While investors debate artificial intelligence valuations and semiconductor supply chains, a quiet corner of the real estate market is doing what dividend investors love most: writing reliable checks. EPR Properties (EPR), a net lease real estate investment trust focused on experiential venues — movie theaters, ski resorts, regional theme parks, and fitness facilities — is up 18% in 2026 and paying a dividend yield of approximately 6.1%, a rate JPMorgan’s equity research team recently called “safe and growing.”

What makes EPR worth studying is not just the yield — it’s the structural story behind it. In 2026, EPR completed the acquisition of seven regional parks from Six Flags Entertainment, adding durable physical assets to a portfolio already built around venues where people choose to spend time rather than just money. That distinction matters enormously for long-term investors: experiential real estate is inherently local, hard to replicate, and insulated from the Amazon effect that has gutted traditional retail REITs. You cannot download a ski trip or a theme park visit. EPR’s tenants operate in spaces where consumer dollars show up in person — and the landlord collects a long-term lease regardless of how the broader economy performs quarter to quarter.

JPMorgan analyst Anthony Paolone, who rates EPR as Overweight with a $62 price target, flagged the REIT’s earnings growth trajectory as particularly noteworthy. He expects EPR’s earnings growth to rank “toward the top of the net lease REIT peer group” — a meaningful statement given that net lease REITs like Realty Income (O) and National Retail Properties (NNN) are considered the gold standard for income investors. For EPR to be tracking ahead of that peer group on earnings growth while still offering a 6%+ dividend yield and a discounted valuation represents a combination that rarely lasts long in efficient markets. The stock is up 18% in 2026, yet still trades below JPMorgan’s target — suggesting the market has yet to fully reprice the post-pandemic rehabilitation of experiential real estate demand.

For long-term investors, the compounding math here is straightforward but easy to underestimate. A 6.1% dividend yield, reinvested over a decade in a sector with pricing power and institutional barriers to new supply, can be a powerful contributor to a portfolio that doesn’t need to chase momentum. REITs distribute at least 90% of taxable income, making them structurally committed to the dividend in a way that discretionary dividend payers are not. EPR’s acquisition of the Six Flags parks adds both geographic diversification and a pipeline of assets that can be upgraded, repositioned, and leased at improved economics over time. The long-term thesis is not complicated: Americans are spending more on experiences relative to things, and EPR owns the physical infrastructure that captures that spending at the lease level — with or without a bull market.

When investors picture the AI infrastructure trade, they typically think of Nvidia chips, hyperscaler capital expenditure budgets, or fiber-optic networks. Far fewer are thinking about gas engines — but that may be exactly the kind of compounding oversight that rewards patient capital over time.

Innio N.V. (INIO), a German industrial company specializing in the design, manufacturing, and servicing of modular gas engines, went public at $27 per share in early June 2026. Within weeks, the stock had surged 37% — and this week, five of Wall Street’s largest banks simultaneously initiated coverage with buy-equivalent ratings. The price targets ranged from $42 (Goldman Sachs, implying 14% upside from recent prices) to $50 (Baird, implying 35% upside). Bank of America set its target at $46, JPMorgan at $44, and Morgan Stanley at $47. The range of conviction across these five firms is notable: it is rare for bulge-bracket banks to initiate a recently-IPO’d industrial name in such unison.

The reason for that agreement is a single structural shift hiding inside Innio’s order book. According to Bank of America, data centers accounted for just 21% of Innio’s equipment revenue over the past twelve months — but they now represent 61% of its recent orders. That 40-percentage-point swing reflects a fundamental change in how hyperscalers are approaching power. As AI workloads have grown, data center operators are increasingly bypassing public utility grids entirely, building their own on-site power infrastructure to guarantee uptime, reduce latency, and maintain quality control over power delivery. Innio’s modular gas engines are engineered precisely for this purpose: they can be deployed rapidly, scaled in segments as capacity grows, and they reduce the “time-to-power” that is now a critical constraint for facilities running large-language-model inference loads. Baird’s analyst projected that Innio’s data center sub-segment would grow at a 103.4% compound annual revenue growth rate.

For long-term investors, the most interesting aspect here is not the short-term price target spread — it is the structural moat question. Innio’s competitive advantage lies not just in its hardware, but in its high-margin servicing model: once its engines are embedded in a hyperscaler’s on-site power infrastructure, switching costs are meaningful. Morgan Stanley described the company as “one of the fastest growing companies in its peer set while also increasing its margin contribution.” That combination — accelerating revenue growth alongside improving margins — is exactly the kind of durable economic characteristic that long-term value investors look for in early-cycle industrial compounders. The risks are real: Goldman Sachs flagged Innio’s $4.8 billion backlog as both a sign of demand strength and a potential capacity bottleneck, noting that if the company cannot fulfill its obligations as fast as orders arrive, execution pressure could weigh on shares. But the broader setup — an industrial manufacturer with a defensible technology niche, sticky service revenue, and a demand driver (AI power needs) that shows no sign of decelerating — positions Innio as one of the more substantive long-term ideas to emerge from the current AI infrastructure cycle. The lesson for patient investors: the AI trade is not just a software story. The physical layer — power, cooling, connectivity — is where durable industrial moats are quietly being built.

While Wall Street spent the first half of 2026 chasing semiconductor moonshots and AI infrastructure plays, a far less glamorous sector quietly mounted one of the most compelling long-term cases of the year. Healthcare stocks rallied more than 6% in June alone — even as the Magnificent Seven group slid 3% and software names broadly retreated. For patient investors, that divergence isn’t just a monthly blip. It may signal the beginning of a durable multi-year rotation toward earnings quality, valuation discipline, and — crucially — growing dividends.

UnitedHealth Group is the anchor of this story. The health insurance giant is up 28% in 2026, yet it remains one of the most under-appreciated compounders in the S&P 500. In Q1, UNH posted adjusted earnings of $7.23 per share — well ahead of the $6.58 consensus — on revenue of $111.72 billion that also topped expectations of $109.43 billion. Management subsequently raised full-year adjusted EPS guidance to $18.25, up from $17.75. That’s not a surprise beat in a boom quarter; it’s a business with structural pricing power and cost-trend visibility improving as medical utilization moderates. Raymond James, upgrading UNH to its top picks list ahead of the July 16 earnings report, cited “moderating inpatient medical cost trend and pharmacy spend” as the conditions that support continued margin expansion at both its insurance segment and Optum Health. In May, UnitedHealth raised its quarterly dividend 5% to $2.32 per share — a signal of management conviction in the durability of its cash flows. The dividend yield is modest at 2.2%, but UNH has compounded that payout consistently for well over a decade.

The broader rotation into healthcare isn’t just a defensive reflex. As UBS strategist Gerry Fowler noted in June, “themes reflecting accelerating growth are now as appealing as the long-running appeal of AI capex beneficiaries — especially from a cheaper and less well-held starting point as earnings revisions turn positive.” That framing matters. Healthcare as a sector entered 2026 trading at a meaningful discount to technology on a forward-earnings basis, with far lower institutional ownership concentration — meaning the money hasn’t fully arrived yet. Janus Living, a senior housing REIT that debuted on the NYSE in March at $20 per share, illustrates the opportunity set even beyond insurance giants: it’s already up 45% from its IPO price, driven by recovering occupancy rates, limited new supply, and demographic demand that only grows stronger as the U.S. population ages. 10 of 11 analysts covering it rate it a buy.

For long-term investors, the so-what is this: healthcare has what technology currently lacks — cheap-enough valuations, rising earnings estimates, dividend growth, and secular tailwinds that compound quietly rather than violently. The sector doesn’t require AI euphoria to sustain it. It requires an aging population, a persistently complex insurance market, and the kind of patient capital that prefers a 5% dividend raise over a 5% options premium. The next rotation isn’t always announced loudly. Sometimes it shows up first in the sectors that were simply too boring to crowd.

Most investing mistakes don’t announce themselves. They arrive disguised as decisiveness — as the confident, urgent feeling that now is the moment to act. A recent essay from Financial Samurai puts a precise dollar figure on that feeling: $120,000, lost not by holding through a crash, but by moving too fast when one arrived.

The story is instructive in its specificity. In early March 2025, the author received $1.65 million in home sale proceeds and set a disciplined plan: invest roughly $1 million (60%) over three months as the S&P 500 declined. But when the index dipped 5% within a week, he deployed $500,000. Then another $700,000 over the next two weeks as the market fell to -10%. By the time Liberation Day hit and the S&P 500 dropped a full 20%, he had invested $1.2 million — $200,000 more than planned and weeks ahead of schedule — with almost no dry powder left to buy at the actual bottom. The result: roughly a -10% loss on the prematurely deployed capital, or about $120,000 gone. Not from bad stock picks. Not from panic selling. From impatience at the precise moment patience was worth the most.

This matters enormously for long-term investors because the behavioral math compounds in both directions. Investors who maintain disciplined deployment schedules — dollar-cost averaging over months rather than days — consistently outperform those who attempt to “time the dip” aggressively. The S&P 500’s worst single-day drops have historically been followed by more drops before the eventual recovery, which means buying-the-dip urgency tends to fire at exactly the wrong moment. The patient investor who kept $200,000 in reserve through that March-April 2025 drawdown could deploy it at a 20% discount instead of a 5% one — a difference that, compounded over a decade at 8% annual returns, translates to roughly $95,000 in additional terminal value on that tranche alone. Patience isn’t passivity. It’s a return multiplier that never shows up on a brokerage statement.

The lesson for long-term investors is not to avoid investing during corrections — it’s to pre-commit to a structure before the volatility arrives and enforce it mechanically when emotion inevitably flares. That means deciding in advance what percentage of dry powder gets deployed at each drawdown threshold (say, 25% at -5%, 25% at -10%, 25% at -15%, final 25% at -20% or deeper), then honoring it regardless of how urgent the opportunity feels. Warren Buffett’s best purchases — Bank of America preferred shares in 2011, Occidental Petroleum in 2022 — were made slowly and deliberately, not in a panic-buy rush. The competitive advantage for individual investors isn’t speed. It’s the willingness to wait until the price is undeniably right, even when waiting feels like losing.

When investors look back at the first half of 2026, many will fixate on the Nasdaq’s 19.9% gain and declare it a banner year for U.S. tech. They’ll be missing the real story. The biggest winners of the first six months weren’t in Silicon Valley — they were in Seoul, Amsterdam, and Taipei. For investors who’ve kept their portfolios anchored entirely in American equities, the H1 numbers should serve as a quiet but powerful prompt to rethink geographic concentration.

The MSCI Emerging Markets Technology index — covering large and mid-cap tech stocks across developing economies — gained more than 90% in the first half of the year. That’s not a typo. South Korea’s Kospi surged 101.1%. TSMC, the Taiwanese chip foundry that produces semiconductors for Apple, Nvidia, and virtually every major AI application, jumped 55.5%. Dutch semiconductor equipment makers ASMI and ASML gained 93.3% and 86.8%, respectively. SK Hynix, the Korean memory chip giant central to AI training infrastructure, soared approximately 300%. Meanwhile, the U.S. S&P 500 gained 9.55%, and the Nasdaq Composite added 12.79%. Even Microsoft — one of the most AI-invested companies on earth — shed 22.9% of its value in the first half. The gap between international and domestic tech returns wasn’t a rounding error; it was enormous.

The underlying thesis here is structural, not tactical. The AI buildout is fundamentally a global supply chain story. The chips Nvidia designs are fabricated almost entirely by TSMC in Taiwan. The memory required for large-scale model training runs through SK Hynix and Samsung in Korea. The lithography machines that make advanced semiconductor manufacturing possible come almost exclusively from ASML in the Netherlands. U.S. companies are designing the software layer of AI, but the physical infrastructure — the actual capital-intensive, hard-to-replicate manufacturing base — lives overwhelmingly abroad. These are businesses with deep competitive moats, long investment cycles, and pricing power that compounds over years. Yet for much of the past decade, U.S. investors largely ignored them in favor of domestic names trading at far richer valuations.

What this means for long-term investors is straightforward: global diversification is not just a risk management exercise — it’s a return driver. The companies enabling the AI revolution that U.S. tech firms are building on top of often trade at materially lower price-to-earnings multiples than their American counterparts, carry less headline risk from domestic policy uncertainty, and benefit directly from any sustained expansion in AI capital expenditure worldwide. The BlackRock Investment Institute, in its midyear outlook, described AI as potentially enabling “a permanent growth breakout by accelerating innovation itself.” If that thesis proves correct over the next decade, the compounding will likely run through TSMC, ASML, and their peers just as much as it runs through Nvidia or Microsoft. Long-term investors who’ve never held a share of any of them may want to seriously reconsider whether home-country bias is costing them more than they realize.

For the first time in a generation, small-cap stocks are outpacing their mega-cap rivals by a wide margin — and the reasons behind the surge have more staying power than most investors realize. The Russell 2000 Index has climbed more than 21% in the first half of 2026, its strongest six-month performance since 1991. That’s not a random fluke driven by speculative froth. It’s a structural rerating backed by improving fundamentals and a broadening of the AI investment cycle that patient, long-term investors should take seriously.

The catalyst is familiar but the beneficiaries are new. While Nvidia and the mega-cap tech giants have dominated headlines for two years, the capital they’re deploying into AI infrastructure is now cascading down the supply chain to hundreds of smaller companies. Semiconductor equipment makers, specialty component suppliers, and connectivity solutions providers — many of them tucked inside the Russell 2000 — are capturing revenue they simply couldn’t access before. Chip-related companies account for 16 of the index’s 50 best performers this year; Aehr Test Systems, Ichor Holdings, and MaxLinear have each gained more than 400%. Critically, these aren’t companies chasing a narrative. They’re booking real orders from real customers with multi-year spending commitments. Earnings growth forecasts for Russell 2000 companies have already been revised upward to 38% for 2026, up sharply from the 23% projection made just at the start of the year, according to LPL Financial.

What makes this moment especially interesting for long-term investors is the valuation setup that preceded the rally. Small caps spent the better part of four years in relative purgatory, battered by higher interest rates that disproportionately hurt companies with floating-rate debt and thin margins. That persistent underperformance created a valuation gap that Amy Zhang, portfolio manager at Alger, described as wide enough to “drive a truck through.” That gap is now closing — not through speculation, but through genuine earnings improvement. The broader implication is important: the AI infrastructure buildout isn’t a winner-take-all story reserved for five or six megacap names. It is a multi-year capital cycle that rewards patient investors willing to look beyond the obvious. Small-cap indexes like the Russell 2000 or the S&P 600 — especially value-oriented slices of those benchmarks — may still offer a meaningful margin of safety relative to the lofty valuations now embedded in large-cap tech. As long as interest rates remain stable and AI spending continues to expand, the structural case for owning a diversified slice of quality small caps alongside larger holdings has rarely looked more compelling.

One of the most reliable blue-chip industrials on the planet is flashing an uncomfortable signal for disciplined investors: Caterpillar Inc. (NYSE: CAT) now trades at nearly 31 times its 2026 forecasted earnings — a multiple that sits miles above where this stock has historically bottomed, and one that deserves serious scrutiny before anyone calls it a value play.

The excitement is understandable. Drive past any data center construction site in America and you’ll see CAT bulldozers and excavators at work. The company posted record revenues in 2025, and Wall Street has piled on a narrative that Caterpillar is also a critical supplier of natural gas turbines — filling the power gap that utilities will take years to address as AI data center electricity demand explodes. The result: a stock that has been re-rated like a tech company, not the cyclical industrial it has always been. At 31x 2026 earnings, it would need to grow into roughly 18x its 2029 estimates just to appear reasonable — and even that assumes no recession, no infrastructure slowdown, and no mean-reversion in the AI infrastructure spending cycle.

Here’s the number that long-term investors should anchor to: historically, Caterpillar has troughed — at cycle peak profits, mind you — at below 12x earnings. That’s not a bear case; that’s the historical floor during good times. The implication is sobering. Even if Caterpillar continues to execute at a high level, the current valuation may already price in several years of favorable outcomes. Heartland Opportunistic Value Equity Strategy flagged this in its Q1 2026 investor letter, noting that AI infrastructure enthusiasm has created “extreme valuation disparity” between perceived AI winners and losers across the industrial landscape — and Caterpillar has become a prime example of a quality company that has drifted into speculative pricing territory. The business is excellent. The price is a different question entirely.

So what does this mean for long-term investors? Caterpillar remains a durable franchise with a strong dividend history, deep competitive moats in heavy equipment manufacturing, and genuine exposure to multi-year infrastructure spending trends. The business fundamentals are not in question. But patience matters enormously here. Investors who bought CAT below 15x earnings in prior cycles captured decades of compounding returns; those who chased the stock at elevated multiples often waited years just to break even. The lesson isn’t to avoid Caterpillar forever — it’s to avoid overpaying for it now. In a market where AI enthusiasm is repricing quality industrials like growth stocks, the disciplined investor’s job is to separate the durable franchise from the temporary narrative premium, and wait for the math to make sense again.

For most of the 20th century, the American auto industry ran on a simple assumption: each year, the country would buy a little more than it did the year before. That assumption is now breaking down — and the implications for long-term investors in automakers, auto dealers, auto-parts suppliers, and even insurance companies are profound.

A new analysis from Bain & Company lays out what they call a “perfect storm” bearing down on the industry. Ten years ago, the U.S. set a record with 17.6 million cars, trucks, and SUVs sold in a single year. According to Bain partner Mark Gottfredson, the country may never come close to that number again. The firm projects that by 2040, U.S. new vehicle sales could fall by more than 2 million units from current levels — not because of a recession or temporary shock, but because of irreversible structural shifts. Separately, AutoForecast Solutions currently expects sales to remain flat at roughly 16 million through at least 2033, after which the trajectory looks increasingly uncertain.

Three reinforcing forces are driving the slowdown. First, demographics: the U.S. fertility rate in 2025 sat at roughly 1.6 births per woman — meaningfully below the 2.1 replacement rate — and Bain expects restrictive immigration policies to cut historical net migration rates roughly in half over the next 15 years. Fewer people means fewer license holders and fewer buyers, and this math is already baked into the census data. As Gottfredson told CNBC: “We already know how many people have been born and how many people will be of vehicle driving age at age 16 in 16 years from now.” Second, behavioral change: today, only half of 16-year-olds have a driver’s license, compared with nearly 70% between 1966 and 1984. Young buyers aged 18–34 accounted for under 10% of new vehicle registrations by mid-2025, down from 12% in early 2021. Meanwhile, buyers aged 55 and older now represent nearly half of all new registrations — a buyer base that will itself shrink over time. Third, affordability: new vehicle monthly payments are up 30% over four years, with nearly one in five new vehicles now carrying a monthly payment above $1,000. Uber, Lyft, and remote work have reduced the urgency of ownership for younger households.

There is one wrinkle that actually extends the decline further: cars are simply lasting longer. The average vehicle on U.S. roads hit a record 12.8 years of age in 2025, according to S&P Global Mobility. The vehicle deregistration rate — the pace at which older cars leave the road — has already dropped from 6% in 2000 to roughly 5% in 2025, and Bain projects it could fall to 4.4% by 2040. Longer vehicle life suppresses replacement demand, which is another headwind for the new-car market. Automakers are competing for a customer base that is both shrinking and holding on to its existing vehicles longer than ever before.

For long-term investors, the takeaway is not a single trade — it’s a framework for questioning the assumptions embedded in the valuations of auto-exposed businesses. Legacy automakers like Ford, GM, and Stellantis have historically traded at low price-to-earnings multiples on the premise that they were cyclical, not secular, businesses. But if the U.S. market is genuinely entering a period of structural contraction, that “cheap” multiple may reflect permanent earnings pressure rather than a buying opportunity. Gottfredson summed it up plainly: “There are too many automakers and too many brands competing for consumers. The market is going to have to consolidate.” Investors who understand structural industry contractions before the market fully prices them in — think newspaper publishing, brick-and-mortar retail, or landline telephony — tend to avoid costly mistakes in seemingly inexpensive stocks. The same discipline applies here.

When Microsoft needs 2.7 gigawatts of power — enough to light up 3 million homes — to fuel its AI data center in Texas, it doesn’t call a software vendor. It calls GE Vernova and orders seven gas turbines. That single transaction captures something most investors scanning tech stocks are missing entirely: the companies best positioned to profit from the AI boom may not be the ones writing the models, but the ones keeping the lights on.

GE Vernova’s largest turbine plant in Greenville, South Carolina, is running at a pace that would have seemed implausible three years ago. The company added 200 workers last year and is hiring 300 more before year’s end. Its order book is fully booked through 2029, with contracts extending into 2031. Every major hyperscaler — Amazon, Google, Microsoft, and Oracle — has sent executives to walk the factory floor. GE Vernova turbines are already powering Elon Musk’s xAI Colossus 1 campus in Tennessee and roughly one gigawatt worth are being deployed for OpenAI’s Stargate project in Texas. Today, roughly 20% of its gas power order book flows to AI-related applications — a figure that was near zero just a few years ago. Meanwhile, turbine prices have surged 300% over the past three years, according to analysts at Melius Research, with a single unit now running more than $250 million. GE Vernova’s stock has gained nearly 60% in the past six months, yet the forward visibility from that locked-in backlog argues the rally has structural legs rather than speculative froth.

For long-term investors, the GE Vernova story isn’t a trade on AI sentiment — it’s a case study in infrastructure moats. The turbines are 31 feet tall, weigh 280 tons, take years to manufacture, and require a specialized workforce that cannot be assembled overnight. Competing at scale in this market takes decades of engineering know-how and supply chain depth that no startup can replicate quickly. Demand for firm, dispatchable power at gigawatt scale is not going away; if anything, the AI capex supercycle ensures it accelerates. A global buildout of AI data centers requiring reliable baseload electricity points directly to gas turbines as the “picks and shovels” of the infrastructure layer. With a multi-year booked order pipeline, pricing power clearly intact, and a customer list that reads like the Fortune 10, GE Vernova has quietly assembled one of the most durable industrial moats of the current technology era — and the market is only beginning to price it in.

While stock market headlines obsess over AI chip cycles and mega-cap volatility, a quieter opportunity has been compounding in the fixed income world — one that Schwab’s top income strategist says long-term investors cannot afford to ignore. With the 10-year Treasury yield stubbornly anchored between 4% and 4.5%, Collin Martin, head of fixed income research and strategy at the Schwab Center for Financial Research, sees three distinct income pockets offering attractive absolute yields that he believes won’t last at these levels indefinitely.

The first opportunity sits in investment-grade corporate bonds, which are currently yielding an average of around 5%. At first glance, tight credit spreads — meaning the yield advantage over Treasurys is historically low — might give cautious investors pause. But Martin argues the tight spreads are a feature, not a warning sign: corporations are entering the second half of 2026 with strong profit growth and healthy balance sheets. “That low risk premium isn’t necessarily scaring us away,” he told CNBC. “We are focusing more on the absolute yields and the income you can earn.” For patient investors who remember the near-zero yield environment of 2020-2021, 5% from high-quality issuers represents a generational reset worth capturing. A diversified basket of investment-grade ETFs remains the most accessible vehicle for most individual investors.

The second idea is a modest tilt toward high-yield bonds — specifically, raising allocation by one to two percentage points beyond what a typical balanced portfolio holds. Martin acknowledges the default risk but points to a structural shift in the Bloomberg U.S. Corporate High Yield Index: higher-rated credits (BB-rated bonds) now make up a meaningfully larger share of the index than a decade ago, improving the overall quality floor of the asset class. ETFs like the Schwab High Yield Bond ETF (SCYB) currently carry a 30-day yield of 6.88% with a rock-bottom 0.03% expense ratio, while the iShares Broad USD High Yield Corporate Bond ETF (USHY) offers a 6.96% 30-day yield with a 0.08% expense ratio — thin enough that the income isn’t being quietly eroded by fees.

The third and most overlooked idea is preferred securities, where yields of roughly 6% come bundled with a meaningful tax advantage: most preferred dividends are qualified, meaning they’re taxed at rates of 0%, 15%, or 20% rather than ordinary income rates. For investors in higher tax brackets, the after-tax yield on preferreds can rival or exceed the pre-tax yield on comparable bonds. The iShares Preferred and Income Securities ETF (PFF) carries a 6.32% 30-day yield, while the Invesco Preferred ETF (PGX) offers 6.33%. Martin also notes that preferred securities, despite their long or perpetual maturities, tend to track credit markets more closely than duration — making them less sensitive to rising long-term Treasury rates than investors typically assume.

The takeaway for long-term investors is structural: for the first time in over a decade, fixed income markets are paying investors meaningfully for taking on credit risk, and the current environment may represent the last window before rate normalization compresses those yields. Martin warns that the Fed’s increasingly hawkish posture under Kevin Warsh could keep long rates elevated — or push them higher — making extended duration in Treasurys the one area to avoid. But across investment-grade corporates, quality high-yield, and preferred securities, a thoughtfully constructed income sleeve of 5% to 7% yield now functions as a compounding anchor for a diversified long-term portfolio — something that simply wasn’t available for most of the past fifteen years.