Institutional Positioning


LuxArbor is a data provider dedicated to tracking major institutional positioning in financial markets. Our goal is to shed light on the very opaque world of holdings. Coverage currently includes country allocation for developed world and emerging market equities as well as sector rotation for U.S., European, Japanese and Canadian large cap equity funds. Fixed income fund positioning is also available for USD-denominated sovereign and corporate emerging market funds. 

Sample chart 1

What's factored in?

Our model tracks average country/sector weighting for major mutual funds. This allows us to not only track the consensus but also the allocation of the best (or lagging) managers.

Sample chart 2

Where is the competition over/underweight?

As a result, our model also identifies which countries/sectors are over/underweighted by fund managers over time.


How do we track market exposure?

Our dynamic model is return-based and calculates the most likely country/sector positions to match a fund’s performance. We do not consider published fund holdings in our model because 1) the information is usually weeks old when announced and 2) window dressing in some funds can lead to misleading results.

Our model is not a rolling regression. The problem with this methodology is that country/sectorweights tend to be concentrated in a few countries/sectors and allocation often exhibits huge jumps from day to day. Our model minimizes unrealistic allocation changes while staying flexible.

We run our model daily, after the close of the market.

How do we know it works?

Out-of-sample calculated returns exhibit a low tracking error with actual fund performance. To test our model, using our calculated daily weights, we “invest” in country/sector total return indices for the next day. On average, for all large cap funds with an S&P 500 benchmark, the annualized tracking error is well below 3.0%. We are continuously tweaking our model to reduce this error (or course, we make sure take the appropriate measures to avoid overfitting.) A 3% tracking error is what you would get if you built a model to track an S&P 500 ETF priced around $10 and that model was off by +/- $0.01 per day over the past decade. In other words, our tracking error is lower than a rounding error.


Is there any persistence in returns?

Following how the consensus moves is interesting to understand what the market has factored in. However, the appeal increases significantly if following what leaders are doing can lead to outperformance.

For individual investors, choosing a fund is often driven by past performance only. Academic researchers have looked at whether or not top funds usually continue to outperform their peers in subsequent months. This is called return persistence. The conclusions of these research papers vary widely. Our own analysis shows that it is possible to find value in understanding how top managers are positioned.


Focus on all top funds, not single funds

Academic researchers have focused on whether or not a single top quartile fund has a high probability of remaining in the top quartile. This is important because investors usually invest in few funds, not all of them. However, the aggregate fund performance is more useful. The following table shows the average spread in forward returns for funds that have been outperforming vs. funds that have underperformed. Since 2005, investing in the top quintile of funds over a three-month horizon and selling the bottom quintile would have resulted in an outperformance in the following months.

Backtest results

Short-term persistence exists

The first thing to note is that persistence works well over the short term. The winners of the past two or three months are likely to continue to win over the near future. We call those “hot hand” managers. We also note that over the long terms, higher returns do not last.

A backtest

Over the past ten years, a strategy that would have followed the sector weighting of the top quartile of funds with the best performance over the past previous three months would have resulted in a significant outperformance.

Quartile backtest

The difference in returns is even greater looking at the top decile vs. the bottom decile. An investor who invested in top decile funds would have outperformed bottom-decile funds by 72% over 10 years.


Finally, we note that the strategy seems to be working constantly through time. Over the past ten years, the cumulative spread in total returns has been increasing at a stable rate over the long run.

Difference backtest