What the number actually says#
EPS = (net income − preferred dividends) ÷ shares outstanding. It answers how much profit each common share earns. Preferred shareholders get paid first at a fixed rate, so their dividends come out before the division.
The comparison it enables is genuinely useful:
| Company A | Company B | |
|---|---|---|
| Net income | $5M | $8M |
| Preferred dividends | $200k | $300k |
| Shares outstanding | 1M | 2M |
| EPS | $4.80 | $3.85 |
Company B earns 60% more in absolute profit and delivers less to each share. On this measure A is the more capital-efficient business. That’s a real insight that raw net income can’t give you — bigger isn’t the same as better per unit of ownership.
It also isn’t sufficient to pick between them. Revenue growth, debt levels and cash flow all sit outside the formula.
Three things move EPS, and only one is earning money#
Profit rises. Higher sales, lower costs, better efficiency. This is the version everyone assumes they’re reading.
Share count rises. Issue more shares and the same profit spreads thinner. Dilution is the cake cut into more slices — every existing holder’s claim shrinks. Stock-based compensation does this quietly and continuously at most technology companies.
Share count falls. Buy back shares and each remaining one claims a bigger slice. Nothing about the business changed.
That third one deserves your attention, because it has become enormous.
The scale of the buyback effect#
S&P 500 buybacks passed $1 trillion over the twelve months ending September 2025 — a record $1.020 trillion, up from $918.4 billion the prior year, and they’re running above $1 trillion annualised into 2026.
The mechanical effect is easy to see. A company earning $10 billion with 1 billion shares reports $10 EPS. Buy back 50 million shares with earnings unchanged and EPS becomes $10.53 — 5.3% “growth” produced entirely by share count reduction.
Buybacks are also concentrated: the top 20 S&P 500 companies accounted for 49.5% of Q3 2025 repurchases, above the pre-COVID average of 44.5%. So the EPS growth of the largest index constituents — the ones driving most of the index’s reported earnings growth — carries the largest share-count effect.
To be fair to buybacks: returning capital to shareholders when a company has no better use for the money is legitimate, and it’s often more tax-efficient than a dividend. The problem isn’t the practice. It’s reading the resulting EPS figure as evidence the business improved.
How to tell the difference in thirty seconds#
Pull two numbers instead of one.
Look at net income growth alongside EPS growth. If EPS grew 12% and net income grew 12%, the business earned it. If EPS grew 12% and net income grew 3%, roughly three-quarters of the improvement came from the denominator.
Then check shares outstanding across several years. A steadily falling count means buybacks are a permanent feature of the EPS trend. A rising count means dilution is quietly working against you, and the reported EPS growth understates the operating performance.
Neither pattern is automatically good or bad. Both change what the headline number means.
Negative EPS isn’t automatically a red flag#
A net loss produces negative EPS, and plenty of growth companies run negative for years while investing heavily before profitability arrives.
The question is what’s producing the loss. Heavy investment in growth with strong revenue expansion and a solid cash position is a different situation from a shrinking business burning through reserves. Same negative EPS, opposite outlook. Revenue growth, cash runway and the reason for the losses do the work the ratio can’t.
Where EPS feeds forward#
EPS is the denominator of the P/E ratio: stock price ÷ EPS. A higher P/E means investors are paying more per dollar of earnings, usually because they expect growth.
Which means any distortion in EPS propagates straight into P/E. A company aggressively buying back shares shows rising EPS and, at a constant price, a falling P/E — looking cheaper without becoming cheaper.
The other limitations are worth listing plainly. EPS ignores debt entirely, so a heavily borrowed company and a debt-free one can post identical figures. Companies with few shares outstanding can look more profitable without being stronger. Accounting choices can shift it. And it’s only meaningful within an industry — comparing a bank’s EPS to a software company’s tells you nothing.
Summary — and what to do about it#
EPS is a ratio, and one side of it is under management’s direct control.
- Always read EPS growth next to net income growth. The gap between them is the share-count effect.
- Check shares outstanding over five years. Falling means buybacks; rising means dilution.
- Treat buybacks as capital allocation, not performance. Sometimes a good use of cash, never evidence of a better business.
- Don’t dismiss negative EPS. Ask what’s causing the loss and whether revenue and cash support it.
- Compare within an industry only. Across sectors the number is meaningless.
- Never use it alone. It ignores debt and growth prospects entirely — pair it with revenue growth, debt levels and cash flow.
A trillion dollars a year is being spent making this number look better. Read the denominator.
Sources & further reading#
- S&P Dow Jones Indices via PR Newswire, Q3 2025 buybacks — the record $1.020 trillion twelve-month total and top-20 concentration.
- Buxton Helmsley, “The Buyback Mirage” — the worked mechanics of buyback-driven EPS growth.
- Wall Street Horizon on 2025 buyback concentration — how few firms account for most repurchases.
