Quarterly Data Estimates
Why We Need Them & Our Methodology for Creating Them
In order for us to build accurate company valuation models for our portfolio management clients, we have to estimate the value of certain data points in quarterly periods. These estimates are necessary because companies often omit footnote disclosures in quarterly filings that they include in annual filings. We take a very conservative approach to these estimates to ensure that our model is driven by disclosed data and not our estimates as much as possible. Items commonly estimated include operating leases payments data, pension data, & employee stock option data, but can include any data reported in the footnotes or MD&A.
We make two types of estimates for quarterly data: forward-looking estimates and backward-looking estimates.
Forward-looking estimates are necessary when the most recent filing is a quarterly filing. For example, ABC Corp’s most recent filing is its 10-Q for the first quarter of 2021. Since the most recent filing is a quarterly filing, forward-looking estimates based on data disclosed in the 2020 10-K may be used when building a quarterly model for the period.
We estimate balance-sheet-like data (snapshots of data at a particular time) by pushing forward the last reported value from a prior period. The last disclosed value can be from a prior quarterly or annual period. For example, if ABC Corp disclosed in its 2020 10-K that it had 100 million ESOs outstanding and did not disclose any ESO data in its 2021 Q1 filing, we would use the last reported value of 100 million ESOs outstanding in our 2021 Q1 model. If ABC Corp had disclosed ESOs data in the Q1 filing, we would use the disclosed data and make no estimates. This is a conservative approach to data estimation. We do not grow (or decrease) balance-sheet-like values with our quarterly estimates.
We do not make forward looking estimates for quarterly income statement related data. Hidden non-recurring charges are unusual and we’ve found no reliable way to estimate them due to their infrequent nature. Therefore, we do not estimate these types of values and only use them in our model when they are disclosed in a filing.
Backward-looking estimates are used when the most recent filing is an annual filing. Once we process the annual filing for a fiscal year, we update our prior quarterly and TTM models with backward-looking estimates that backfill quarterly data based on data only disclosed in the annual filing. This backfill is necessary because annual filings contain additional disclosures that are often omitted in the quarterly filings. For example, ABC Corp’s most recent filing is its 10-K for 2021 and contains disclosures for a large unusual expense that was not disclosed in any quarterly filings. So, we backfill the quarterly models for the 1st, 2nd, 3rd and 4th quarters of 2021 with data so that the 2021 quarterly models add up to the results in the full-year 2021 model.
Point-in-Time Backtest Data
In our backtest data files, only data available on the publish date is used to generate models. No data from any future filings is ever used to estimate data in quarterly periods (or any periods).