The sweet spot trading time horizon

The investment philosophy is to trade narratives over very short time periods, post economic data, policy and news releases. “We do not let ourselves become wedded to narratives, because often one narrative is quickly superseded by another,” says Kokkila.

For instance, traditional fundamental narratives around exchange rate valuation can falter. “In 2023, Swiss Franc to Japanese Yen is at the highest level since 1979, and the Norwegian Krona is weak despite a huge current account surplus,” says Kokkila. Thales have no firm view on “correct” valuation for a currency since their time horizon is measured in hours or days.

If some currency managers fixate on long run views, at the other extreme, Thales believe that G10 FX trading volumes are dominated by algorithms. “High frequency traders have become key players with hardly any staff. And initial price reactions to events tend to be very linear, immediately pricing in a surprise rate decision,” says Kokkila, who has been trading and perfecting the strategy over the past eight years and is an ex-fund manager for Forte Securities. Kokkila started trading FX for his personal account as a university student in 2012 during the European sovereign debt crisis.

This still leaves nuances for humans to exploit. “Despite all the hype on language learning models, we think it is particularly difficult to interpret complex scenarios algorithmically as the number of scenarios is infinite and market conditions change. The algos and LLMs you use are only as good as the data they have been taught with,” points out Kokkila, who has final responsibility for trading and risk management; analyst Antti Maimanen is responsible for pre-trade research and interpreting events, while trader Kari Kumpulainen is responsible for monitoring risk levels, execution and entry prices. The team sometimes work at night in shifts to trade market moving events, such as key policy decisions in Japan, Australia and New Zealand.

Currency universe

Thales have historically traded all spot G10 FX, with some temporary exclusions. “The Swiss Franc was excluded between 2015 and 2021, since its monetary policy decisions did not move markets because rates were pinned at zero and it followed the ECB very closely. Since 2022, we have traded it relatively actively as the monetary policy and intervention stance took a fundamental turn,” explains Kokkila. (Thales would not have traded the Swiss Franc before it de-pegged in January 2015 since they do not trade pegs.) Additionally, GBP was briefly excluded when it was driven by Brexit headlines, but trading has resumed since 2021. And at times during 2020, AUD, NOK and SEK were also eschewed due to illiquidity, but now less liquid units are traded in smaller size and with wider stops. Thales find that the largest and most liquid currency pairs have contributed most, and the correlation between return streams per pair has been low. The strategy might venture into emerging markets by adding some modest Mexican Peso exposure.

Events traded

Thales trade events include economic data releases, such as jobs or inflation; monetary policy decisions, statements, reports, projections, press conferences, speeches and minutes of meetings. Risk budgets per event vary between countries and over time. “Some events have always been very tradeable, such as the Fed and ECB. Between 2015-2017, the most important economic releases across the main G10 moved the corresponding pairs. Now since 2022, it has been mainly about the US data,” says Kokkila. One year can make a big difference: “Bank of Canada was our best event in 2021, and our worst in 2022. The reaction function of CAD has changed, and Canadian monetary policy has become very correlated to the US stance,” he says.

The devil is often in the details of data releases: “Interpreting detail versus reacting to headlines is the edge of our process. For example, now on inflation reports, the services ex-housing is a key metric to watch,” says Kokkila.

Forward-looking events, such as statements, projections and follow-up press conferences, are often the most market moving. In contrast, speeches and minutes are more variable: “Usually speeches stick to the script from preceding statements, but they can be quite market moving if information has changed in the interim. Minutes are very rarely market-moving, as the information is stale,” says Kokkila.

Ad hoc, special, unannounced, emergency etc. central bank meetings and actions are a special case that Thales have begun trading more in 2022 where they were confident about extrapolating from history: “For the Japanese Yen interventions, we could prepare and subsequently capture alpha because we had a lot of historical data and could roughly guess when they would intervene. Similarly, the BOE action after the sterling crisis was expected and we had a rough idea of how to trade it,” explains Kokkila.

Beyond economic data, financial stability can come to the fore in crises: “In 2023, the regional banking crisis in the US moved EUR/USD higher, firstly because Europe was perceived to have higher financial stability, and secondly because markets anticipated a Fed pause to protect banks’ balance sheets, whilst the ECB had headroom to hike rates. We captured this event after the ECB decision, which was impacted by financial stability,” says Kokkila.

New leaders of central banks can also be important: “New BOJ Governor, Ueda, has been quite dovish, highlighting a non-zero chance of inflation moving above 2% in the next 2 years in the April 2023 press conference. This differed from Kuroda and was interpreted very dovishly, since the markets had ascribed a higher probability to inflation overshooting,” Kokkila explains.

Events are nearly always uncorrelated, except for multiple events on the same day, which receive special attention: “Careful analysis is undertaken as to which ones to trade and with how much risk budget, and whether the events have spillover impacts, which can sometimes lead to de-risking,” says Kokkila.

Backward and forward-looking analysis

Thales has built a historical database of over 2,000 events, which informs analysis and decisions but cannot always be extrapolated from. “History sometimes rhymes but does not repeat itself. And history sometimes repeats itself but does not rhyme,” says Kokkila. The database helps Thales to determine similarities and differences between current and historical events. This is balanced against forward-looking analysis of positioning, trends and risk appetite, to build out potential scenarios. One playbook often highlighted by financial media such as Bloomberg radio is the knee jerk reaction, when the initial response to a data headline is rapidly reversed upon closer examination of the numbers. “The knee jerk reaction occurs when an event is mixed. It is just one pattern and there is no such thing as the most common pattern. We analyse every single situation,” says Kokkila.

Gauging market colour and opinion

Thales itself can also reverse an initial position in response to subsequent news events and market action, relative to expectations. Thales carefully interpret various data to monitor sentiment and positioning. Option markets are not traded but can be one source of intelligence to determine broad brush market context, such as high and low volatility regimes “However, option implied moves around data releases are not always a reliable guide to the actual market action,” says Kokkila.  Similarly, CFTC Commitment of Trader reports are only one piece of the jigsaw and can overlook other important data. Similarly, Bloomberg Consensus data can be useful for data events, though it can differ from implied futures market pricing and sometimes needs to be modified by “whisper numbers”.

Regime analysis: monetary policy dispersion

Events also need to be percolated through a prism of regime sensitivities, in terms of policy convergence/divergence, inflation/disinflation, and recessions/crises.

Some judgement is needed to define the FX regime, which can shift around even during a steady inflation regime. “In 2022 we had high inflation, moderate monetary policy divergence, and higher volatility in FX. Currently in 2023, there is enormous monetary policy divergence between Japan and the rest of the world, but overall, we see most central banks moving in lockstep in terms of rate rises, so we think the actual divergence is moderate. We are in a low volatility, and high inflation, regime,” says Kokkila, who has witnessed the opposite regime. “In the financial crisis of 2009-2011 and the subsequent recovery phase we had high volatility, moderate inflation and high divergence. Even though all banks eventually took rates to zero, the lag between policy responses created enormous trading opportunities, especially if you understood the impact of unconventional monetary policy.”

The divergence and volatility regimes for currencies can gyrate even in steady macroeconomic conditions: “During 2014-2017 we were in a deflation, high monetary policy divergence regime, with high volatility. The divergence mechanics were mainly based on unconventional policy tools and future rate expectations. Thereafter 2018-2019 was a low monetary policy divergence, deflation, and low volatility regime,” recalls Kokkila. That climate also resumed after the Covid crisis: “Between H2 2020-2021, we were in a recovery regime, with minimal inflation, low volatility and low monetary policy divergence,” says Kokkila.

Recessions and crises

Recessions can reverse normal relationships between rates and currencies: “In the first half of 2020, we had a crisis regime, where easing decisions could be interpreted counterintuitively positively for a currency. This helped us develop new insight into crisis regimes, which were turned into profitable trading,” recalls Kokkila.

Systems and discretion

All of this is synthesized into a substantially systematic framework: risk management, trading rules, position entry and exit, and leverage per scenario, are all systematically determined. Discretion, however, is exercised in unanticipated situations, which are run with lower risk, but even then, the structure of risk, leverage and timeframes apply. “Our motto is we trade based on rules not based on views,” says Kokkila. The rules-based trading embedded in the strategy also controls behavioural finance biases, such as confirmation and availability bias.

Refinement of risk management rules

Thales have honed and refined techniques for scaling in to and out of positions. A staggered ladder of stops is used, and gap risk is rarely experienced, in part because Thales do not hold events close to stops. Stops can be moved to breakeven but will not be trailed beyond that, and some of the longest running positions have been held for a week or two.

Thales has spent many years optimizing volatility, diversification and risk management for the strategy. Risk limits for position sizes and stop losses are partly based on the risk class conviction category of the event, and Kokkila has found that sizing higher conviction events larger has been additive for returns.

(A trade can sometimes move up and down the risk classes during its life, though this is relatively infrequent. If price action is completely random, Thales may refrain from trading or could downsize. Positions might also be reduced due to surprise risks such as whisper numbers deviating from consensus numbers. It could be increased if the actual scenario is even more dovish than the one anticipated).

Risk limits are also scaled according to the volatility target of the strategy, which can be accessed at 95% monthly Value at Risk (VaR) levels of 6.5%, 9% and 12%; Kokkila is personally invested in the strategy.

There were double digit drawdowns in 2018, 2019 and 2021, which led to a revision of the risk limits, introducing greater diversification across more events, and reduced risk limits (e.g. from 4% to 2% per event for the 6.5% VaR strategy), designed to prevent another double digit drawdown.

Seeding sought

The Finbou strategy had been operating under a Swiss corporate structure but is transitioning to Finland in 2023 for easier marketing and regulation through fund passporting. The institutional version of the strategy will be launched early in 2024. It will only trade currencies and earn interest on cash. On top of existing firm assets of circa EUR 35 million, seed capital of around EUR 30 million is sought and family offices are showing a keen interest in the strategy. The first institutional fund will only trade FX events, but Thales is currently researching the additions of other asset classes, such as gold or energy or bonds to a new strategy.

Opportunity set and outlook

The strategy is not long volatility per se, but levels and types of volatility can influence the opportunity set. The strategy has made lower but positive returns during low volatility regimes such as the first half of 2023. Thales expect that inflation could stay higher for longer. “In the past couple of years, there have been irreversible geopolitical changes (trade wars, war in Ukraine). Add to that, changing supply chains, green transition, and labour shortages due to demographics, and there is more reason to believe that currency flows will become more volatile,” says Kokkila.

A regime of higher and more persistent inflation could have profound implications for currency markets. “Whereas 2019 saw very little monetary policy divergence and very low volatility, there is now potential for much more market-moving events. For instance, if inflation targets of 2% prove to be unrealistic and the targets are formally, or informally, revised up to even 3%, this has dramatic repercussions for bond markets. Just a 1% overshoot in inflation could reduce ten-year bond returns by 12% over a decade,” points out Kokkila.

In addition, the composition of inflation is starting to change, which makes it more challenging for markets to analyze and react to the drivers of inflation. “We think markets will move to high volatility, higher inflation, and higher monetary divergence regime as inflation proves more persistent as the last few percentages of core inflation will be harder to squeeze out. Central banks are caught between a rock and a hard place in terms of killing inflation without killing growth,” says Kokkila.

This could spark “currency wars” of competitive depreciation. “Inflation targeting is not enough to stabilize exchange rates, so currencies depreciate contributing to higher inflation. JPY, NOK, SEK are all good examples,” says Kokkila. A further market mover is that the USD is also challenged as a reserve currency. Overall, there are plenty of reasons for larger and more divergent currency moves around economic events.

The strategy is currently available to investors on Darwinex.com under the name Thales. Past performance statistics and current AUM can be found there. Darwinex is regulated by the Financial Conduct Authority in the UK and the Comision National del mercado de Valores (CNMV) in Spain.