Hi Ihor,
Book portfolios can sometimes be difficult to troubleshoot because their constituent portfolios can have different rules limiting their trades that are not immediately apparent. Sometimes we cannot see these limitations because they might be built into the model by the limitations of a particular parameter.
Upon the first impression, it sounds like at least one of your input models could have a final date of 6/30/21 (rather than 7/2/21). You’ll need to go through each ‘feeder’ model and make sure that for each of them, their last update occurred at the close on July 2 (last Friday).
Also, make sure that you don’t have a parameter limiting the last assessment to last Wednesday (June 30). While I can’t think of an example off the top of my head that determines the end of a simulation, I have seen this happen at the start of a Book’s performance because (for example) one of the feeder models uses a long-term moving average that limits the beginning of the run.
For example, in one feeder strategy I used a Composite Breadth Indicator that combined the Advance-Decline Percent, High-Low-Percent, High-Beta-Low-Beta Percent, and Percent-of-Shares-Above-200-DMA – then required the composite signal for all of these indicators to be above its 250-DMA (one-year avg) for a BUY signal. At first, I couldn’t figure out why the backtest wouldn’t begin until it was a year past the start date I entered. Then I realized it was because that 250-day parameter was hidden in the Breadth Composite, and that was the aspect requiring the model to have 250 days of live trading (assessment) before issuing its first BUY signal.
Books can provide exceptional advantages despite the occasional frustration when you combine multiple strategies configured to work together with uncorrelated results. I have a Book Strategy called the ULTIMATE 6-Model (9 ETF) Combo Strategy that captures synergies and attains a CAGR of about 30% and a Max Drawdown of only -13% since 2007. The Avg. Annual Max Drawdown (what you might expect any given year) is only -8%!
The correlations among most of the feeder models are low, but when they’re not (EQUITY+ and EQUITY++), it is by design.
[img]https://etfoptimize.com/images-external/p123/210705-ult-6-book-correlations.png[/img]
And here’s the equity-curve results of this diversified Book of ETF-based models:
Books can be a real challenge to develop properly with uncorrelated diversification that identifies different types of risk, but as shown by the example above, it can be well worth the time and effort if you have enough experience to build it to be adaptive and protect you against the market’s ever-changing primary risks.
I can’t give you a more accurate assessment of what could be wrong without seeing your Book’s particulars. You can DM me if you don’t want to make your model public to the entire community. Or you can also directly contact Paul (paul@portfolio123.com) or another P123 staff member (yuval@portfolio123.com) and I’m sure they can help you.