How to Find Undervalued Stocks, Step by Step
Julia Villaran
To find undervalued stocks, screen for companies trading at a low multiple of earnings or cash flow compared with their own industry, then estimate what the business is worth and buy only when the price sits well below that estimate. The steps below show how to value a stock with the main stock valuation methods, using real sector data and a fully worked Microsoft example taken from its annual report.
Most guides stop at "look for a low P/E." That is where the work starts, not where it ends. A cheap-looking ratio is only a lead. You still have to check that the earnings are real, that the cash is real, and that the price is not pricing in a decline you have missed.
Key takeaways
"Undervalued" only means something next to a benchmark. Compare a stock's ratios with its own industry and its own history, not with the whole market.
Use two kinds of stock valuation methods together: relative multiples (P/E, EV/EBITDA, free cash flow yield) to find candidates, and an absolute method (a discounted cash flow, or its reverse) to test them.
A reverse DCF asks what growth the current price already assumes. It is the fastest way to tell a bargain from a value trap.
Insist on a margin of safety and a written reason the market is wrong. Without both, a low price is just a low price.
What does "undervalued" actually mean?
A stock is undervalued when its market price is below its intrinsic value: what the business is worth based on the cash it will produce for its owners over its life. The CFA Institute puts it simply. If your value estimate is above the market price, you infer the stock is undervalued; below it, overvalued.
Intrinsic value is the present value of all the cash a business will hand to shareholders in the future. Present value means future dollars converted into today's dollars by discounting them at a rate that reflects risk and the time value of money.
Two things follow from that definition. First, intrinsic value is an estimate, never a fact, so two careful analysts can disagree. Second, a stock can look cheap on a ratio and still be fairly priced or expensive, because the market may be correctly expecting lower profits ahead. That is why the process below has a "find" stage and a separate "prove it" stage.
The main stock valuation methods, compared
Stock valuation methods fall into two families. Relative valuation compares a price multiple (such as price-to-earnings) with similar companies. Absolute valuation estimates intrinsic value from the company's own cash flows, with no reference to peers. Relative methods are fast and good for screening. Absolute methods are slower but answer the real question: what is this business worth?
| Method | What it measures | Best for | Weak spot |
|---|---|---|---|
| P/E ratio | Price ÷ earnings per share: dollars paid per $1 of profit | Profitable, stable companies; quick screens | Useless when earnings are negative; distorted by one-off gains and cyclical peaks |
| PEG ratio | P/E ÷ expected earnings growth rate (%) | Comparing growth companies with different growth rates | Depends entirely on a growth forecast that is often wrong |
| Price-to-book (P/B) | Price ÷ book value (assets minus liabilities) per share | Banks, insurers, asset-heavy firms | Misses intangible assets such as software, brands and patents |
| EV/EBITDA | Enterprise value (market cap + debt − cash) ÷ earnings before interest, tax, depreciation and amortization | Comparing companies with different debt levels | Ignores capital spending, which can be huge |
| Free cash flow yield | Free cash flow per share ÷ price | Seeing the real cash return on your purchase price | Lumpy year to year when investment spending swings |
| Dividend discount model | Present value of expected future dividends | Utilities, consumer staples, steady dividend payers | Does not work for companies that pay little or no dividend |
| Discounted cash flow (DCF) | Present value of all future free cash flows | Any business with forecastable cash flow | Very sensitive to the discount rate and long-term growth you assume |
| Reverse DCF | The growth rate the current price already implies | Testing whether a "cheap" stock is really cheap | Still needs a discount rate; tells you what is priced in, not what will happen |
We recommend pairing one relative multiple with a reverse DCF. The multiple finds the candidate. The reverse DCF tells you whether the price demands heroic growth or very little. No single model is enough, and the CFA Institute's guidance says the same: analysts use more than one model because each one rests on inputs that can be wrong.
How to find undervalued stocks in 6 steps
Step 1: Screen for candidates
Start with a free stock screener and a handful of filters: positive earnings over the last year, a P/E or EV/EBITDA below the industry median, positive free cash flow, and a market value large enough to trade easily. The goal is a short list of 10 to 30 names, not an answer.
Step 2: Compare each stock with its own industry
This is the step most investors skip, and it matters more than any other. Industries trade at very different multiples because they grow at different speeds and carry different risks. A P/E of 20 is expensive for a life insurer and cheap for a chipmaker.
How to find undervalued stocks: forward P/E by US industry, January 2026
The full US market traded at 27.66 times forward earnings in January 2026. Life insurers traded at 9.30 and biotech firms at 63.76. If you screened the whole market for "P/E under 15," you would mostly get insurers, banks and homebuilders, not undervalued companies.
Step 3: Compare the stock with its own history
A company trading well below its own 5- or 10-year average multiple deserves a closer look, but only if the business has not changed. The market as a whole swings widely too, which is why "cheap" in 2001 and "cheap" in 1974 meant very different things.
S&P 500 trailing P/E at year end, 1960–2025
The S&P 500 ended 2025 at 25.2 times trailing earnings (6,845.50 ÷ 271.52), well above its 1960–2025 median of 17.4. The low was 7.3 at the end of 1974; the high was 29.6 at the end of 2001, when earnings collapsed after the dot-com bust. When the whole market is expensive, a stock that is "cheap versus peers" can still be expensive in absolute terms.
Step 4: Check the quality of the earnings
Open the latest annual report (the 10-K, free on SEC EDGAR) and compare net income with free cash flow. Free cash flow is cash from operations minus capital expenditures, the money spent on buildings, equipment and data centers. Over several years, healthy companies turn most of their profit into free cash flow. If profit keeps rising while cash flow does not, find out why before you trust the P/E.
Step 5: Estimate value with a reverse DCF
Instead of guessing the future and calculating a value, take the price as given and solve for the growth it implies. If the implied growth is below what the business has already been delivering, the stock may be undervalued. If it is far above, the "cheap" multiple is a mirage.
Step 6: Demand a margin of safety and write your thesis
A margin of safety is the gap between your value estimate and the price you pay. Illustrative example: if you value a stock at $100 and want a 25% margin, you buy at $75 or less. Write down, in two or three sentences, why the market is wrong and what would prove you wrong. That note is what keeps you from selling at the bottom or holding a broken idea.
Ask "why is this cheap?" before you ask "how cheap is it?" Someone is selling you these shares. If you cannot name the worry behind the low price and explain why it is overdone, you do not yet have an edge.
How to value a stock: a worked Microsoft example
Here is the full process on a real company. On 8 April 2026, Microsoft (MSFT) traded at $375.38, the price at which TruthSayer AI's Valuations AI© model bought it. The latest annual report at the time was Microsoft's 10-K for the fiscal year ended 30 June 2025. Every figure below comes from that filing or is calculated from it.
- Find earnings per share. The 10-K reports diluted earnings per share of $13.64 for fiscal 2025.
- Calculate the P/E. $375.38 ÷ $13.64 = 27.5×. The inverse, the earnings yield, is $13.64 ÷ $375.38 = 3.6%.
- Compare with the industry. Damodaran's software (system & application) group traded at 37.52 times trailing earnings, counting only profitable firms. Microsoft's discount: (37.52 − 27.52) ÷ 37.52 = 26.7% below its peers. Against the whole market's 26.56, it was roughly in line.
- Calculate free cash flow. Operating cash flow $136,162 million − capital expenditures $64,551 million = $71,611 million. Divided by 7,465 million diluted shares = $9.59 per share.
- Calculate the free cash flow yield. $9.59 ÷ $375.38 = 2.6%. That is lower than the 3.6% earnings yield, so cash flow lagged reported profit.
- Find out why. Free cash flow was $59,475 million in fiscal 2023, $74,071 million in 2024 and $71,611 million in 2025, a 3.3% fall in the last year. Over the same year EPS rose 15.6% (from $11.80 to $13.64). The reason is in the cash flow statement: capital spending jumped 45.1%, from $44,477 million to $64,551 million. Profit was real; more of it was being reinvested in the business.
- Run a reverse DCF. Assume a 9% discount rate and 3% growth forever after year 10 (both are assumptions, not facts). Solving for the 10-year growth rate that makes the present value equal $375.38 gives 13.8% a year in free cash flow per share.
- Judge the implied growth. Microsoft's revenue grew 15.7% in fiscal 2024 and 14.9% in fiscal 2025. The price needed free cash flow to grow a little slower than revenue had recently grown. That is not a heroic assumption, which is what you want to see in an undervalued stock.
Here is the reverse DCF in full, so you can check every number. Free cash flow per share starts at $9.59 and grows 13.8% a year. Each year's cash is multiplied by a discount factor of 1 ÷ 1.09 raised to the year number.
| Year | FCF per share | Discount factor (9%) | Present value |
|---|---|---|---|
| 1 | $10.92 | 0.9174 | $10.02 |
| 2 | $12.42 | 0.8417 | $10.46 |
| 3 | $14.14 | 0.7722 | $10.92 |
| 4 | $16.09 | 0.7084 | $11.40 |
| 5 | $18.31 | 0.6499 | $11.90 |
| 6 | $20.84 | 0.5963 | $12.42 |
| 7 | $23.71 | 0.5470 | $12.97 |
| 8 | $26.98 | 0.5019 | $13.54 |
| 9 | $30.71 | 0.4604 | $14.14 |
| 10 | $34.94 | 0.4224 | $14.76 |
| Years 1–10 | $122.52 | ||
| Terminal value | $34.94 × 1.03 ÷ (0.09 − 0.03) = $599.87 | 0.4224 | $253.39 |
| Total value per share | $375.91 |
The total, $375.91, matches the $375.38 price to within the rounding of the growth rate to 13.8%. Note that 67% of the value ($253.39 ÷ $375.91) comes from the terminal value, the cash beyond year 10. That is normal for a DCF, and it is why small changes in assumptions move the answer so much.
Keep the same 13.8% growth but change the discount rate by one point and the value moves sharply: $463 per share at 8% (23% above the price) and $314 at 10% (16% below it). Any DCF that gives you a single precise number is hiding this range. Always test at least three discount rates.
What happened next? Valuations AI© closed the Microsoft position on 25 September 2026 at $517.51, a 37.82% gain after 170 days, as listed on the Valuations AI© page. One winning trade proves nothing on its own, which is why the full record below matters more.
What a cheap reading looks like, by sector
There is no universal "good" P/E. Use the sector's own level as the yardstick, and watch how many firms in the sector lose money, because a P/E screen silently drops all of them.
| Industry (Damodaran) | Forward P/E | PEG ratio | Money-losing firms | Better yardstick |
|---|---|---|---|---|
| Insurance (Life) | 9.30 | NA | 25.0% | Price-to-book |
| Banks (Regional) | 11.00 | 0.90 | 14.1% | Price-to-book |
| Homebuilding | 14.35 | 5.72 | 20.0% | P/E across a full housing cycle |
| Utility (General) | 18.13 | 2.96 | 0.0% | Dividend yield and DDM |
| Machinery | 24.06 | 1.97 | 42.9% | EV/EBITDA |
| Total Market | 27.66 | 1.90 | 57.2% | Compare within the industry first |
| Software (System & Application) | 34.13 | 1.65 | 70.2% | Free cash flow yield, reverse DCF |
| Semiconductor | 37.29 | 2.13 | 62.1% | EV/EBITDA across the chip cycle |
| Drugs (Biotechnology) | 63.76 | 0.84 | 89.5% | Pipeline value; P/E rarely applies |
Source for the numbers: Damodaran, NYU Stern, January 2026. The "better yardstick" column is our guidance. As a rule of thumb, treat a stock as a candidate when its multiple is clearly below its industry's level and its free cash flow is positive and stable. Treat it as a warning when the multiple is low but cash flow is falling, debt is rising, or earnings are at a cyclical peak.
Common mistakes when hunting for undervalued stocks
Comparing across industries. A bank at 11 times earnings is not "cheaper" than a software company at 34. They are different businesses with different growth and risk.
Trusting a P/E built on peak earnings. Homebuilders, chipmakers and miners look cheapest when profits are at the top of the cycle, just before profits fall. Use average earnings over a full cycle.
Ignoring the cash flow statement. Earnings are an accounting figure; cash is not. The Microsoft example shows a 15.6% EPS rise alongside a 3.3% fall in free cash flow. Know which one you are paying for.
Falling for a value trap. A value trap is a stock that looks cheap because the business is shrinking. The price keeps falling with the profits, and the ratio never improves.
Treating a DCF as precise. A one-point change in the discount rate moved Microsoft's value by $149 per share ($463 − $314). Use ranges, not single numbers.
A low P/E can be the market correctly pricing in bad news. If a stock has fallen 40% and trades at 6 times earnings, read the last two earnings call transcripts before buying. The answer to "why is it cheap?" is usually there.
How TruthSayer AI uses this
Our Valuations AI© model automates the steps above across US-listed stocks. It combines proprietary fair-value models with price action, institutional activity and fundamental financial data to flag stocks that look mispriced. Every alert is traded in our own funded account before it reaches users, and losing trades are published alongside winners.
As of 30 October 2026, the Valuations AI© page lists every trade, open and closed, including the losing ones. It shows a 73.5% historical win rate, a 68-day average hold and a 78.65% return on investment since inception. Our performance reporting follows the Global Investment Performance Standards, and our firm-wide policies were independently verified for September 2024 to June 2026; see our GIPS compliance page for the reports.
The rules matter as much as the model. We cap each position at 5% of the portfolio, use a 3× ATR (average true range) trailing stop on non-core trades, and average down on core trades only when the price falls 2–3 standard deviations, as described in our hedge fund strategy for retail investors. If you are weighing signal services, our guide to AI stock signals and how to judge them explains what to ask for, and our comparison of the best AI stock apps covers the wider field. For how value investing fits next to other approaches, see our investment strategies explained.
The bottom line: Finding undervalued stocks is a two-stage job. Screen with a multiple compared against the stock's own industry, then prove the case with free cash flow and a reverse DCF that shows what the price already assumes. Buy only with a margin of safety and a written reason the market is wrong. If you would rather see that process run systematically, with every trade and every loss on the record, look at Valuations AI©.
Frequently asked questions
How do you know if a stock is undervalued?
Compare its price with an estimate of intrinsic value. In practice, check whether its P/E, EV/EBITDA or free cash flow yield is clearly better than its industry's, then run a reverse DCF. If the growth the price implies is below what the company has been delivering, and the balance sheet is sound, the stock may be undervalued.
What is the best ratio to find undervalued stocks?
No single ratio works everywhere. P/E is the most common for profitable companies, price-to-book suits banks and insurers, and EV/EBITDA is better when debt levels differ. Free cash flow yield is the hardest to fake. Use one ratio to screen and a cash-flow check to confirm, always against the same industry.
Is a low P/E ratio always a good sign?
No. A low P/E can mean the market expects profits to fall, that earnings are at a cyclical peak, or that a one-off gain inflated the "E." It is also meaningless for the 57% of US-listed firms that lost money over the trailing year. Treat a low P/E as a reason to look, not a reason to buy.
What are the main stock valuation methods?
There are two families. Relative methods compare multiples such as P/E, P/B, PEG and EV/EBITDA with similar companies. Absolute methods estimate intrinsic value from cash flows: the discounted cash flow model, the dividend discount model and asset-based valuation. A reverse DCF uses the absolute approach to show what growth today's price implies.
How long does it take for an undervalued stock to go up?
There is no set time, and some never recover. The market may need a catalyst such as an earnings report, a buyback or new management. Holding periods for value strategies are usually weeks to years. Our Valuations AI© trades show a 68-day average hold, but individual positions have run far longer.
Where can I find undervalued stocks for free?
Free stock screeners from most brokers let you filter by P/E, P/B and free cash flow. Annual reports are free on SEC EDGAR, and Aswath Damodaran publishes free industry multiples each January. Combine a screen with your own check of the 10-K.
Sources
- Aswath Damodaran, PE Ratio by Sector (US), NYU Stern, data as of January 2026
- Aswath Damodaran, Historical Implied Equity Risk Premiums (S&P 500 level and earnings), NYU Stern, January 2026
- Microsoft Corporation, Form 10-K for the fiscal year ended 30 June 2025 (filed 30 July 2025), SEC EDGAR
- SEC EDGAR XBRL company facts for Microsoft (EPS, operating cash flow, capital expenditures, diluted shares, revenue)
- CFA Institute, Equity Valuation: Concepts and Basic Tools, 2026 refresher reading
- Investopedia, Stock Valuation Techniques: A Guide to DDM, DCF, and Comparables
- TruthSayer AI, Valuations AI© performance and trade list, viewed 30 October 2026
- TruthSayer AI, GIPS compliance and independent verification reports
This article is for educational purposes only and is not investment advice. Past performance does not guarantee future results.
Julia Villaran
With a strong foundation in business development and consultancy, Julia has a keen analytical approach to risk-managed portfolio growth. Her strategic insight and market expertise make her a trusted advisor for individuals and organizations navigating the complexities of today's investment landscape. She combines deep market understanding with a research-driven approach to help uncover high-potential investment opportunities and build resilient financial strategies.
