What Wall Street’s $40 Trillion Actually Runs On (And Why It Breaks When AI Arrives)

The Corporate Jets Problem

Here’s a fact that sounds like an investing strategy: companies that own corporate jets outperform companies that don’t.

It’s true. Statistically significant. You can backtest it for decades and it holds up.

So should you buy every company with a Gulfstream in the hangar? Obviously not. The jet isn’t why they outperform. The jet is a proxy. Companies with corporate jets tend to be large, profitable, well-managed, and generating enough free cash flow that a $50 million aircraft is a rounding error. The jet is a side effect of quality. It has zero causal relationship with returns.

This is the problem with almost everything Wall Street runs on. And most investors — including most professional fund managers — don’t fully understand it.

People chase signals. Insider buying. Lobbying spend. Patent filings. Short interest. And yeah, some of these correlate with outperformance. But the vast majority are corporate jets — proxies for something deeper, not independent predictors of anything.

The entire $40+ trillion quantitative investing industry is, at its core, a machine for separating the jets from the engines. And it’s a machine that works beautifully — right up until the world changes underneath it.

What $40 Trillion Actually Runs On

Let’s get specific, because this matters.

The backbone of modern institutional investing is something called factor models. The idea is simple in principle: identify measurable characteristics of stocks that predict future returns, then build portfolios that systematically harvest those characteristics.

The academic version starts with Fama and French in 1993. They showed that three factors — the overall market, company size, and book-to-market ratio (basically “cheapness”) — explained the vast majority of stock return variation. Since then, researchers have been hunting for more factors. And they found them. Hundreds of them.

By 2019, the academic literature had documented over 400 published factors that supposedly predict stock returns. Four hundred. That’s not a toolkit. That’s a junk drawer.

The finance world calls this the “factor zoo” — and the name tells you everything. Most of these factors are garbage. Data-mined artifacts. Researchers ran thousands of regressions against historical data, found patterns that looked statistically significant, published papers, and moved on. When you test those same factors on new data they’ve never seen? About 390 of them collapse.

The Surviving Factors: What Actually Works (With Caveats)

Out of 400+ published factors, here’s what actually survives rigorous out-of-sample testing. This is the real toolkit — and its limitations matter more than its strengths right now.

Five factors. That’s the real list. And notice something important: every single one of them is backward-looking. They measure what has already happened — past cheapness, past momentum, past profitability, past volatility, past size. They assume the relationships that held over the last 30 years will continue to hold over the next 30.

For most of financial history, that was a reasonable assumption. Industries evolved slowly. Disruption played out over decades. Mean reversion was the dominant force — what went down came back up, what went up came back down.

Then AI showed up.

The Value Trap Machine

This is where it gets really important. Pay attention, because this is the single biggest blind spot in quantitative investing right now, and it’s about to cost a lot of people a lot of money.

Here’s how factor models see a company in a disrupted industry:

• Earnings: Still positive (for now). Check.
• Price-to-Earnings ratio: Low — trading at 6x. The model sees “cheap.”
• Book value: High relative to market price. The model sees “undervalued.”
• Cash flow: Still generating. The model sees “quality.”
• Verdict: Buy signal.

Now here’s what the factor model can’t see:

• The company’s core service is being automated by AI at 90% lower cost
• Three of its top five clients are piloting AI replacements
• Its talent pipeline is drying up because new graduates want to work in AI, not legacy consulting
• The revenue looks stable because enterprise contracts are 3-year commitments that won’t renew
• The business model is structurally dying

This isn’t theoretical. It’s already happening. Look at the legacy IT services sector. Companies like DXC Technology, Unisys, and Conduent have been “cheap” on every value metric for years. Quant funds have been buying them. And they keep getting cheaper, because “cheap” and “dying” look identical to a factor model.

The factor model sees the price. It cannot see the reason for the price.

That’s not a minor limitation. During a period of structural disruption across dozens of industries simultaneously, it’s a catastrophic one.

Why Can’t They Fix This?

You’d think smart quants would just add a “disruption” factor. And some have tried. The problem is fundamental: factor models work by finding statistical patterns in historical data. AI disruption at this scale has never happened before. There is no historical data to train on. You can’t backtest an event that hasn’t occurred.

This is the core epistemological problem: the most important variable for stock selection in the next decade — a company’s vulnerability or resilience to AI disruption — is inherently qualitative. It requires understanding business models, competitive dynamics, talent flows, client behavior, and technology adoption curves. None of which fit in a regression.

Industry Momentum: The Signal We’re Leading

In 1999, Tobias Moskowitz and Mark Grinblatt published a paper that should have changed how everyone invests. It showed that buying winning industries and selling losing industries generates significant alpha — and that a large chunk of individual stock momentum is actually just industry momentum in disguise.

Think about what that means. When a tech stock is “hot,” it’s often not because of anything company-specific — it’s because the entire sector is rotating. When a bank stock is falling, it’s usually not the CEO’s fault — it’s a sector-level move. Industry momentum explains roughly 50% of individual stock momentum.

This finding has been replicated dozens of times across global markets and time periods. It’s one of the most robust findings in all of quantitative finance.

But here’s the thing about momentum: it’s a lagging indicator. By definition, momentum strategies buy industries after they’ve already started winning. You’re following the signal, not leading it.

Traditional momentum strategies bought cloud computing stocks after they’d already run up 200%. They sold legacy retail after Amazon had already eaten the margins. The momentum premium is real — ~4.5% annually — but you’re always late. And during AI disruption, being late means buying semiconductor stocks at peak euphoria and selling healthcare services right before AI makes them more valuable, not less.

Why 88% of Fund Managers Lose

S&P Global publishes the SPIVA Scorecard every year. It compares active fund managers against their benchmarks. The results are devastating and consistent:

• Over 1 year: ~60% of large-cap fund managers underperform the S&P 500
• Over 5 years: ~80% underperform
• Over 15 years: ~88–92% underperform

These aren’t amateurs. These are people with Bloomberg terminals, proprietary data feeds, Wharton MBAs, and teams of 50 analysts. They have every advantage you’d think matters. And almost nine out of ten of them can’t beat an index fund you can buy for 3 basis points.

The conventional explanation is fees and transaction costs. And sure, a 1% management fee and 0.5% in trading costs makes it hard to outperform after expenses. But that only explains a few percentage points of underperformance. The deeper problem is structural.

The Mean Reversion Trap

Most active strategies — whether they use factor models explicitly or not — are built on an assumption of mean reversion. Cheap stocks get less cheap. Expensive stocks get less expensive. Industries that fell behind catch up. Industries that got ahead slow down.

And for most of market history, this was true. Mean reversion was the dominant force in equity markets. The problem is that disruption breaks mean reversion.

When AI structurally destroys the business model of legacy IT consulting, that industry doesn’t mean-revert. It declines permanently. When AI gives cloud platforms an entirely new revenue stream, those stocks don’t “revert to the mean” either. They re-rate to a new level.

Fund managers who buy “cheap” disrupted industries and sell “expensive” beneficiary industries are systematically getting it exactly backward. And the factor models that guide their decisions are telling them to do it.

The Renaissance Exception

There’s one fund that makes the 88% statistic look almost quaint.

Renaissance Technologies’ Medallion Fund returned approximately 66% annually before fees for over 30 years. After their famously steep fees (5% management, 44% performance), investors still netted roughly 39% per year. Over three decades. Consistently. Through dot-com crashes, financial crises, and every market regime in between.

It is, by any measure, the most successful investment fund in the history of capital markets.

And here’s what matters for our purposes: Renaissance doesn’t use traditional factor models.

Jim Simons didn’t hire economists or MBA finance graduates. He hired mathematicians, physicists, signal processing engineers, and computational linguists. The team applied techniques from code-breaking and speech recognition to market data. They looked for patterns in places the academic finance world wasn’t looking — microstructure effects, cross-asset correlations at unusual frequencies, non-linear dynamics that linear regression can’t capture.

The most successful fund in history looked at the same market everyone else was looking at and said: “Your tools are too primitive. The real signals are in places your framework can’t reach.”

The “Quantamental” Sweet Spot

The smartest institutional investors have figured this out. The current holy grail in professional money management is called “quantamental” investing — a hybrid approach that uses quantitative factors for filtering and screening, combined with qualitative structural analysis for actual stock selection and portfolio construction.

The logic is straightforward:

• Quant factors are good at: Processing thousands of stocks quickly, eliminating obviously bad candidates, identifying broad characteristics worth investigating
• Quant factors are bad at: Understanding business model disruption, assessing competitive dynamics, evaluating management adaptation capability, anticipating structural industry shifts
• The solution: Use quant to narrow the universe from 5,000 stocks to 500, then use qualitative structural analysis to pick the 50 that actually matter

Firms like Point72, Citadel, and Bridgewater are spending billions building these hybrid systems. They’re hiring AI researchers not to build better factor models, but to build systems that can process qualitative information at scale — earnings call transcripts, patent filings, satellite imagery, supply chain data.

But here’s the thing: even these sophisticated hybrid approaches are still reactive. They’re processing qualitative data about what’s already happening. They’re reading earnings calls where the CEO already talked about AI disruption. They’re analyzing patent filings that were already published. They’re faster than pure quant, but they’re still following, not leading.

What This Means for Your Portfolio

Let’s pull this together, because the implications are concrete.

If you own index funds, you’re fine — you’ll get the market return. But if you’re trying to beat the market, or even tilt your portfolio toward the right sectors, the tools available to retail investors are almost entirely backward-looking. The ETFs you can buy are factor-based. The stock screeners you use are factor-based. The analyst ratings you read are built on factor models with qualitative overlays that are still mostly reactive.

None of them answer the question that actually matters right now: which industries are being structurally reshaped by AI, and which companies within those industries are positioned to win or lose?

That question lives in a bucket that the quant world calls “inherently qualitative.” It requires:

• Understanding how AI actually works in each industry (not just “AI is coming”)
• Mapping cross-industry cascade effects (when transportation gets disrupted, what happens to insurance? to real estate? to retail?)
• Tracking the psychology of adoption — workforce resistance, regulatory response, public sentiment
• Analyzing geopolitical dynamics — US-China compute restrictions, EU regulation, export controls
• Identifying historical parallels — how past technology revolutions played out, and where we are in that arc
• Separating hype from deployment — what’s actually being used in production vs. what’s still a demo

The quantamental sweet spot that Point72 and Citadel are spending billions to build? That’s what our engine is for retail investors. Quantitative structure for filtering and scoring. Qualitative depth across 8 analytical dimensions for actual insight. Cross-industry mapping that no factor model captures. Forward-looking, not backward-looking.

We’re not competing with quant funds on their turf. We’re operating in the space they can’t reach — the space where the next decade of returns will actually be determined.

The Bottom Line

Wall Street’s $40 trillion runs on factor models. Those models assume stable relationships. AI disruption breaks stable relationships. The 88% of fund managers who rely on these tools will continue to underperform — and the underperformance will get worse as disruption accelerates.

The Medallion Fund proved that the biggest returns go to whoever finds signals outside the conventional toolkit. The quantamental movement proves that the industry knows the toolkit is broken. And our engine provides what both of those insights point toward: a structured, multi-dimensional, forward-looking analysis of which industries win and which lose as AI reshapes the global economy.

The question isn’t whether your portfolio is exposed to AI disruption. It is. The question is whether you understand how it’s exposed, and whether you’re positioned on the right side of the biggest structural shift since the internet.

Related reports: How We Score 28 Industries  |  Why Math-Only AI Predictors Fail  |  AI Fear vs Your Portfolio