Trailing P/E divides a stock's price by the GAAP earnings per share it actually reported over the last four quarters; forward P/E divides the same price by what analysts expect it to earn in a coming fiscal year. One is a fact with a timestamp, the other is a forecast with a margin of error. Neither is "the" P/E. The useful signal is the gap between them, and on the October 16, 2026 close that gap ranged from almost nothing at Apple to a canyon at Nvidia and Exxon Mobil. Below are the formulas, a worked table on eight large US stocks, and the traps that make the two numbers lie.
The two formulas, side by side
Trailing P/E = share price ÷ trailing twelve months (TTM) EPS. TTM EPS is the sum of diluted GAAP earnings per share from the last four reported quarters. It changes only when the price moves or a new quarter gets reported.
Forward P/E = share price ÷ consensus EPS for a future fiscal year. The denominator is an average of analyst estimates, and it changes whenever any analyst revises a number, which can happen on any trading day.
A quick illustration with a hypothetical $100 stock: if it earned $4.00 over the past four quarters and analysts expect $5.00 next year, the trailing P/E is 25 and the forward P/E is 20. The market price is identical. The only difference is which year of earnings you believe in.
That distinction matters most when earnings are moving fast, in either direction. When profits are flat, the two ratios sit close together. When profits are surging, collapsing or distorted by one-off items, they can tell completely different stories about the same stock.
Worked example: eight large caps on the October 16, 2026 close
The table uses closing prices from October 16, 2026, TTM diluted GAAP EPS from each company's last four reported quarters, and the FMP analyst consensus for the first fiscal year the company had not yet reported as of that date. Fiscal years do not line up, so the FY-end month is listed. The "n" column is the number of analysts behind the EPS estimate. Consensus figures may be built on adjusted rather than GAAP earnings, so treat the trailing-versus-forward comparison as approximate.
| Ticker | FY ends | Close (Oct 16, 2026) | TTM GAAP EPS | Trailing P/E | Consensus EPS (FY) | Forward P/E | n |
|---|---|---|---|---|---|---|---|
| $AAPL | September | $343.46 | $8.73 | 39.3 | $8.84 (FY ending Sep 2026) | 38.9 | 29 |
| $MSFT | June | $529.05 | $17.95 | 29.5 | $19.75 (FY ending Jun 2027) | 26.8 | 31 |
| $NVDA | January | $214.13 | $7.91 | 27.1 | $9.28 (FY ending Jan 2027) | 23.1 | 33 |
| $AMZN | December | $250.00 | $12.43 | 20.1 | $12.79 (FY 2026) | 19.5 | 39 |
| $GOOGL | December | $340.59 | $19.91 | 17.1 | $20.59 (FY 2026) | 16.5 | 41 |
| $JPM | December | $339.72 | $23.84 | 14.3 | $24.80 (FY 2026) | 13.7 | 11 |
| $XOM | December | $178.98 | $7.77 | 23.0 | $12.76 (FY 2026) | 14.0 | 11 |
| $COST | August | $989.11 | $20.76 | 47.6 | $22.81 (FY ending Aug 2027) | 43.4 | 27 |
Two quirks jump out. Apple's fiscal year ended in late September 2026, but the company had not reported it by October 16, so its "forward" P/E of 38.9 is mostly a look at a year that is already over. Costco, by contrast, reported its August fiscal year in early October, so its forward column already points at a year that runs to August 2027. Same label, very different time horizons.
Where trailing and forward P/E split apart, and why
The table looks tidy until you add the following fiscal year. That is where the real story sits.
| Ticker | Trailing P/E | Forward P/E (first unreported FY) | Forward P/E (following FY) | Following-FY consensus EPS |
|---|---|---|---|---|
| $NVDA | 27.1 | 23.1 | 13.6 (FY ending Jan 2028) | $15.79 (n=32) |
| $XOM | 23.0 | 14.0 | 15.2 (FY 2027) | $11.81 (n=14) |
| $GOOGL | 17.1 | 16.5 | 22.3 (FY 2027) | $15.30 (n=41) |
| $AMZN | 20.1 | 19.5 | 23.4 (FY 2027) | $10.66 (n=41) |
Nvidia: growth compresses the multiple. The FMP analyst consensus for the fiscal year ending January 2028 is $15.79, roughly double the $7.91 of TTM EPS. Put that against the $214.13 close on October 16, 2026 and a stock trading at 27.1 times trailing earnings drops to 13.6 times. That is the bull argument for high-growth names in one line. It is also the risk: the cheap-looking number depends entirely on earnings that do not exist yet. For context on how the market has been pricing that growth, see our coverage of Nvidia's record high amid bond-market tension.
Exxon Mobil: a weak trailing year. Exxon's TTM EPS of $7.77 includes a $1.00 quarter (Q1 2026) and a $1.53 quarter (Q4 2025). Consensus for calendar 2026 is $12.76. Since Exxon booked $4.48 in the first half ($1.00 plus $3.48), hitting that consensus would take about $8.28 in the second half. That is why the trailing P/E of 23.0 and the forward P/E of 14.0 look like two different companies. Whether the second half delivers is exactly what the forward number assumes.
Alphabet and Amazon: the reverse trap. Here the forward P/E for 2027 is higher than the trailing P/E, because consensus EPS for 2027 is below 2026: $15.30 vs $20.59 at Alphabet, and $10.66 vs $12.79 at Amazon. The reason sits in one quarter. Alphabet reported Q2 2026 diluted EPS of $9.11, and its July 22, 2026 earnings release says other income "reflected a net gain of $98.0 billion, primarily the result of net unrealized gains on our equity securities." Amazon reported Q2 2026 diluted EPS of $5.75, and its July 30, 2026 release says that quarter's net income "includes non-operating pre-tax other income of $53.4 billion," primarily from its investments in Anthropic. Those gains sit inside GAAP EPS, so they inflate the trailing figure and make the trailing P/E look cheaper. The lower 2027 consensus is consistent with analysts not modeling gains of that size again. Read alone, a 17.1 trailing P/E on Alphabet flatters the stock; the 22.3 multiple on 2027 estimates is the less flattering view.
Five mistakes that make P/E comparisons meaningless
- Mixing GAAP and adjusted EPS. Trailing P/E is usually computed on GAAP diluted EPS. Many forward estimates and company guidance use adjusted figures that strip out items like stock compensation or investment gains. Dividing price by GAAP for one stock and adjusted for another is not a comparison.
- Using stale estimates. Consensus numbers move every time an analyst updates a model. A forward P/E quoted from a three-month-old screen may reflect a world that no longer exists, especially right after an earnings report.
- Ignoring fiscal-year mismatches. On October 16, 2026, "next fiscal year" meant a year ending September 2026 for Apple, June 2027 for Microsoft, January 2027 for Nvidia and Walmart, and August 2027 for Costco. Comparing those forward P/Es head-to-head compares different calendar windows.
- Confusing NTM with FY. Some screeners use next-twelve-months (NTM) EPS, a blend of two fiscal years. As a simple time-weighted approximation for Microsoft on October 16, 2026, roughly 8.5 months of the FY ending June 2027 ($19.75) and 3.5 months of the FY ending June 2028 ($23.54) gives about $20.86, or a P/E near 25.4 versus 26.8 on the fiscal-year number. Same stock, same day, different "forward P/E."
- Trusting thin coverage. JPMorgan's 2026 consensus rests on 11 analysts and Exxon's on 11, versus 41 for Alphabet. One outlier estimate moves a small-sample average more. Check the analyst count before treating a forward P/E as settled.
One more trailing-side trap: a single odd quarter can distort the TTM sum. Meta's Q3 2025 diluted EPS was $1.05, against $6.18 to $10.44 in the three quarters after it. That quarter still sits in Meta's TTM EPS of $26.54, which puts the trailing P/E at 26.7 on the October 16, 2026 close of $709.33, versus 23.2 on 2026 consensus. It drops out once Meta reports its Q3 2026 results.
How to calculate both yourself
- Pull the closing price for a specific date and write the date down.
- Find diluted GAAP EPS for the last four reported quarters in the company's earnings releases or 10-Q and 10-K filings, and add them up. That is TTM EPS.
- Divide price by TTM EPS for the trailing P/E.
- Look up consensus EPS for the next fiscal year the company has not reported yet. Note the fiscal-year end date, the analyst count, and whether the figure is GAAP or adjusted.
- Divide price by that consensus for the forward P/E. Repeat for the following year to see the trajectory.
- Scan the four trailing quarters for outliers. If one quarter is far above or below the others, read that quarter's release before trusting either ratio.
Rates matter too. A forward P/E of 25 means something different when Treasury yields are climbing; our piece on rising Treasury yields and equity valuations covers that angle. For daily market coverage, start at the What's On Finance homepage, or browse more analysis pieces.
Bull/Bear Verdict
Bull Case: Used together, the two ratios show the earnings trajectory the market is pricing. Nvidia's drop from 27.1 trailing to 13.6 on FY January 2028 consensus, as of the October 16, 2026 close, puts a number on the growth case instead of leaving it as a vibe.
Bear Case: Either ratio misleads when read alone. Trailing P/E swallows one-time gains, as Alphabet's and Amazon's Q2 2026 investment gains show, and forward P/E is only as good as a consensus that can be thin, stale or on a different fiscal calendar.
FAQ
Is forward P/E or trailing P/E better?
Neither is better in isolation. Trailing P/E is based on reported results, so it is verifiable but backward-looking. Forward P/E reflects expected growth but depends on analyst estimates. Comparing the two, and checking why they differ, is more informative than picking one.
Why is forward P/E usually lower than trailing P/E?
Because consensus usually expects earnings to grow; in the first table, all eight forward P/Es sit below their trailing P/Es. If next year's consensus EPS is higher than the last four quarters' EPS, dividing the same price by a bigger number produces a smaller ratio. When the forward P/E is higher, as with Alphabet and Amazon on 2027 estimates, consensus expects lower earnings ahead.
Does trailing P/E use GAAP or adjusted earnings?
The standard version uses GAAP diluted EPS from the last four reported quarters, which is what the tables here use. Some data providers substitute adjusted EPS, so check the source before comparing numbers across sites.
How often does forward P/E change?
Every trading day, through the price, and whenever analysts revise estimates. That is why every forward P/E here is tied to the October 16, 2026 close and the FMP analyst consensus at that time.