William O’Neil: The 8% Rule, and the Fund That Had to Live By It
In March 1997 a reporter telephoned William O’Neil to ask whether shutting his mutual fund meant momentum investing was finished. “That is dumb,” O’Neil growled down the line. He was 63, and he had just handed the assets of the only fund that had ever let outsiders measure his famous system to somebody else. Over the life of that fund it had returned 90.0%. The average growth fund had returned 98.4%.
KEY TAKEAWAYS
- The rule is sound and incomplete. 8% caps one loss, not a run of them. Eight trades in a row each stopped out at 8% compound to −48.68%, a hole needing +94.85% to climb out of.
- It silently sets a position cap nobody quotes. If a stopped-out trade may cost 2% of your account, an 8% stop means the position can be at most 25% of the account. At 1% risk, 12.5%.
- His founding number does not reconcile with itself. Bloomberg’s obituary reports both “a 20-fold increase” and Schwager’s $5,000 to $200,000 — which is forty-fold — two sentences apart.
- The only public record of the system in his own hands was mediocre. New USA Growth returned 90.0% over its life against 98.4% for all growth funds: rank 162 of 276 (Lipper, March 1997).
- He named his own mistake out loud. On the fund he lost in the late 1960s: “we bought too many small names… We were growing too fast.”
Who was William O’Neil?
Born 25 March 1933 in Oklahoma City, raised in Texas, a business degree from Southern Methodist University in 1955, then the Air Force. In 1958 he started as a stockbroker in the Los Angeles office of Hayden, Stone & Co., and there he began doing something almost nobody else was doing: feeding price and earnings history into a computer to ask what the biggest winning stocks had in common before they ran.
“I believe in the next three to five years the computer will be the big thing in the securities market. You go through fads in this business, and the computer is the next one.”William O’Neil, quoted by Martin Mayer in New Breed on Wall Street (1969).
He said it from California, not New York, and he was early by about two decades rather than three to five years.
He founded William O’Neil + Co. in 1963, bought a seat on the New York Stock Exchange in 1964 at the age of 30 — the youngest to do so at that time — and stayed in California anyway. In 1984 he launched Investor’s Daily, renamed Investor’s Business Daily in 1991, to put his database in print. He died on 28 May 2023, aged 90.
The seven-part screen he built from that research he called CAN SLIM. But the thing that outlived the acronym, and the reason he belongs in this section rather than a stock-picking one, is a single number.
What is the 7–8% rule?
Never let a loss on a position exceed 7–8% below the price you paid. The level is fixed at the moment you buy. If it is reached, the position closes: no averaging down, no waiting for the bounce, no moving the level. He set it out in How to Make Money in Stocks, first published in 1988, and repeated it for the rest of his life.
Where the number came from matters more than the number. It was not chosen for elegance; it came out of his database work on what the largest winning stocks did on the way up. In his framing, a stock doing the thing you bought it for does not normally hand you an 8% loss first. So 8% is not a pain threshold but a falsification threshold: past it, the reason you bought is no longer true. That is the structure taught in avoiding the big loss, reached from the stock’s side rather than the account’s — you decide what evidence ends the trade while you still have nothing at stake.
Does the arithmetic hold?
On its own terms, yes — and the check is worth doing, because the second half of it is the half that gets left out.
Losses are asymmetric. Recovering an 8% loss takes an 8.70% gain. Recovering 20% takes 25%. Recovering 50% takes 100%. Capping the loss at 8% keeps you inside the region where the repair is roughly the same size as the damage, and that is a real and underrated property.
But the rule caps one loss, and markets hand out losses in runs.
Read the bottom pair. Eight consecutive stop-outs at 8% of the position leaves you down 48.68% and needing to almost double what is left; the same eight-loss run sized so each costs 2% of the account leaves you down 14.92% and needing 17.54% — a bad quarter rather than a career. So O’Neil’s rule and the one to two percent rule are not rivals. His is a rule about the stock; the other is a rule about the account. Obey only the first and an entirely ordinary losing streak can still finish you, without your breaking a rule you know about.
The number the rule never states
The two rules join at one division, and it is the most useful line in this article. If a stopped-out trade may cost at most 2% of the account and the stop sits 8% below entry, then the position may be at most 2 ÷ 8 = 25% of the account. At 1% account risk it is 12.5%. Below those sizes the 8% rule protects you; above them it is decoration.
That is why we teach size before strategy: an 8% stop means nothing on its own, because 8% of what is the whole question. Run the division the other way and it hands you the size directly — allowed account risk divided by stop distance. That is exactly what position sizing does, and what the position size calculator does for you.
PRACTICE CORNER
Two measurements, both doable in one sitting, and both taken directly from the two numbers above. First, the size cap: divide the loss you will accept as a share of your account by your stop distance in percent. If you accept 2% and your stop is 8% away, no position exceeds a quarter of the account — write that ceiling down before your next entry. Second, the sanity check on the number 8 itself: download the last 90 daily candles for the pair you actually trade — every major exchange lets you export candle history — and work out the average daily high-to-low range as a percentage. Divide 8 by it. That tells you how many ordinary days of noise your stop sits away from your entry. If the answer is under two, an 8% stop is not a stop on that instrument; it is a coin toss, and you need the structural approach in stop-loss placement instead.
Referral links — they never change our assessment. Education only; most retail traders lose money.
Legend against record
The founding story is that in the early 1960s O’Neil turned a small stake into a large one and built his firm with the proceeds. True in outline, almost certainly. But the numbers attached to it do not agree with each other, and a single obituary carries two of them.
Bloomberg’s obituary in May 2023 says he founded William O’Neil + Co. in 1963 “after recording a 20-fold increase in his own account”. Two sentences later, the same obituary reports Jack Schwager’s account in Market Wizards (1989) that O’Neil “parlayed a $5,000 investment into $200,000 through three consecutive bets” — short the discount chain E.J. Korvette, long Chrysler, long Syntex. Five thousand to two hundred thousand is forty-fold, not twenty. Other accounts put the same result over one year, others over two.
Nobody here is being accused of dishonesty. The point applies to every trading legend you will read about: the multiple is not a checkable fact, it is a figure that arrived through retelling. The reporter who covered his fund’s closure in 1997 put the underlying problem plainly — because O’Neil’s own finances were private, nobody had ever been able to track his results.
Two things are checkable, and both are in his favour. He twice bought full-page advertisements in The Wall Street Journal to announce a bull market was starting: March 1978, ahead of a six-month rally, and February 1982, months ahead of one of the longest bull markets in American history. Dated, public, and right.
What his own funds did
He ran public money twice, and that record is the closest thing we have to a measurement of his system in his own hands.
The first fund, the O’Neil Fund, launched in the mid-1960s — the Wall Street Journal’s obituary says 1965, Bloomberg’s says 1966, and we print both because we could not settle it. Within two years it held about $10 million. In 1967 it rose 116%, making it that year’s best-performing mutual fund by FundScope’s measure. Then 1968–69 took it apart, and when the market came back the fund did not. He sold it in 1975 with $6 million in assets, down from a $49 million peak — an 87.8% fall. His own post-mortem went to the Los Angeles Times.
“We were buying phenomenal small companies that no one had ever seen before, but we bought too many small names. … We were growing too fast.”William O’Neil on the O’Neil Fund, to the Los Angeles Times, quoted in his Wall Street Journal obituary, 30 May 2023.
That is a liquidity confession, not a stop-loss confession. An 8% rule assumes you can get out at 8%. In small, thinly traded names held in size, the exit is not where you left it.
The second fund was New USA Growth, opened in 1992. On the strength of his following it took in $170 million inside a month. In March 1997 he transferred its assets to another manager. These were the numbers on the way out.
Rank matters more than return here. Over the fund’s life, 162 of 276 is the 59th percentile — below the median of funds doing the same job — and annualised over five years that is 13.70% a year against 14.69%, a percentage point behind for five years. The year in progress was 772 of 786: the bottom 1.8%. Morningstar’s analyst noted 400% annual turnover, meaning an average holding period near three months, an expense ratio nearly a point above the group, and a fund “even riskier than its average peer”; Morningstar called it “a study in extreme investing”.
Be fair about what that proves. A public fund is not the same test as a private account: redemptions arrive on other people’s schedule, there are fees, cash drag and a stated mandate, and this one was run day to day by a protege rather than by O’Neil at the screen. None of it shows the 8% rule fails.
What it does show is smaller and still worth knowing. The only measurable record of this system in the hands of the man who wrote it did not beat the average fund doing roughly the same thing — and a book of positions each cut at 8% still finished a year down 4% while peers made 5%. A stop rule limits how badly one position can hurt you. It does not make a strategy profitable, and it was never advertised as doing so.
What this rule costs you
It costs you winners that shake first. A fixed 8% stop will remove you from positions that would eventually have worked. That is not a flaw to be engineered away; it is the price of the cap, paid in every account that uses one.
It costs re-entry discipline most people do not have. O’Neil’s answer was that you can always buy the stock back. In practice, traders stopped out at 8% either stand aside for the whole subsequent move or climb back in badly — the behaviour examined in revenge trading. Cheap to state, expensive to live with.
It costs turnover, and turnover costs money. Four hundred percent a year at his own fund is what this rule looks like followed literally: fees, spreads and slippage on every exit and re-entry, plus funding on any perpetual you hold through it.
Where it stops working
Eight is not a constant of nature. It was calibrated on US growth stocks in a particular era. On an instrument whose ordinary daily range is much larger it stops being a decision and becomes a coin toss. The relation: if your instrument’s typical daily high-to-low range is R percent, an 8% stop sits 8 ÷ R ordinary days of noise from entry. At R = 2 that is four days; at R = 5 it is 1.6. We have not printed a figure for R on any crypto pair, because we have not measured it — measuring your own is the second half of the Practice Corner.
A percentage stop ignores structure. The market has never heard of your entry price. A round 8% can land just inside an obvious level where a great many other stops are also resting. Putting the stop where the idea is actually wrong, then sizing to that distance, is the method in stop-loss placement.
Leverage changes what the number means. An 8% adverse move is 8% of the position; on ten times leverage it is 80% of the margin behind it. Carrying an unleveraged-equity rule unmodified into a leveraged account is a fast way to be liquidated while believing you are being careful — see leverage and margin.
It says nothing about selling a winner. O’Neil published many sell rules and most are about taking profits. This is only the loss half, and the loss half alone does not produce a positive expectancy — see risk to reward.
Two funds is not a study. The record above rests on one contemporaneous column and two obituaries: the best public evidence we found, and still a small, indirect sample.
FAQ
Who was William O’Neil?
William J. O’Neil (25 March 1933 – 28 May 2023) was an American stockbroker who founded the research firm William O’Neil + Co. in 1963, bought a seat on the New York Stock Exchange in 1964 at the age of 30, and founded the newspaper Investor’s Business Daily in 1984. He created the CAN SLIM stock-selection method and wrote How to Make Money in Stocks, first published in 1988.
What is O’Neil’s 7–8% sell rule?
Never let a loss on a position exceed 7–8% below the price you paid. The exit level is set at the moment you buy, and if it is reached the position is closed — no averaging down and no waiting for a bounce. He set it out in How to Make Money in Stocks (1988) and repeated it for the rest of his career.
Does the 8% rule protect your account?
It protects one position, not the account. Recovering an 8% loss needs an 8.70% gain, which is manageable. But eight trades in a row each stopped out at 8% compound to a 48.68% drawdown, which needs a 94.85% gain to recover. The 8% rule is a rule about the stock; a rule about the account, such as risking 1–2% of it per trade, is a separate thing you also need.
What position size does an 8% stop imply?
Divide the account risk you accept by the stop distance. If a stopped-out trade may cost at most 2% of your account and the stop sits 8% below entry, the position can be no more than 2 divided by 8, or 25% of the account. At 1% account risk it is 12.5%. Almost no summary of the rule mentions this number, and without it the rule does not limit anything.
How did O’Neil’s own funds perform?
He ran two public funds and neither is a strong advertisement. The O’Neil Fund rose 116% in 1967 to be that year’s best-performing fund by FundScope’s measure, then fell apart in 1968–69; he sold it in 1975 with $6 million in assets, down from a $49 million peak. New USA Growth, opened in 1992, returned 90.0% over its life against 98.4% for all growth funds, ranking 162 of 276 by Lipper’s count in March 1997, when he moved its assets to another manager.