyesight? People do trust computers when driven through safe hands. I could write a whole book on this, the fact is computers are trusted by smart people and are indispensable in almost all walks of life – and this now includes, some, forward-thinking asset managers. Even those not accepting A.I. are still totally dependent on computers to do the hard grind on data; the problem is – everyone gets the same data, and so the interpretations don’t vary much across asset managers – hence the similar results. It has been shown many times they are chasing the same assets.
Asset managers are getting smarter, but this brings with it a problem; they all interpret data similarly.
Don’t Trust Computers
“People don’t trust computers” perhaps they mean people don’t trust computers to make decisions alone. Come on – we are talking about efficiency and recognition of signals not evident to the current tools they use. Computers can take in high dimensional data and make sense of it, humans cannot. A bit of repetition of argument, without computers making thousands of decisions per minute, your aeroplane journeys would be a whole new experience. In this context, computers reduce accidents and make flights a much smoother experience, and safer.
When SigmaPlus gets something not quite right, it may sell early, or buy at the start of a dip, but it would correct the errors very quickly without remorse – the markets are stochastic, so some poor trades are inevitable. Humans often buy at the wrong time and justify themselves by holding on to poor performing assets believing they were right, and if given enough time, this will prove correct. That’s a performance killer.
Why did so many professionals continue to hold the Woodford Fund? Because they believe in fairy tales? They believed the man was bigger than his results – the computer would not make this “mistake.”
There is so much evidence that humans are not good at making decisions even under simple uncertainty scenarios, now add a continual varying flow of information and we are then in computer space, not human. The idea mere mortals can digest, analyse and marshall vast volumes of data and speedily direct this into optimal decisions is self-evidently wrong.
Changing Risk Profiles
I recently wrote a response to a Citywire article, highlighting the expressions asset managers used to justify their asset allocation decisions. A salutary lesson in human strategies – some dominant examples were, Intuition, Guesswork and Heuristics. I need to be careful; there are very many smart people working in this field, using tech and impressive strategies to overcome the hurdles robbing them of performance. Valiant as this may be, performances do not support them beating their benchmarks over time.
Risk profiles can change dramatically when asset managers are under pressure to solve an underperformance problem or maintain a good one, especially under challenging conditions. There are many examples of this over media history and even more under cover of darkness. Changing risk profiles is a very human foible – not so with a computer; they keep the game on track as directed by humans.
The fact is asset managers are usually making judgements on portfolio risk mainly based on asset classes, their quality and mix – pity this didn’t work with Neil’s fund – underlying risk divergence is not uncommon. Concerns typically arise when they see the mix deviates from the benchmark(s) or risk codified by percentages in equities and bonds, and most certainly at the level of simplicity, it makes good sense. But these measures do not work well in our current environment. “It worked in the past, so will work now” is not a good strategy.
The mix of Bonds and Equities being a good proxy for limiting risk (volatility) is right, but historically we have not had to face negative yields.
Those asset managers who held onto bonds in the face of zero interest rates are now seen as super-human by their peers as interest rates went negative, but what now? Well here is what Brian Funk head of credit research at MetLife Investment Management said
“This is a global reach for yield. Every rock has been turned over at least three times,” said Brian Funk head of credit research at MetLife Investment Management. For Funk, emerging market debt, both corporate and sovereign is a “least bad” choice for investors. “It’s offering value on a risk-adjusted basis at this stage in the cycle,” he said. Source CNBC
SigmaPlus is not designed to work in the direct Fixed Interest space; all assets, including cash, are pitched against each other. If we imagine a seesaw – at its most simplistic level volatility can be controlled thus.
The man on left (Equity Investor) calls for Cash when opportunities arise and give back when there are none.
There are times when CASH IS KING, but is it the ongoing complexity of factors that determine the right level? Not really. If SigmaPlus determines there are fewer and fewer opportunities in the securities space when taking into account the risk profile, then the cash level adjusts automatically.
In our world, the Cash element of any portfolio is managed by the asset manager.
That said, we do run portfolios of fixed interest funds and do very well.
This was not covered in the discussion except in the context of changing risk profiles, but it is worth covering.
Our system is entirely Agnostic; it holds no preconceived notions or second guesses on its perspective of the state of the markets. Every day the system continues to reassess the convenience of keeping the same stocks (or funds) and sectors in the portfolio, selling and replacing as needed. On average, just under 10% of the portfolio is modified every month.
A typical daily question by the digital team is, where and how much should be invested today while minimising stock turnover, and considering exchange rate risk and hedging costs (if appropriate) for, say, a euro investor? An answer is returned such as 57% US Equities, 37% European equities and 6% Cash, with a turnover of once per year. The same questions materialise each day. A second question is, in which Sectors and with how many securities should be invested today. Let’s look at a historical response – Headings: Sector – Weight – How many titles.
For those well-practised in analysing sectors, this model equipped with artificial intelligence may seem quite defensive. But it had an overall gross return of more than 50% as of 1st January 2017 and a positive return in 2018.
A very brief interrogation of this sectoral allocation. The first twelve sectors weigh more than 50% of the portfolio and none of them is an important historical component of the relevant indices. There are no banks, financials, industrial or oil stocks. Inevitably, the investment landscape is changing, driven by sustainability and the technology revolution. But, new technologies in many cases still need old technologies. Even electric cars to communications need electricity; this is why the most significant presence in the portfolio is that of electric utilities and multi-utilities.
Compare this with frequent comments of the following type, ‘we are a little nervous of xxxx market, so we are presently underweight,’ ‘YYY Fund Manager has had poor performance for a while, but we like him, we are sure it will be turned around,’ ‘X is looking over-priced so we are considering reducing our exposure,’ I could go on.
Suck your finger and put in the air to see which way the wind is blowing?
We support active asset managers, but we need to be honest. Active asset managers have been heralding the return of volatility to bring the buy/hold brigade back into the fold, well now is that time. The reality is, asset managers have been struggling in rising markets, so not falling as far as benchmarks is not a reward cherished by investors. Sure, buy/hold investors will likely/maybe suffer more in the short to medium term, but is this enough? They have had years of outperformance to cover short term downturns. A prolonged downturn may bring, some, back to the active management fold. Don’t bet your shirt on it.
A Robust Repeatable Set of Outcomes is Needed
Investors generally have very reasonable expectations of their asset managers, and they deserve every effort from them to secure robust performances.
Humans Don’t Do Well Under Uncertainty
They also don’t do well at spotting complexity in the market; A.I. is uniquely able to “see” what humans cannot. Intelligent design with a “digital team” able to digest this information and turn it into a robust decision-making process working within well-defined parameters is the future, well for some it’s the present.
“Head in the sand” means running risks with your clients’ portfolio performances, and this will manifest more trouble for the active industry as has already been seen by the massive shift from active management to Buy-Hold.
This is a tough-talking article, but the field needs a shake-up, it is what needs to be said, and I believe, fairly reflects reality. Don’t let your clients endure; take a positive step forward.
Clients Must Win.
Come on, set aside the grumps and explore. Clients are on the receiving end, embracing a move away from old practice to the new could mean more stable and rewarding outcomes.
A.I ASSET MANAGEMENT IS HERE – ITS ROBUST AND STAYING
Call us, and we will work with you to build a framework between our A.I. advisory support and your management team.
I used the content from the video and interpreted statements into meanings. I aimed to highlight mindsets within wealth management and expose erroneous thinking. One person in the video seemed more clear on the realities than the others. The video only served to give me a brief guide on content. I have deliberately kept names out of the matter because this is not about them – its the mindset of the industry that needs shaking.
Harry Holt BSc MSTA CFTe, Director at Perimeter Consultancy