Senaca Capital’s Ascinaa G10 Currency Futures program has won The Hedge Fund Journal’s CTA and Discretionary Trader Award for Best Performing Fund over 5 Years ending in December 2023 in the Currency – Statistical Pattern Recognition Strategy, based on risk-adjusted returns.
The return profile is remarkably consistent with no losing calendar years since it started in 2019. While currency trend following strategies have in recent years benefited from the resurgence of trends most notably in the Japanese Yen, Senaca is not a trend following strategy and does not need persistent or large moves in currencies.
It is disarmingly refreshing to find a manager trading a single signal on a relatively concentrated investment universe of liquid currencies.
The reported returns understate potential returns in a fund format. The numbers are purely trading profits and as the strategy trades FX futures on margin, in a fund structure it would also be earning considerable interest, always on unencumbered cash and sometimes on the entire NAV. On average the strategy only has risk on for two months per year and was only positioned for three months in its most active year. It might in theory even have zero risk for an entire year if the signal is not triggered.
In a hedge fund world of increasing complexity with a myriad of models, signals and exotic or alternative markets, often deploying machine learning, statistical learning, artificial intelligence and other techniques, it is disarmingly refreshing to find a manager trading a single signal on a relatively concentrated investment universe of liquid currencies.
Before designing the Senaca strategy, manager Martin Delahunty had never done any discretionary trading of any asset class, because he was too busy running a business in the marine systems sector in the US. His engineering background was of some use in structuring the systematic analysis, back testing and risk management.
5
The system is based on data for five currencies versus the USD since 2000. “The pattern we seek is based on how they move against the USD. The signals for entry points are very similar for all five currency pairs and do not change from day to day,” says Delahunty.
Delahunty spent over 10 years developing the distinctive currency trading strategy, based on a systematically identified pattern. The signal has a very high hit rate, but the more unpredictable element is that its frequency varies considerably from year to year. Nonetheless, he expects to profit every calendar year.
The system is based on data for five currencies versus the USD since 2000. “The pattern we seek is based on how they move against the USD. The signals for entry points are very similar for all five currency pairs and do not change from day to day,” says Delahunty, who devised a proprietary coding methodology for the back test.
Delahunty was in no rush to launch the strategy and spent years patiently and cautiously testing it with his own capital before marketing to outside investors. The strategy was piloted with proprietary capital between 2012 and 2017 to understand how the signal and currencies move in real time and fine tune risk management rather than maximise profits. “I was trading to understand and test the pattern and would not be trading in the same way now. I wanted to understand how to mitigate risk,” recalls Delahunty.
The strategy is not predicated on Delahunty’s trading prowess. “I could see how traders operate, but I am not a trader in the true sense. I have never done any discretionary trading. I spent 8 or 9 years looking at the data before launching the product for outside investors. I do not want to take any unnecessary risk, so the statistical framework has to be really solid, and the trading should be secondary, within that framework,” he explains. Indeed, as assets have grown, Delahunty has not noticed increased trade execution slippage.
The framework is mainly mechanical but there is some limited degree of discretion around how to trade the positions once the signals happen. Very occasionally discretion could also be exercised to stay out of a market, such as the British Pound around Brexit. Delahunty judges that, “The element of discretion is so small that a back test before live trading began is still substantially useful”.
“I do not want to take any unnecessary risk, so the statistical framework has to be really solid, and the trading should be secondary, within that framework.”
Martin Delahunty, Manager, Senaca Ascinaa G10 Currency Futures
The frequency of signals per calendar year has ranged from 60-70 times in 2019 down to only 8-9 times in some other years. Since the signal has a very high hit rate, its frequency also quite accurately predicts the overall level of profits and the first year of the strategy, 2019, has been the best calendar year to date.
Performance drawdowns have been very small and only occurred once every 2 or 3 years. Ascinaa has a worst drawdown of 0.89% between February and March 2021, which was less than its annual volatility of 1.54%, while Oceanaa, which runs at four times higher volatility, correspondingly has roughly four times these metrics: a worst drawdown of 3.99% over the same period, which is also less than its annual volatility of 5.55%.
Round turns per million average 1,500 in a busy year for the lower volatility strategy Ascinaa and 6,000 for Oceanna, but they will vary dependent upon the number of signals annually.
Even when the signal is activated for some currencies, the stars are virtually never aligned for all of them. Very seldom does the strategy have simultaneous exposure to all five currency pairs. Around 80% of the time it is in only one pair, and 20% of the time it may be only in two or three of the pairs. “We always trade the pair with the highest probability of best positive capture during the expected move,” says Delahunty. Holding periods can range from intraday to 20 days and most trades are between intraday and 5 days.
The strategy tends to scale in and out of trades once moves become dynamic.
The signal predicts a minimum level of potential profit, and trades might be exited after this initial target is attained. The amount of profit beyond the minimum target does vary. “Sometimes we will stay in the trade after the minimum level is met, even though the probability of further profits has declined,” says Delahunty.
Levels of exposure and stop losses vary: “We may sometimes trade very low margin to equity and keep a very tight stop loss,” says Delahunty.
Unlike some currency managers, Delahunty does not fret about a lack of trends or volatility in currency markets. He has observed the signal occurring during trending, range bound and mean reverting currency markets, and believes the strategy is unique to the foreign exchange markets he trades against the USD.
The approach is based on statistical probabilities but at a relatively digestible level as no machine learning, statistical learning, AI or generative AI is used.
“It is a mathematical model, rather than a global macro strategy,” says Delahunty. No fundamental or economic data is used, though Delahunty does closely monitor currency market reactions to economic data releases in terms of how prices move relative to the data. He sticks to his circle of competence as defined by the signals: “Even if I had a hypothesis about economic data, I would not trade it because I am not an economist”.
Divining the secret sauce of the signal sounds like finding a golden needle in a haystack and the painstaking research involved a considerable amount of trial and error. “The biggest problem was seeing patterns emerge that only worked for a short period. We might have categorised 200 movements and find 199 of them have a very low probability of positive capture,” says Delahunty.
Data inputs cover a range of shorter term and higher frequencies. Data used can be high to low ranges, for periods including open to close, four hourly, hourly or even down to tick levels. “It is price data rather than price action in a true sense over many periods. We know what starts to happen when certain criteria are met,” says Delahunty.
The Ascinaa G10 Currency Futures program runs with volatility of 1-2% and an average margin to equity of 1-4% while the Oceanaa G10 Currency Futures program simply has four times the exposure of Ascinaa: volatility of 4-8% and an average margin to equity of 8-16%. Fees are calibrated to the volatility target: Ascinnaa charges 0.25% management fee and Oceanaa 1%.
The minimum account size of USD 330,000 for Ascinaa is based on contract sizes for currency futures. The minimum account size for the Oceanaa is USD 82,500 or one quarter of the level. For notionally funded accounts the minimums can be lower, though Senaca do recommend certain minimum levels of funding.
The strategy is currently marketed to US investors. The investor base includes funds of funds, high net worth individuals and family offices, as well as a Chicago proprietary trading house. A $5 million ticket was garnered earlier this year taking assets to over $18 million.
Delahunty is exploring various platforms for asset raising and could consider launching a fund as assets grow.
Certified Public Accountant (CPA) Lisa Casagni manages operations. She is also Delahunty’s wife, and the two worked successfully together for many years in the marine systems business.
Service providers on the broker side include several US brokers active in the CTA space: StoneX Financial, Wedbush Securities and Straits Financial Group.
Senaca has offices in the UK and US. Regulation is typical for a CTA: NFA and CFTC. Assets are currently below the threshold for SEC registration.