Dan Benton's 1991 Tech Rules: What Held Up and What Didn't
Dan Benton's 20 rules for technology investing, written at Goldman Sachs around 1991, still circulate. In 2026 he rewrote them. The rules about earnings expectations held up; the summer-seasonality rule fit the six summers he had seen and then faded, as 40 years of Nasdaq-100 data show.
ON THIS PAGE ▾
Some time around 1991, a technology analyst at Goldman Sachs named Dan Benton wrote down twenty rules for investing in technology stocks. They fitted on a single page. Thirty-five years later they still circulate as photocopies on old Goldman letterhead, and Benton has said they end up pinned to cubicle walls. In March 2026 he finally published a revised edition.
I find the pair more instructive than either list alone. A rule that survives thirty-five years of change is telling you something about how markets work. A rule its own author drops is telling you something about how rules get made. And a few of Benton's rules can be tested against the data that has accumulated since he wrote them, which is what I've tried to do here.
What the 1991 rules said
The original page grouped the twenty rules under six headings: momentum, valuation, seasonality, management, "old world/new world", and a one-line summary. Three ideas carry most of the weight:
- Expectations drive the stock. Rule 1, in capitals on the original page, says to sell technology stocks when estimates are being reduced. Rule 2 says to buy only for positive earnings surprises, and Rule 3 describes what produces them: accelerating growth, rising selling prices and widening margins.
- Cheapness is a trap. "Value investors don't make money in technology," reads Rule 7, and Rule 8 warns against buying on relative valuation, especially when estimates are falling.
- The calendar matters. Rule 9: "Technology stocks perform poorly in the summer." Seasonal slowdowns, the next rules add, tend to be mistaken for something worse.
The page ends with Rule 20: "Don't forget Rule 1." By CNBC's account, Benton went on to manage technology hedge funds whose returns from 1994 to 2001 averaged 53% a year, against 13.3% for the S&P 500, a figure CNBC attributes to The Wall Street Journal. Whatever one makes of that record, the rules were written by someone who acted on them.
What he kept: estimates move stocks
The rule Benton never let go of is the first one. His 2026 edition puts it this way: "Technology stocks move in the direction of estimates. Forecast at least two years out." He describes how well the old framework held up in one sentence: expectations still drive stocks, acceleration still creates operating leverage, and value investing still doesn't work.
This is the rule I'd expect to last, because it rests on a mechanism rather than a pattern. A stock's price embeds a forecast of future earnings. When that forecast is revised, the price must move, and analysts and investors tend to revise in steps rather than all at once, so revisions come in runs. The tendency of prices to keep drifting after an earnings surprise has been studied in academic finance since the late 1960s and is one of the most persistent patterns in the literature. Benton wrote it down as a practitioner's rule.
Second-level thinking adds the qualification Benton himself added in 2026. Knowing that estimates matter is not an edge; everyone knows it now. His new Rule 8 is the sharper version: "Invest when your estimates are out of consensus and correct." Elsewhere he puts it more bluntly: "The information edge is now zero." The durable part of Rule 1 isn't the observation. It's the discipline of asking what the consensus already expects, and why yours differs.
What he dropped: the summer rule
Rule 9 is the one that can be tested directly, and it is also the one Benton abandoned. His 2026 list says: "Seasonality is a distraction. Forget 'sell in May and go away.'" I wanted to know whether the rule was wrong when he wrote it, or right then and wrong since. So I looked at the Nasdaq-100, the longest-running index weighted towards technology, month by month from 1986 to October 2026, and compared June, July and August with the other nine months.
The summer effect Benton saw, and what came after
Average monthly Nasdaq-100 price return, summer (Jun–Aug) versus other months
- 1986–1991, summer months−0.43%
- 1986–1991, other months+2.21%
- 1992–2026, summer months+1.21%
- 1992–2026, other months+1.32%
| 1986–1991, summer months | −0.43% |
|---|---|
| 1986–1991, other months | +2.21% |
| 1992–2026, summer months | +1.21% |
| 1992–2026, other months | +1.32% |
Price index, excludes dividends. 1986–1991 is the period Benton could have seen; 1992 onwards came after the rules. Author's calculations.
Source: Nasdaq-100 (^NDX) monthly closes, Yahoo Finance, through October 2026.
Table 1: Rule 9 before and after it was written
| Period | Summer months, average | Other months, average | Summer minus S&P 500, average | Summers the Nasdaq-100 lagged the S&P 500 |
|---|---|---|---|---|
| 1986–1991 (6 summers) | −0.43% | +2.21% | −1.32% | 5 of 6 |
| 1992–2026 (35 summers) | +1.21% | +1.32% | +0.73% | 10 of 35 |
| 1986–2026 (41 summers) | +0.97% | +1.45% | +0.43% | 15 of 41 |
Source: author's calculations from Nasdaq-100 (^NDX) and S&P 500 (^GSPC) monthly closes, Yahoo Finance, January 1986 to October 2026. Monthly price returns; summer is June to August.
How often tech lagged the broad market over the summer
5 of 6
summers, 1986–1991
the data behind the rule
10 of 35
summers, 1992–2026
after the rule was written
Source: Nasdaq-100 versus S&P 500, June–August price returns; author's calculations from Yahoo Finance data.
The answer is both. In the years Benton could see, the pattern was real and large: tech averaged a loss in summer months and a healthy gain in the rest, and it trailed the broad market in five of six summers. After 1992 the effect disappeared. Summer months have averaged almost exactly what other months have, and the Nasdaq-100 has actually done slightly better than the S&P 500 in summer, not worse.
I don't think the lesson is that Benton was careless. Six summers is what he had. The lesson is about how rules get made. A pattern drawn from a handful of observations can be genuine for that period and still have no reason to persist, because nothing in the mechanism of a business or a market requires July to be worse than October. Compare that with Rule 1, which rests on how prices absorb new information. One kind of rule describes a cause; the other describes a coincidence that hasn't been disproved yet.
What changed meaning
A few rules didn't fail so much as get overtaken by changes in the industry.
- Stock options. Rule 15 said insider selling doesn't matter because management receives new stock options every year. The 2026 edition keeps "insider selling doesn't matter", but adds a new rule: "Stock-based compensation is an expense." The same annual grants that once explained away insider sales are now treated as a real cost to shareholders. Benton adds that buybacks which only offset that dilution create no value.
- Sector to market. In 1991 technology was one sector among many. Benton's 2026 Rule 3 reads: "Technology is now the market." That is not a figure of speech. As our recent look at Howard Marks's memo on AI noted, 42 AI-related companies produced about two-thirds of S&P 500 price gains since January 2024, by J.P. Morgan's count. A rule written for a sector now governs most of an index fund.
- Old world, new world. Rule 16 called traditional mainframe and minicomputer companies a secular decline, and history agreed. The more durable point sits in Rules 18 and 19: when products become standardised, execution is what separates companies, and execution is hard to forecast. That reads as well about today's commoditising AI models as it did about personal computers.
Where this could be wrong
- The Nasdaq-100 isn't a pure technology index. It is weighted towards technology but holds other large non-financial companies. A narrower tech index might show a different summer pattern, though its history would be shorter.
- Small samples cut both ways. Six summers were too few to establish Rule 9, and averages can be moved by a few extreme months. Medians are a little less kind to summer: since 1992 the typical June–August month has returned about 1.4% against about 1.9% for the other months, a gap far smaller than in 1986–1991. The honest conclusion is that there is no reliable summer effect, not that summers are now good for tech.
- Price only. Yahoo Finance series exclude dividends. That matters little for a month-by-month comparison, but the absolute averages understate total returns.
- The past isn't a promise. A pattern can reappear for reasons no one anticipates. Testing Rule 9 says the calendar has not been a dependable signal since 1992; it can't say it never will be.
- Benton's record is his. The returns quoted are second-hand, from CNBC citing The Wall Street Journal, and belong to a particular era, a particular manager and the rules as he applied them.
Closing thoughts
Read side by side, the two lists make a quiet argument about what kind of knowledge lasts. The rules that held up describe why prices move: expectations, revisions, the leverage that accelerating revenue gives to margins. The rule that faded described when prices had moved, over a period too short to know whether it meant anything. Benton's own summary of thirty-five years is worth keeping: the market is faster than it was, but it isn't smarter. If that's right, the advantage still belongs to whoever understands the mechanism, and the calendar was never one.
This article is educational and is not personalized financial advice. Investing involves risk of loss, and past performance does not predict future results. The rules discussed are Dan Benton's, quoted from his published lists; the analysis and data tests are the author's own. Index data is not Shipra results. Data as of 31 October 2026.
Related questions
What are Dan Benton's rules for technology investing?
They are twenty rules Dan Benton wrote as a technology analyst at Goldman Sachs around 1991, grouped under momentum, valuation, seasonality, management and "old world/new world". The best known is Rule 1: sell technology stocks when estimates are being reduced. He published a revised 2026 edition that keeps the focus on earnings estimates and drops seasonality.
Do technology stocks perform poorly in the summer?
They did in the late 1980s: from 1986 to 1991 the Nasdaq-100 averaged −0.43% a month in June to August against +2.21% in other months. Since 1992 the difference has disappeared, with summer months averaging +1.21% against +1.32%. Benton himself dropped the rule in 2026, calling seasonality a distraction.
Why do earnings estimates matter so much for technology stocks?
A stock's price reflects expected future earnings, so revisions to those expectations move it. Revisions also tend to come in runs, because analysts adjust gradually. Benton's 1991 Rule 1 and his 2026 Rule 7 both rest on this, and the drift of prices after earnings surprises has been studied in academic finance for decades.
Sources & references
- Dan Benton's Rules for Technology Stock Investing, 2026 Edition (4 March 2026), Dan Benton, Substack
- Pro Strategy: 20 investing rules from hedge fund legend (29 March 2016), CNBC
- Nasdaq-100 (^NDX): historical monthly data, Yahoo Finance
- S&P 500 (^GSPC): historical monthly data, Yahoo Finance
The materials on this website are for illustration and discussion purposes only and do not constitute an offering. An offering may be made only by delivery of a confidential offering memorandum to appropriate investors. PAST PERFORMANCE IS NO GUARANTEE OF FUTURE RESULTS. Disclosures
Investing in the US, Canada, India, China and Japan involves currency, emerging-market, liquidity and regulatory risks. International and emerging-market risk
More in Factor Research
View all →Compounding Builds Roots Before It Shows
Twenty years of the Nifty 50 turned ₹1 lakh into about ₹6.7 lakh, yet less than a quarter of the gain came in the first decade. That is the arithmetic of compounding, not bad luck. It needs nothing complex, only a long time without interruption, and the same shape shows up well beyond money.
Howard Marks on AI: 11 Questions Before You Invest
Howard Marks's memo "A Layman Thinks About AI" makes no forecast. It asks eleven questions instead. The odds math says a six-link AI thesis rarely holds in full, concentration and credit exposure are already high, and price, not conviction, is the protection.
Rates Are Rising Again: Where Credit Still Offers Value
The Fed hiked to 3.75%–4.00% and the 10-year briefly topped 5%, for different reasons at each end of the curve. Short-duration credit lets you collect the higher yield without forecasting the hard part. Here is how higher rates reshaped high yield, loans, RMBS and CLOs, and where value remains.