Ford Equity Research - The Quantitative Validation Process™

Quantitative Validation Process™

At Ford Equity Research, we base everything on publicly available numbers, because we think facts yield the best results. Our proprietary research and model development techniques, the Quantitative Validation Process, serve institutional and individual investors by ensuring every investment decision is made with the best possible information.

Collection

In our fundamental database operations, we collect emerging data from sources we believe to be accurate and reliable: Company press releases, SEC filings, and third-party data suppliers. Redundancy is built into the system; we look at online, printed, and faxed resources to ensure accuracy.

Ford Equity Research has been collecting and storing fundamental data as it was reported, without restatements, since 1970. Our database, the only one of its kind in the equity research universe, is the framework upon which Ford builds its successful equity valuation models. Each model represents a tested, validated, descriptive historical relationship between fundamental factors and equity returns.

Comparison

After data collection, we compare various data to test for accuracy. In this step, data can be compared to expectations (earnings estimates, for example), previously reported data for the same company, and relationships with other data items for the same company. Any data that demonstrate major deviation from baselines are flagged.

We also test our research ideas through analysis techniques, including single-variable and multivariate regression, sector rank performance, correlation analysis and model portfolio analysis. We compare actual results with expected results based on previous research results, experience, published research, and informed intuition.

Investigation

In the investigation phase, data that were flagged during Comparison are further investigated, based on reviews of original sources. We carefully re-read, looking for reasons that reported amounts seem out-of-line with expectations and predictions. Secondary sources of the same data may also be checked.

We likewise investigate research results to reconcile them with expected outcomes. We use sampling techniques to confirm the accuracy of underlying data used in a study, and we review research steps to verify that the test methods used were not designed or carried out in error.

Evaluation

Following the investigation process, we evaluate whether the data is correct as stands or needs to be adjusted based on a precise set of pre-determined rules. Some instances require a judgment call; in these cases, the analyst must determine whether to adjust suspect information alone, through consultation with peers, or with the advice of a senior analyst.

In model development, research results are shared with colleagues who ask probing questions about the data and research techniques used. Typical discussions involve results, expectations, and investigation steps. Commonly, a colleague will attempt to independently evaluate research results by replicating the study.

Validation

Once the previous steps are completed, we validate the data, either making a correction or leaving the original value unchanged. Investment models are validated on the basis of their ability to reliably predict equity returns. Once backtesting has established that the model has broad applicability in the equity markets, we publish the results. When a model is especially strong, we add the model results to the Ford database.