What is alpha?

The 10-second definition

If the S&P 500 returned 10% last year and your portfolio returned 15%, your raw outperformance was 5 percentage points. Of those 5, some came from taking on more risk (leverage, concentration, small-caps). The rest — the part that came from actually picking better — is your alpha.

Formally: Alpha = Your return − (Risk-free rate + Beta × (Market return − Risk-free rate)). But don't get hung up on the formula. The thing you want to internalize: alpha is the part of a return that skill produced, not exposure.

Why most investors have none

The uncomfortable truth from four decades of research: 85% of active US fund managers underperform their benchmark after fees over any 10 year window. The median "professional" investor has negative alpha.

This isn't because they're dumb. It's because:

• Markets are close to efficient. When thousands of smart people with trillions of dollars all look at the same public data, the easy edges get arbitraged away fast.
• Fees eat excess return. A 1.5% management fee on a strategy that earns 1% alpha/year is a net −0.5% vs. the benchmark. The math has to clear a high bar.
• Style drift kills compounding. Most funds drift toward benchmark-hugging because deviating costs them jobs if they underperform for 18 months — even if the long-run thesis was right.
• Survivorship bias inflates the average. The funds that blew up don't show up in today's "average active fund" number. The true failure rate is higher than the brochures suggest.

What the 15% have in common

The managers who do produce persistent alpha share surprisingly few traits. Looking at the 30 tracked superinvestors on HoldLens — Buffett, Ackman, Klarman, Druckenmiller, Burry, Marks, Tepper, Hohn, Coleman, Li Lu — the common threads are not what most retail investors think:

1. Concentration. Most run 20-40 positions, not 300. Their top 10 positions are typically 60-85% of the book. You can't outperform by owning everything.
2. Patience. Average holding period for Buffett's core positions is 15+ years. Druckenmiller famously holds some trades for weeks — but when he's wrong, he's out the next morning, not hoping.
3. A structural information edge. Hohn's TCI has a reputation for regulatory/legal depth no retail analyst can match. Druckenmiller has a macro network most generalists can't replicate. Value investors like Klarman run screens on esoteric securities (distressed debt, cross-holdings, special situations) that most funds don't touch.
4. Willingness to look wrong. Burry shorted housing in 2005 while his fund was down 20%. Buffett sat on $150B+ in cash through 2024's melt-up. Being alone with a thesis for 2+ years is what generates alpha — not consensus correctness.
5. Explicit loss aversion over gain seeking. Howard Marks wrote 20 years of memos before anyone cared. His central idea — "the best investors make fewer mistakes" — is boring, obvious, and almost never practiced.

Alpha vs. beta, in pictures

Beta is your portfolio's sensitivity to the market. A beta of 1.0 means you move with the S&P. A beta of 1.5 means you move 1.5× as much. A beta of 0 (theoretically) means you're uncorrelated.

Alpha is everything else. If your portfolio's beta is 1.2 and the market returned 10%, your "market exposure return" was 12%. If your actual return was 15%, your alpha is 3 percentage points. If your actual return was 9%, your alpha is −3 — you underperformed even adjusting for your higher market exposure.

This is why a 30% year for a tech-heavy fund during a tech bull market isn't impressive. And it's why a 4% year for a macro fund when the market dropped 20% is exceptional.

How HoldLens uses alpha

Every manager on HoldLens has a quality score derived from their historical alpha-over-benchmark. When you see the signed −100..+100 ConvictionScore, it's weighted by this quality — a buy from a manager with a decade of +8% alpha counts more than a buy from a manager with a decade of +0.2% alpha.

That's the honest use case. We don't claim to predict tomorrow's alpha — we rank today's smart-money conviction by yesterday's demonstrated ability to produce it.

What not to do with this knowledge

Mechanically copying a high-alpha manager's 13F doesn't transfer their alpha to you. The copy-trading myth explains why in detail, but the short version: alpha lives in decisions you don't see (cost basis, position sizing, timing of trims), in instruments that never touch a 13F (shorts, options, foreign equities), and in the willingness to hold through drawdowns that retail copycats almost never stomach.

Treat 13F signals as research starting points, not trades. A consensus buy from five high-alpha managers tells you where to spend your own research time. It doesn't tell you when to click buy.

Further reading

• Can you actually copy Warren Buffett? — the full case for why transferring alpha doesn't work.
• How to read a 13F filing in 5 minutes — so you can evaluate alpha claims yourself from primary source data.
• What is a ConvictionScore? — our alpha-weighted aggregation of manager buy/sell signals.
• Manager backtests — concrete alpha numbers for Buffett, Ackman, Druckenmiller, Klarman, Burry.
• The 30 tracked superinvestors — profiles with philosophy, longest-held position, and current portfolio.