Some of the strategies we track use economic data, like the unemployment rate, when making investment decisions. Like 99.99% of strategy backtests you’ll encounter, we’ve always taken the shortcut of basing our historical results on that economic data as it looks today.
The problem is that introduces a degree of “lookahead bias”. Economic data is often initially reported at one value and then later restated (often multiple times). Basing our analysis on the economic data as it looks today, which includes all those later revisions, means that we’re assuming an investor had access to data they couldn’t have had at that moment in time.
To remedy this, starting in early May, we’ll begin using “vintage” economic data for all strategies that trade based on economic data.
What is “vintage” economic data?
Vintage economic data is data as it looked at that moment in time. So, for example, if a strategy is calling for a 12-month average of the unemployment rate on 12/31/1999, we’re going to calculate that based on the 12-months of unemployment data that was actually available on 12/31/1999.
That’s more complicated than it might first appear. For example, the data for November may have been revised by the end of December, and we would include that revision, but we wouldn’t include any revisions to November that happened after the end of December. Of course, it’s our job to manage that complexity so it’s seamless to you.
Impact to historical results:
A total of 8 strategies will be affected by this change. We track other strategies that trade based on economic data, but it’s either data that’s not subject to revisions, or in one case data that we already vintage (Link’s Global Growth Cycle).
Below we’ve listed those 8 strategies and the % of historical monthly allocations that will change:
|Strategies Impacted by Vintage Economic Data|
|Growth-Trend Timing – Original||1.7%|
|Growth-Trend Timing – UE Rate||0.9%|
|Lethargic Asset Allocation||0.9%|
|Movement Capital’s Composite Strategy||3.3%|
|Novell’s Bond UI1||1.9%|
|Resilient Asset Allocation||0.6%|
In no cases will good strategies become bad strategies, or bad ones become good, but knowing that 2-3% of your backtest is incorrect is still quite significant.
All backtests involve a degree of uncertainty, especially when we try to estimate results in the distant past, but we strive to come as close to reality as possible. We’ve talked a lot about steps we take to ensure better results, like investing in quality data and not using subpar data sources.
This is another step towards that goal. We know that we have a special responsibility as the leader in Tactical Asset Allocation analysis, and we hope that this change will inspire others to follow suit.
Expect this change to be rolled out early next month. We’ll post an announcement in the members area when the change is complete. If you have questions, please don’t hesitate to reach out.
We invite you to become a member for about a $1 a day, or take our platform for a test drive with a free membership. Put the industry’s best tactical asset allocation strategies to the test, combine them into your own custom portfolio, and follow them in real-time. Learn more about what we do.