London (HedgeNordic) – Within the tactical trading space, CTAs were traditionally systematic and used technical data while global macro funds were discretionary and used fundamental data. These boundaries have become blurred in many cases as managers broaden out their data inputs and blend data types – and some discretionary managers have quantitative substrategies. IPM – Informed Portfolio Management, which marked its 20th anniversary in 2018, has been consistently systematic and fundamental. IPM’s approach differs from traditional, discretionary fundamental macro managers in many respects, but also has some overlap.
“As a systematic manager, I believe we have a broader view of the economy across all geographies and asset classes, whereas discretionary managers are often more focused on a few particular themes that are in vogue at the moment”, says IPM’s CIO and Head of Research, Björn Österberg. IPM, which now runs the largest liquid single hedge fund strategy in the Nordics, nearly always has active positions in all the markets it trades: equity indices; volatility index; Global Macro: Systematic or Discretionary? Björn Österberg, CIO and Head of Research – IPM government bonds; developed market currencies and emerging market currencies. Positions are driven by a wide number of themes, whereas some discretionary macro managers can be concentrated into four or five themes.
This is because, “IPM’s models are designed to identify a broad set of inefficiencies that may be off the radar screen of discretionary managers looking at big themes”, he adds. “IPM focuses on items out of the spotlight”, says acting CEO, Lars Ericsson (the new CEO, Arne Hassel, starts in July). “Our models are also based on a consistent view of historical relationships between fundamentals and asset prices, whereas discretionary managers are more likely to be forming new expectations based on beliefs about the future”, Österberg continues. When market regimes change suddenly, the best discretionary managers may have an edge.
Österberg acknowledges that good discretionary managers may be adept at interpreting new events, such as Trump’s election victory, or other political surprises, and they may outperform during such phases. Equally, a potential pitfall of discretionary investors, identified by behavioural finance, is overconfidence: “discretionary managers may be tempted to form an opinion in situations where they do not have a comparative advantage, such as political or geopolitical events where it is hard to claim an informational edge over the aggregate market”, he says.
“Discretionary managers may be tempted to form an opinion in situations where they do not have a comparative advantage, such as political or geopolitical events where it is hard to claim an informational edge over the aggregate market.”
IPM is also different from directional discretionary macro managers in that 85% of IPM’s strategy’s risk budget is based on relative value, with only 15% taking an outright, directional view. IPM’s four families of relative value models take views within its four asset classes, and also take care to avoid accidental beta bets which can come about through exposures such as the currency carry trade. “For instance, going long of emerging market currencies has an implicit beta of 0.5 to global equities”, explains Österberg. IPM’s directional sleeve can however result in the portfolio becoming net long or net short of emerging versus developed market currencies; equities; bonds, or the VIX, but these wagers are fairly small.
IPM’s four families of models contain a total of 85 distinct ideas. The number has grown over time and each individual idea evolves over time. Two of IPM’s families of models – valuation and risk premia – gauge fair value, yield and income rather than predicting market direction. The allocations to models move up and down according to the opportunity set. For instance, the risk premia models have a greater weighting when risk premia are higher, whereas other managers often size position based on risk contribution. IPM’s opportunistic approach to risk-taking is probably more typical of a discretionary manager whereas many systematic managers target constant volatility.
IPM’s biggest model family is macroeconomic, based mainly on traditional economic data such as trade and leading indicators. The market dynamics model acknowledges that investment decisions are not always driven by hard economic data – and therefore looks at risk sentiment and flows. IPM provides transparent performance attribution, including by asset class and model. In recent years, a short stance in the Swedish Krona has been a big winner for IPM. In 2019 year to date the strategy’s winning positions have included longs in the British Pound, Mexican Peso and Russian Rouble, paired against shorts in other currencies. Its losing positions have included being long of European equities versus US equities, and being short of
Australian Government bonds vis a vis other government debt. “Discretionary managers may be tempted to form an opinion in situations where they do not have a comparative advantage, such as political or geopolitical events where it is hard to claim an informational edge over the aggregate market.”
Data and Modelling
Data inputs are one area of common ground between IPM and discretionary macro managers. IPM’s fundamental data inputs include macroeconomic releases, prices, news, sentiment, estimates, and forecasts. “Even with nowcasting, the data is rather low frequency and the information content is sparse and slow moving”, says Österberg. This most naturally explains why IPM tends to trade over multi-month time horizons.
The time lags and gaps in data also mean it does not make sense to follow, say, an unsupervised machine learning/statistical learning approach, which might ‘let the data speak’ and let models identify possible relationships between data and markets. In common with a typical discretionary manager, a plausible and intuitively sensible hypothesis is the starting point for IPM’s models. “We are very theory-based. When we model things, we always make sure we incorporate a prior belief”, he points out. The data landscape is rapidly evolving. “Our database is growing exponentially, with a lot of new datasets becoming available, and we have a whole team dedicated to scouting for new data”.
IPM makes use of some alternative datasets, including unstructured data such as satellite images, which have been structured by data vendors. “We plan to add at least four new datasets this year”, he says. Even so, the alternative data basket is less than 10% of the total. Some 70% of data inputs are classic economic data, and 20% are more advanced economic data. “We expect data costs will continue to rise. These costs are paid out of management fee income”, says Ericsson.
“We are very theory-based. When we model things, we always make sure we incorporate a prior belief.”
“Systematic and discretionary macro are complementary approaches”, says Österberg. Indeed, IPM’s return profile has historically shown no correlation to global macro, (and nor to CTAs, other hedge fund strategies, nor conventional asset classes). IPM is also lowly correlated to nearly all other systematic macro managers. ”But this is just one of the dimensions. Historically our primary objective was to help diversify traditional portfolios of equities and bonds. Nowadays the lack of correlation versus other diversifying strategies is probably more important.”, he points out.
Many allocators will place IPM in their systematic, tactical trading bucket, which could include systematic macro, CTAs, short-term traders and possibly strategies based on machine learning and artificial intelligence). IPM’s client base now spans the globe, including institutions in the US, Canada, Australia, and China.
This article featured in HedgeNordic´s special report on systematic strategies in June 2019.
Title Pic By Liu-zishan—shutterstock.com