Institutional Edition – Release Notes 20231212.8

The Finish Line

Parameters available for the first launch

In light of the latter consideration, we have found it necessary to make certain compromises with respect to the parameters. Specifically, the following parameters:

• Target cash
• Sector-based rebalancing
• Fees
• Options related to liquidity checks
• Reselect all will be concealed or rendered inactive for the time being. These parameters will automatically assume values of “0,” “none,” or “no.”

Furthermore, certain parameters will be immutable:

• The trim up and trim down parameters are set at upper and lower rebalance boundaries of 33%.
• Testing is limited to equally weighted strategies.
• Selection of benchmarks is confined to the S&P/TSX Composite and S&P 500 (depending on the country)

Limitations

We acknowledge certain limitations inherent in our current backtesting framework.

Backtest with Variables with no Data:

The backtest will not work properly if a variable is missing data at the start date. Transactions will not be executed until the entire model is supplied with data.

Backtesting Capacity: – Corrected on Dec 19, 2023

Conducting extensive backtests involving a high number of maximum stocks and an extended historical period may trigger rejection by the backtest engine. For example, a portfolio comprising 60 stocks per period since 1985 is likely to yield no results. In such instances, users will need to observe a waiting period of 7 days before reinitiating the backtest for the specific strategy. Nevertheless, users can continue to conduct backtests on other strategies unaffected by this constraint.

Hidden Parameters: The deployment of backtest settings occurred some time ago, and since then, several settings have been hidden. This concealment of parameters may yield unexpected outcomes. To preemptively address this potential issue before executing an existing strategy, we strongly recommend users to click on the “reset to default” button, which will reset the backtest settings to defaut, and subsequently update their parameters.

The “reset to default” button can be found by:

1. Clicking on “edit strategy”.
2. In the ranking/buy/sell rule settings, click on continue to be redirected to the backtest settings.
3. The “reset to default” button is in the top-right corner.

U.S. Estimate Variables: Added on Dec 13, 2023

Currently, U.S. estimate variables are not available for the backtest. However, they should be added in about a week.

Our Methodology to Review our Backtest Results

We have determined that the outright rejection of our backtest results due to minor deviations from CPMS was an inaccurate approach. Recognizing that a slight variance at the outset of the backtest could trigger a cascade of differences—referred to as the butterfly effect—we deemed it necessary to adopt an independent validation methodology instead of solely relying on direct comparisons with CPMS.

In evaluating the outcomes of our backtest, we adhered to key principles throughout the verification process:

1. Perfect Match Criterion:
   • Were the results identical to CPMS? If yes, deemed perfect.
2. Non-Identical Results Analysis:
   • If the results were not identical:
   • 2.1. Consistency Validation: – Could the result be validated based on Inovestor’s buy/sell rules, Inovestor’s data, and selected parameters, demonstrating internal consistency?
   • 2.2. Systematic Comparison: – Could it be demonstrated that Inovestor’s backtest is systematically superior or inferior to CPMS?
   • 2.2.1. If superior, deemed perfect.
   • 2.2.2. If inferior, considered unacceptable.
   • 2.2.3. If impossible to conclusively determine superiority, deemed reasonable.

These principles in action include instances where the portfolios at initiation were identical, specific parameters and variables produced slight differences, and certain buy/sell rules resulted in nuanced variations between Inovestor and CPMS.

For example:

• Out of 10 stocks, one transaction differed due to nuanced disparities in buy/sell rules.
• The assessment of the trim up and trim down algorithm, governing the quantity of shares transacted when acquiring a new company, revealed complex dynamics, with outcomes sometimes aligning perfectly and at other times exhibiting marginal differences. The decision to launch was made in the absence of conclusive evidence favoring one algorithm over the other.

We approached these instances with a nuanced perspective, considering factors such as platform speed and stability enhancements, standardized user experiences, and the inherent complexity of algorithms governing share transactions. Where feasible, an immediate resolution was undertaken, and parameters deemed unnecessary were removed. In cases where parity between Inovestor and CPMS could not be definitively ascertained, a decision to proceed was made, acknowledging the inherent complexity of the trim up and trim down the algorithm.

Track Changes

We are pleased to outline the features and enhancements scheduled for release alongside our upcoming backtest:

1. Partial Summary Page:
   • Displaying returns exclusively.
2. Transactions Tab:
   • Providing a detailed overview of transactions.
3. Holdings Tab (Formerly Named “Portfolio” in CPMS):
   • Offering insights into the current portfolio holdings.
4. Sector Allocation:
   • Illustrating the distribution of holdings across sectors.
5. Backtest PDF Report:
   • Despite remaining typos, we have decided to release this feature to provide users with a preview.
6. User Interface Improvements:
   • Addressing typographical errors.
   • Introducing the ability to skip backtest settings in the strategy edition.
   • Substituting “latest data” with the production date in the buy/sell list, among other refinements.

In addition to these planned releases, we also highlight recent updates from the last week or two:

• Professional Licenses (“Advisor Edition users”):
  • Users with professional licenses now receive emails from Inovestor regarding the models.
• Factor Attribution Section:
  • Launched two weeks ago, this section provides insights into the attribution of factors.

We remain committed to continually enhancing our platform and providing our users with a seamless and comprehensive experience.