Processing forward looking data (analyst estimates)

Many analysts and investors prefer working with forward looking estimates. For example, forward looking valuation (1-year forward P/E) is a widely used valuation metric. But the challenge here is what constitutes 1-year forward?

Next 12-month estimate calculation

In the above example, what is 1-year forward EPS to be used for calculating the P/E ratio? Suppose we are in October 2022. Does EPS estimates for March 2023 constitute 1-year forward EPS? But March 2023 is just 6 months away. Neither does estimates for March 2024 constitute 1-year forward since March 2024 is 18 months away.

To solve this problem, we use next-12-month (NTM) methodology for all financial metrics for which we have forward looking estimates (e.g., EPS and Sales). NTM estimates are calculated as weighted average of estimates for the next 2 financial years. The weight assigned to the next financial year estimate is equal to the number of days remaining in the financial year divided by 365. The rest of the weight is assigned to next-to-next financial year estimate.

In our example, on 1st October 2022, EPS (NTM) will be calculated as follows:

EPS (NTM) = 0.5 x EPS Estimate (FY23) + 0.5 x EPS Estimate (FY24)

And on 1st January 2023, this formula will change to:

EPS (NTM) = 0.25 x EPS Estimate (FY23) + 0.75 x EPS Estimate (FY24)

This method ensures an accurate depiction of 1-year forward estimate and hence the derived valuation (or any other) metrics.

Note that we could have used sum of next 4 quarterly estimates as our NTM estimate. But there are a few issues:

1. Quarterly estimates are generally not available for stocks
2. If available, it is only available from a few analysts and hence average and median calculations are rendered unreliable.
3. Analysts generally build yearly models and focus on the next quarter. Rest of the quarters (especially those further away) tend to be balancing items. Hence, quarterly estimates that are 3 or 4 quarters away are extremely noisy and not reliable.

Other important points

When using NTM derived metrics (like forward P/E) in screener and signals, those stocks that have less than 5 analysts covering them are dropped. This ensures that NTM estimates are not a spurious estimate of few analysts.

• NTM estimates are consensus estimates. We use the median of all estimates as consensus. Use of median ensures that consensus is not impacted by any large outlier (potentially faulty) estimate.