Use Case 4: The Buy-Sell Decision

This series of Deception And Truth Analysis (D.A.T.A.) articles has covered actionable investment insights use cases for the outputs of our DATA algorithm. In this piece we cover “the buy-sell decision.” Admittedly this article is a bit more speculative than our other articles. Here’s why:

• We consider ourselves a screening tool and not a diagnostic. That is, we seek to assist investors in their investment process by rapidly surfacing actionable investment insights, rather than making decisions for D.A.T.A.’s Clients. We are your partners; we are not you.
• In the portfolio management asset allocation decision, our suggested use cases are entirely speculative. That’s because to our knowledge no one is currently using D.A.T.A.’s metrics and algorithm to construct/asset allocate their portfolios. Nonetheless, we think you will see that our proposals are compelling.

As a review, there are five major due-diligence steps in almost every investment process. They are:

1. Establishing your investible universe.
2. Relying on high-level criteria to identify state changes.
3. Deeper dive analyses that establish an investment thesis.
4. The buy-sell decision.
5. Ongoing due-diligence of your investments.

Deception And Truth Analysis (D.A.T.A.) provides a powerful assist in each of the above steps. Now let’s discuss the buy-sell decision…we think you will be pleasantly surprised.

Deception And Truth Analysis provides a number of outputs on its platform. That said, our primary output is our DATA Score. It ranges between -100% and +100%, with any negative score indicative of a document’s level of deceptiveness and any positive score indicative of a document’s level of truthfulness. Our back tests have demonstrated that DATA Scores are predictive of future securities price returns. Below we discuss seven possible asset allocation/portfolio management ideas making use of DATA Scores.[i]

Below we discuss seven possible asset allocation/portfolio management strategies that can be implemented where the S&P 500 is the investment performance benchmark. For each strategy we use:

1. DATA Score assessments of each of the S&P 500 components’ 10-Ks issued in 2021.
2. Then the asset allocation/portfolio management strategy is implemented on the first trading day of the following year, that is: 2 January 2022.
3. We then look at the performance of the asset allocation/portfolio management strategy on the last trading day of the year, 31 December 2022.

     Note: This article is not meant as a validation of our algorithm, just a demonstration that our outputs may be used in constructing/asset allocating for portfolios.

Clearly, above, there is a lag in the implementation of the strategy from a document being assessed by us and its DATA Score then being used to implement an asset allocation/portfolio management strategy many months later. The average lag from a 10-K being assessed with a DATA Score and then the purchase of the security is 9.17 months, or 279 days before implementation. The longest lag is 11.60 months and the shortest lag is 0.56 months.

As you can see below that there remains signal in securities prices and as predicted by DATA Scores with such long lags is surprising. Yet, this is a result that we consistently find with DATA Scores. Our theory is that numbers are outcomes driven by managements’ choices. In turn, managements’ choices are driven by their behaviors. DATA Scores measure deceptive behavior and thus, we believe they are predictive. Or, at least that is our theory for why we consistently see the results we do in our tests.

In 2022, the performance of the S&P 500 was -19.6445%.[ii]Each of the strategies below is compared with this performance.

Strategy 1: Long-Short, Weights Determined by Median DATA Score

For asset allocation / portfolio management Strategy 1 we propose that DATA Scores may be used to establish the weights within a portfolio. In this instance we:

1. Calculate the median DATA Score for the S&P 500’s component companies’ 10-Ks issued in 2021, which is 12.17%. This calculation is in cell G508.
2. Next, in Column H we calculate a figure that is each company’s DATA Score divided by the median DATA Score of 12.17%. This creates a series of weights to include in the portfolio.
3. The average weight for S&P 500 companies would be 100% divided by 495 holdings (some of the companies have dual share classes) = an average weight of 0.2020% for each holding if it were equal weighted. This calculation is in cell I503.
4. In column I we then multiply the average weight of a holding of 0.2020% by the overweight calculated in Column H. This tells us the weighting for each company to include in the portfolio. Note: in some instances, because the company’s DATA Score is below the median score it has the effect of creating a short position for the security.
5. In Column Y the stock price for each company on the first trading day of the following year, 2 January 2022 is shown. In Column Z the stock price for each company on the last trading day of the following year, 31 December 2022. In Column AA, the change in stock price year over year is shown.

The performance of Strategy 1 for 2022, described above is: -15.3878% vs. S&P 500 performance of -19.6445%. Thus, the return advantage of Strategy 1 is:

+4.2566% or +425.66 bps.

Strategy 2: Long-Short, Weights Determined by DATA Scores

For asset allocation / portfolio management Strategy 2 we propose that DATA Scores may be used to establish the weights within a portfolio. In this instance we:

1. Calculate the total of all DATA Scores for the S&P 500’s component companies’ 10-Ks issued in 2021, which is 59.0674. This calculation is in cell G502.
2. Next, in Column J we calculate a figure that is each company’s DATA Score divided by the total of all DATA Scores of 59.0674 to establish the portfolio weights for the strategy. Note: in some instances, because the company’s DATA Score is negative, meaning in the aggregate its 10-K is assessed as deceptive it has the effect of creating a short position for the security.
3. In Column Y the stock price for each company on the first trading day of the following year, 2 January 2022 is shown. In Column Z the stock price for each company on the last trading day of the following year, 31 December 2022. In Column AA, the change in stock price year over year is shown.

The performance of Strategy 2 for 2022, described above is: -15.6937% vs. S&P 500 performance of -19.6445%. Thus, the return advantage of Strategy 1 is:

+3.9507% or +395.07 bps.

Strategy 3: Long Only, Weights Determined by Median DATA Score

For Strategy 3 we employ exactly the same asset allocation / portfolio management strategy as in Strategy 1, except that we only look at the performance of the long companies; that is, the companies whose DATA Scores were assessed as Truthful (i.e. DATA Scores >= 0%) in the aggregate.

The performance of Strategy 3 for 2022, described above is: -15.3872% vs. S&P 500 performance of -19.6445%. Thus, the return advantage of Strategy 1 is:

+4.2572% or +425.72 bps.

Strategy 4: Long Only, Weights Determined by DATA Scores

For Strategy 4 we employ exactly the same asset allocation / portfolio management strategy as in Strategy 2, except that we only look at the performance of the long companies; that is, the companies whose DATA Scores were assessed as Truthful (i.e. DATA Scores >= 0%) in the aggregate.

The performance of Strategy 4 for 2022, described above is: -15.6931% vs. S&P 500 performance of -19.6445%. Thus, the return advantage of Strategy 1 is:

+3.9514% or +395.14 bps.

Strategy 5: Long Only Truthful Companies, Equal Weight

For Strategy 5 we shift gears entirely by taking away the asset allocation / portfolio management decision entirely and simply equal weight all of the companies whose DATA Scores are assessed as truthful (i.e. DATA Scores >= 0%). This strategy features 441 truthful 10-Ks from companies and excludes 54 deceptive 10-Ks from companies. Thus the portfolio weights for each holding are 1 divided by 441, or 0.2268%.

The performance of Strategy 5 for 2022, described above is: -12.4322% vs. S&P 500 performance of -19.6445%. Thus, the return advantage of Strategy 1 is:

+7.2122% or +721.22 bps.

Strategy 6: Long Only Companies with DATA Scores > 1 Standard Deviation Above the Mean, Equal Weight

For Strategy 6 we indulge an assumption. Namely, that the companies issuing the most truthful documents probably outperform the companies issuing less truthful documents. Thus, we build a portfolio that only includes companies whose documents score 1 standard deviation above the mean. The mean DATA Score was 11.93% and the standard deviation was 9.78%. Thus, the threshold DATA Score for inclusion is 21.71%, as shown in cell N506.

Again, we take away the asset allocation / portfolio management decision entirely and simply equal weight all of the companies whose DATA Scores are assessed as truthful. This strategy features 67 10-Ks from companies and excludes the remainder of the 495 companies. Thus the portfolio weights for each holding are 1 divided by 67, or 1.49%.

The performance of Strategy 6 for 2022, described above is: -19.1358% vs. S&P 500 performance of -19.6445%. Thus, the return advantage of Strategy 1 is:

+0.5086% or +50.86 bps.

Strategy 7: Use DATA Score Year on Year Change Rank, Equal Weight

Our very last example, Strategy 7, does not directly use our primary output, the DATA Score. Instead, we create a portfolio of the securities that have the greatest year on year change in DATA Scores. Our thought here is that companies whose performance is improving likely want to tell the world about this performance. No need to be bashful if you are kickin’ it performance-wise.

Once again, we equal weight the companies. But here there is a bit of trickiness. Namely, what constitutes the “greatest improvement” in DATA Scores year on year? Because this decision is arbitrary we show a number of possibilities below. To calculate the weighting for each security, simply divide 1 by the number of companies included in the inclusive category of: greatest year on year change.

For the following x companies to include here is the performance vs. the S&P 500:

• 40 companies: -18.3863% vs. -19.6445%, +1.2582% or +125.82 bps
• 50 companies: -17.4071% vs. -19.6445%, +2.2374% or +223.74 bps
• 75 companies: -15.7633% vs. -19.6445%, +3.8812% or +388.12 bps
• 100 companies: -13.6297% vs. -19.6445%, +4.2566% or +425.66 bps

     Note: you can download our spreadsheet and toggle the figure yourself by changing the number of companies to include in cell G526.

We have demonstrated in this article the ways in which outputs from Deception And Truth Analysis may be used to help investors engaged in “the buy-sell decision.” Specifically, by proposing ways that our scores may be used in the asset allocation / portfolio management decision. While the above work is far from complete, and is certainly not robust from a validation perspective, we believe the above results are intriguing.

We would love to hear about your adventures in asset allocation / portfolio management using our DATA Scores and other outputs. Let’s have fun out there!

__________ 

[i]Each of these strategies is shown in the following spreadsheet:  Deception And Truth Analysis – Use Case 4 – The Buy-Sell Decision.xlsx 

[ii]Comes from Yahoo! Finance download of S&P 500 open on 2 January 2022 of 4,778.14 and the adjusted close on 31 December 2022 of 3,839.50.