6 Stock Research Mistakes New Investors Make (and What Each One Costs)

Over the past ten years, only 28% of the companies in today's S&P 500 beat the S&P 500 itself. The typical one turned $10,000 into $28,530. An index fund turned it into $41,290. And that count flatters the stocks, because companies that fell out of the index aren't in it. Picking winners from that pool is hard enough. These six stock research mistakes make it harder, and each has a cost we could measure: not checking your pick against an index fund, confusing a great company with a great stock, trusting earnings without checking cash flow, ignoring dilution, copying Congress, and putting too much in one stock.

Watch it play outOnly 28 in 100 of today's S&P 500 companies beat the index

Each dot is 1% of the companies in today's S&P 500 with ten years of prices. Buy each one ten years ago and reinvest the dividends.

Now hold each one against SPY, an index fund that owns the whole S&P 500. Only 28% beat it.

The typical company turned $10,000 into $28,530. The index fund made $41,290 with no research at all.

And this is the flattering count. Companies that were dropped from the index, usually after falling, aren't in our data. Counting them would push 28% lower.

Replay

1. Not checking your pick against an index fund

Judging a stock by whether it went up is the wrong test. The right one is whether it beat what an index fund paid you for doing nothing, and most stocks fail it, even the ones good enough to be in the S&P 500 today.

We took every company in today's index and measured its total return, price plus reinvested dividends, against SPY over three windows. The result barely moves with the window: about 7 in 10 trailed.

About 7 in 10 of today's S&P 500 companies trailed the index

Past 3 years27%

Past 5 years29%

Past 10 years28%

Share of today's S&P 500 companies whose total return beat SPY, an S&P 500 index fund, over each window. Dividends reinvested in both.

Why so few? A handful of giant winners carry the index, and the typical stock never gets near them. Over ten years the median company returned 185%. SPY returned 313%.

One caveat applies to every price figure in this article. Our price file holds only the companies in the index today. Companies that were dropped, usually after falling, are missing, and companies that grew into the index are included, so the true share that beat the index is lower than 28%. Over very long windows the bias takes over: over 20 years the count rises to 43%, because a company had to survive two decades to be counted at all. Research that includes the dead companies points the same way. A study of nearly a century of US stocks found that most returned less than Treasury bills over their lifetimes, and the best 4% accounted for the market's entire net gain.

The fix: when you buy, write down the price of the stock and the price of an index fund on the same day. Judge the pick against that line after three to five years, dividends included. If most of your picks trail it, an index fund would have served you better.

2. Confusing a great company with a great stock

A dominant, growing business is not a good buy at any price. You can be right about the company and still lose years. At the top of the dot-com bubble, Cisco, Intel, Microsoft and Qualcomm were the companies everyone agreed would win. The businesses mostly delivered: by the time their stocks recovered, Cisco earned about seven times as much per share, Qualcomm about six times and Microsoft three times. Intel grew least, at 1.3 times. The stocks still took 12 to 21 years to climb back to the peak, even with dividends reinvested.

EPS is diluted earnings per share, net income divided by every share that could exist, from each company's annual report and adjusted for later stock splits. Falls and recoveries use monthly closes.

Their earnings grew, and the stocks still took 12 to 21 years to get back

Cisco (EPS up 6.9×)21.4 years

Intel (EPS up 1.3×)17.6 years

Microsoft (EPS up 3.0×)14.0 years

Qualcomm (EPS up 5.9×)12.0 years

Years from the dot-com peak until the stock, with dividends reinvested, closed a month above it again.

Cisco's numbers show how that happens. Its annual report for that year lists a highest closing price of $80.06, against earnings of $0.36 a share. That's a price-to-earnings ratio of 222: buyers were paying for 222 years of that year's profit. Profit per share then rose almost sevenfold. The multiple shrank by about as much, so the stock went nowhere for two decades. And these four are the survivors: companies from that era that later failed aren't in our price file at all.

The fix: before you buy a company you love, work out what the price is promising. Enter its EPS, its P/E, a growth rate you believe and a lower P/E ten years out into the P/E multiple compression calculator, and see what return is left. If the return only works when the P/E stays high, the price leaves no room for error. Our guide to calculating intrinsic value covers five fuller ways to value a stock, including a reverse DCF.

3. Trusting earnings without checking cash flow

Investors who read only the profit line and never open the cash flow statement miss the gap between what a company earns on paper and what it collects. Net income is an accounting figure built partly on estimates. Operating cash flow is the cash the business actually collected from customers, minus what it paid to run itself. In a healthy company the two move together over time.

Super Micro Computer, which builds servers for AI data centers, is a clear case. Over its last three fiscal years it reported $4.43 billion of net income. Over the same three years its operations used $7.64 billion of cash.

Super Micro's profit kept rising while its cash went the other way

Net incomeCash from operations

−7.5−5.0−2.5$2.5B0

$1.15B−$2.49B

Two years agoNet income$1.15BCash from operations−$2.49B

$1.05B$1.66B

Last yearNet income$1.05BCash from operations$1.66B

$2.23B−$6.81B

Latest yearNet income$2.23BCash from operations−$6.81B

Two years agoLast yearLatest year

Where did the cash go? Into inventory and unpaid customer bills. Over those three years, inventory and receivables (money customers owe) absorbed $17.4 billion of cash, far more than all the profit. In the latest annual report, sales rose 78% and profit more than doubled to $2.23 billion, while inventory nearly tripled to $12.9 billion. To fill the gap, Super Micro raised a net $15.4 billion from borrowing and selling stock over the three years.

The warning was public early. The quarterly report filed weeks before the stock peaked showed $453 million of profit for the half-year and $325 million of cash going out. Since its peak monthly close, the stock is down 59%, while SPY is up 50%. The fall had more than one cause. Later that year its auditor, Ernst & Young, resigned partway through the annual audit. But the cash gap was in the filings months before that, for anyone who compared the two lines.

Same company, same year, two different stories

Income statementWhat the headlines saw

• Sales up 78%
• Profit more than doubled, to $2.23B
• EPS $3.26, up from $1.68

Record year

vs

Cash flow statementWhat the cash did

• Operations used $6.81B of cash
• Inventory up $8.88B, receivables up $3.92B
• Net $9.48B raised from lenders and shareholders

Cash going out

That pattern is rare. Among the 285 profitable S&P 500 companies outside finance with three comparable years of data, operating cash flow beat net income over the past three years at 93% of them, and the typical one brought in $1.54 of operating cash for every $1 of profit (more on what's normal). Super Micro had the lowest ratio of all, and it was one of only two with a profit and negative operating cash flow. A gap can have an innocent cause, such as a one-time gain from selling a business, so a low ratio is a reason to check where the cash went, not proof of trouble.

The fix: put net income and operating cash flow side by side for the last three years. If profit keeps running ahead of cash, find where the cash went, usually inventory or receivables, and whether the gap is closing. Our list of 10-K red flags covers the other warning signs in an annual report.

4. Ignoring share dilution

The mistake is watching a company's total profit and forgetting that you own a slice of it. When a company issues new shares, to pay staff in stock, buy another company or raise cash, each existing share owns a smaller piece. That's dilution, and it's why earnings per share matter more to you than net income.

Most big companies do the opposite and buy back stock. Across 398 S&P 500 companies with five straight annual reports, the median diluted share count fell 3.9% over five years. But a third grew their share count, and 14% grew it by more than 10%.

One in seven S&P 500 companies grew its share count by more than 10%

Fell more than 10%27%

Fell 0 to 10%39%

Rose 0 to 10%20%

Rose more than 10%14%

Change in diluted weighted-average shares over each company's last five fiscal years. 398 companies.

Carnival shows why the share count matters even after the crisis is over. It sold shares to survive when the pandemic stopped its ships, and those shares kept it alive. Today its profit has nearly recovered, so a reader who stops at net income would think the business is back. Compare its latest annual report with the year before the pandemic:

Diluted EPS as reported. It assumes convertible notes turn into shares and adds back the interest they'd no longer cost, so it's slightly above net income divided by diluted shares.

Sales are 28% higher and profit is 92% of its old level. But there are twice as many shares, so each one earns less than half as much. With dividends counted, the stock is down 44% since the end of that pre-pandemic year, while SPY is up 168%. Super Micro, from mistake 3, grew its diluted share count by 25% in three years as it raised cash.

The fix: check the diluted share count on the income statement every year, right next to EPS. Growth of more than 10% over five years, about 2% a year, puts a company among the heaviest 14% of diluters in the S&P 500, so find out why: stock pay, acquisitions or capital raises. Then judge growth per share. If profit grows 10% a year and the share count grows 3%, your slice grows only about 7%. The dilution and buyback calculator shows whether a company's buybacks actually cover the shares it hands out.

5. Copying Congress or famous investors

Buying a stock because someone famous bought it doesn't work, at least in the most popular version we could test: copying members of Congress.

In our study, we took 4,967 stock purchases disclosed by members of Congress and "bought" each one the day after it became public. Of the 3,742 old enough to measure at six months, the typical copied trade had trailed the S&P 500 by 3.5 percentage points. Only 42% beat the index.

The longer you held a copied trade, the further it fell behind

30 days−0.6 pts

90 days−1.9 pts

180 days−3.5 pts

365 days−8.1 pts

Median return of a copied congressional stock purchase minus the S&P 500, bought the first trading day after the filing.

The filing delay doesn't explain it: buying on the day the member actually traded gave almost the same result, with the typical buy trailing by 3.6 points. Nancy Pelosi's record was no better at six months: of 17 spouse purchases we could price, 6 beat the S&P 500 after the filing date. Seventeen trades can't prove skill either way, and that's the trouble with copying a famous name: the record is too thin to trust.

Copying a famous fund manager has the same built-in problems. You see each trade late, you don't know why it was made, and you hear about the sale late too.

The fix: use a famous name's trade as a lead, not a reason. Read the company's filings and write your own one-sentence thesis. If you can't explain why you'd own it without the famous name, don't buy it.

6. Putting too much money in one stock

Even good research is wrong some of the time, and position size decides what being wrong costs. Big falls are normal even among the strongest companies: 3 in 4 of today's S&P 500 companies have lost half their value at some point. Letting one stock grow into a big share of your money turns an ordinary fall into a disaster.

Three in four of today's S&P 500 companies have fallen by half

75%of today's S&P 500 companies fell by half or more at some point

40%fell by 70% or more

1 in 4of the times a stock halved, it went on to halve again before recovering

Here the survivor-only data understates the risk. Every company that went bankrupt or was dropped after a collapse is missing, and each of those fell by more than half, so the real odds are worse. Of the times a stock in our data halved and we had ten years of prices afterward, 19% still hadn't regained its old high a decade later.

What a fall does to you depends on how much you own. A drawdown in one stock becomes a smaller one in your portfolio:

Assumes the rest of the portfolio holds its value. A loss always needs a bigger gain to recover: after losing a quarter, you need a third to get back.

The fix: decide the most one stock is allowed to cost you, then size from that. If a halving shouldn't cost more than 5% of your portfolio, no single stock gets more than 10%. The position size calculator does the math, and the concentration stress test shows what a fall in your biggest holding would do to the whole portfolio. When a winner grows past your limit, trim it back.

Our read

Paying any price for a great company costs the most, because its cost is counted in years: Cisco's buyers were right about the business and still waited two decades to break even. And because these feel like the safest stocks to own, they're the ones investors let grow into their biggest positions, which piles mistake 6 on top.

The easiest to fix is the index test. It's one extra line in a spreadsheet, and if you keep it honestly for a few years it catches the other five: when your picks keep trailing the index fund, something in your research isn't working.

Go deeper with member research

To run each of these checks properly on a real company, our member course, Fundamental research, from the ground up, teaches them in order across 40 chapters (12+ hours). It follows one example company the whole way: reading its filings, testing whether its earnings turn into cash, tracking its share count, valuing it and sizing the position. The chapter on testing the quality of earnings is the full version of mistake 3, and what the price is already assuming is the full version of mistake 2. The first lesson, Think like an owner, is free.

Keep reading

• Does copying Congress's stock trades work? The full study behind mistake 5.
• How to research stock picks: a seven-question checklist for doing the research right.
• Numbers to check before buying a stock, with what's normal across the S&P 500.
• Total return vs dividend yield: why the yield alone can mislead you.

FAQ

What percentage of stocks beat the S&P 500?

Over the past ten years, 28% of the companies in today's S&P 500 beat it, dividends included. The real share is lower, because companies that left the index aren't counted.

How much of my portfolio should be in one stock?

There's no single rule. Size each stock so that a 50% fall costs an amount you'd accept: at 10% of your portfolio a halving costs 5%, at 25% it costs 12.5%.

How we built this

• Prices: Yield Theory's monthly closes for current S&P 500 members from January 2000, adjusted for splits and dividends (Yahoo Finance chart API), as of October 19, 2026. Index-beating windows and every "since" figure end at the September 2026 month-end close. SPY is the benchmark. The file holds only companies in the index today, so every price-based result is survivorship-biased in the direction we describe in each section.
• Dot-com peaks: the highest monthly close in 2000 and 2001; "got back" is the first monthly close at or above it. Peak-year EPS comes from the original fiscal 2000 10-K filings (Cisco, Microsoft, Intel, Qualcomm), with Microsoft's $1.70 and Qualcomm's $0.85 halved for their later 2-for-1 splits. Later EPS comes from SEC XBRL data: Cisco fiscal 2021, Intel 2017, Microsoft fiscal 2013, Qualcomm fiscal 2011.
• Super Micro: fiscal years ended June 2024, 2025 and 2026 from its 10-K filings via SEC XBRL; the half-year figures from its 10-Q for the quarter ended December 2023. Its peak monthly close was March 2024.
• Carnival: fiscal 2019 (year ended November 2019) against fiscal 2025, from its 10-K filings via SEC XBRL.
• Share counts: diluted weighted-average shares from each company's 10-K filings via the SEC XBRL API. To stay clear of stock splits, each yearly change compares two numbers from the same filing, and five yearly changes are chained. 398 of the 460 companies in our SEC dataset had five straight years ending in fiscal 2025 or later.
• Cash conversion: net income and operating cash flow summed over each company's last three fiscal years from SEC XBRL, leaving out banks, insurers, REITs and other financial companies (SIC codes 6000 to 6799), whose operating cash flow isn't comparable. 285 companies had three comparable years and a positive total profit.
• Congress: every figure is from our congressional trades study, which covers purchases transacted from June 2024 to September 2026.
• Halvings: a halving starts the first month a stock closes 50% or more below its prior high; it ends when the stock closes back at that high. "Halved again" means it closed below half of that first halving price before recovering.
• What we tested and left out: "chasing last year's biggest winners" and "buying a stock because it fell a lot." In our data, last year's top 10% of performers beat the rest the following year in 11 of 24 years, a coin flip, and stocks bought after halving did well on average. Both results lean on companies that survived, which is exactly what this file over-represents. The same bias makes mistakes 1 and 6 look smaller than they are, so we kept those; it would flatter the two we left out, so we dropped them.