Artificial Intelligence Research Reporting – YEAR 2016


Successful picking on the long leg
While the Financial Times dated 23/10/2016 reported that 99% of actively managed US equity mutual funds failed to outperform their benchmark, our long-only strategy has generated in 2016 a price return of 24.82% in excess over the SP500, without any leverage, and any small cap or style bias.

Of course, because Bramham Gardens is a research firm, we do not account for transaction costs and slippage. With about 10 rotations per year in the portfolio and as we work on a liquid and cost effective market like the US large cap universe, costs would however remain below 1 or 2% per year.

Price based – ex dividends SP500 Long only
Return in 2016 9.53% 34.36%

A robust long-short
Obviously, with a long-leg generating such outperformance, running an unleveraged long-short strategy with low directionality constitutes the natural next step. In 2016, we obtained a gross-of-fee-and-cost performance of 17.04%, with a positive skewness (meaning limited drawdowns).

Gross of fees & costs Long-Short
Return in 2016 17.04%

No trick

In order to evaluate how much directionality we carry, we compute on a yearly basis the 12-month correlation between the Deutsche Bank market-neutral investment style driven risk-premia and our long-short strategy. We can observe two things: first of all, a very limited dependency on the traditional style factors in 2016, but also a slight shift away from “low-beta” towards more “value” in the portfolio, which was the right move ex-post to have implemented.

As a Research Lab, Bramham Gardens concentrates on building investment strategies based on Artificial Intelligence.

The team is expanding, with Arnaud Apffel joining us as a partner. Arnaud brings us 25 years of experience in Investment Banking and Asset Management, and will provide immense value in terms of investment structuring, academic content delivery and business development.
Two young scientists are joining Bramham Gardens as well, helping us refining several Artificial Intelligence related technical aspects on which we wish to develop.

As was already mentioned in the end of November 2016 Bramham Gardens Research Report, Boussard & Gavaudan, the well-recognized asset management firm, is working to open a UCITS Sicav fund, which follows our long-short US strategy. This fund is scheduled be launched at the start of February 2017, pending CSSF approval.

We are glad to announce that assets invested on our long-short US strategy are reaching $80 million in Q1-2017.

3. INSIGHT IN THE BRAMHAM GARDENS DNA: no magic, just work!

Let us articulate the core principles, which drive our asset allocation choices:

  • Comparing all stocks on the basis of an equal level of information: in our universe, there are around 400 stocks monitored live. Avoiding asymmetrical information among stocks is key for us, as emotional proximity to a given company can be a source of misallocation of capital.
  • Taking no industry rotation bet: a long experience in equities has shown us that sector rotation tends to generate a lower Sharpe ratio than single stock selection. This year has been particularly telling in this respect, requiring remarkable skill to get in the financial sector while moving away from healthcare. Our view is that proper sector diversification should be kept at all times.
  • Taking no specific “investment style” bet: there are many money managers who favor value or momentum as key drivers for their investment decisions. Again, getting the timing right for investment style rotation looks too difficult to be a dependable source of performance, from our perspective.
  • Focusing on relative bets within homogeneous industrial peer groups: we do not consider that we should be paid to predict the dynamics of the stock market indices overall. Our job is to identify resilient outperformers / underperformers and focus on them without being too concentrated on any of them.
  • Relying on Machine Learning to select stocks: adaptive learning is paramount to be able to distinguish between an expected future outcome, pure noise and an anomaly. Using altogether historical information and the multi-dimensional analysis of the peer group dynamics enables us to make complex judgment calls based on rules that evolve and adapt over time.
  • How does it compare with the typical investment process of active managers? In efficient markets like the US market is, it should not be the privileged access to information that drives performance. Defining a robust and sensible metric to capture company talent and applying it with a rigorous and agile methodology can be a game changer.
  • Avoiding accidents by averaging: qualitative analysis often leads to concentrated portfolios, as gaining superior information requires time, focus and capacity. On our side, we make every effort to spread bets looking at different entry points, different universes, various metrics for talent measurement. Averaging is thus an important element of risk management in our view.

We wish you a happy 2017 and are of course at your entire disposal to discuss about what we do :