The Numbers: Does Tilting Actually Beat Owning the World?

The one-line answer

On paper, tilting toward cheap and rising countries looks like it adds a huge +6 to +8% a year. In the real world, the funds that actually do it have lagged simply owning the whole world for the past decade. A realistic honest edge, over a full cycle, is roughly −1% to +2% a year — small, uncertain, arriving in sudden bursts after long droughts, and paid for with a bumpier ride.

This confirms the “0–2%” range I gave you earlier — now with the actual numbers behind it, refined slightly to −1% to +2%.

What the backtests promise (the seductive part)

These are the numbers that sell books and funds. They are real studies — and they are backtests, which means they end at the moment the strategy went live.

Value tilt — buying the cheapest-CAPE countries

• Meb Faber's Global Value: the cheapest countries returned ~16–18%/yr vs ~9.6%/yr for the international benchmark, 1980–2013 — a +6 to +8%/yr spread. (Before taxes, fees, and trading costs.)
• Academic backing (Keimling/StarCapital, 17 countries, 1881–2015): a starting CAPE under 10 was followed by ~11.7% real/yr over the next 10–15 years; a CAPE over 30 by only ~0.5–3%/yr. The catch: CAPE explains only ~48% of what happens next — half the outcome is noise, and it only works over 10–15-year horizons.

Momentum tilt — rotating into rising countries

• Gary Antonacci's Dual Momentum (GEM), backtest 1974–2013: ~17.4%/yr vs ~9.5%/yr for global buy-and-hold — a +8%/yr edge, with smaller drawdowns.
• Country momentum studies: a real but more modest ~2–5%/yr gross premium.

If these held up, you'd be foolish not to tilt. They don't hold up. Here's why.

Why you don't get that: the three leaks

Between the backtest and your account, the edge leaks away in three places:

1. Decay — the edge shrinks once it's known. The landmark study (McLean & Pontiff): published market edges are ~26% smaller out-of-sample and ~58% smaller after publication. Rule of thumb: you keep maybe 40–50% of a backtest at best — and some of it was never real, just curve-fitting.

2. Costs — the tilt is expensive to run. Single-country funds cost 0.5–0.7%/yr (vs 0.06% for a world fund), momentum trades a lot (turnover + taxes), and the whole tactical-fund category averages **1.4%/yr** in fees. That comes straight off the top.

3. Behavior — you have to survive the drought. The average investor already loses ~1–8%/yr to badly-timed buying and selling. Tilts make this worse because they trail longest right before they pay — so people quit at the bottom.

What actually happened to real money

This is the part that matters most. Backtest vs live:

                        BACKTEST said   →   REALITY delivered
Value tilt (GVAL)       +6 to +8 %/yr   →   −7 %/yr for a DECADE
                                            (~−2 %/yr even after 2025)
Momentum (GEM)          +8 %/yr         →   −5 %/yr vs owning stocks
                                            (live, 2014–2026)
Tactical funds overall   —              →   30 of 34 did WORSE than
                                            doing nothing (10 yrs)

• GVAL (the real global-value ETF): from its 2014 launch to April 2024 it returned just 1.9%/yr while the world index did 8.8%/yr — behind by 7 points a year, every year, for ten years. Then in 2025 it jumped +56% in a single year — lifting its lifetime return to ~7.2%/yr, still short of just owning the world (9%/yr).
• GEM (the famous momentum system): since it was published it has returned ~8.4%/yr vs ~13.6% for simply holding stocks — and even its signature “smaller crashes” advantage mostly vanished live.
• Trend-following turns out to be a drawdown tool, not a return booster: historically it roughly matches buy-and-hold returns while cutting the worst crash from ~−46% to under −10%. In the long bull market since 2009, it mostly cost return.

The lag is the real price

The hardest truth isn't the size of the edge — it's the waiting:

• A country tilt typically swings ±8–16% vs the benchmark in a single year, and trails it for 3–5 years routinely, sometimes 10+.
• Value stocks trailed growth for ~13 straight years (2007–2020) before turning.
• GVAL's entire payoff came in one year (2025) after a decade of pain — rewarding only those who never flinched.

“Three years is a long time, five years very long, ten years an eternity” — yet these strategies routinely demand all three. Most people can't hold that long, which is why most people don't capture even the small edge.

The honest number

Putting it together, a realistic expectation for a disciplined, low-cost country tilt held through a full cycle:

• Roughly −1% to +2% a year versus just owning the world.
• Reliably negative for multi-year stretches — even a full decade.
• Back-loaded — the payoff, if it comes, arrives suddenly after a long wait.
• Bumpier — more volatility for that small, uncertain reward.

And in dollars, at a $750 account: a +1%/yr edge is about $7.50 a year. Even in the good case, the tilt is a rounding error next to how much you contribute.

So what does this mean for us?

Not “don't do it” — but “prove it, keep it small, and know what you're signing up for.”

1. The plain world index (VT) is the champion to beat. It's cheap, hard to beat, and has beaten the tilt funds for a decade.
2. This is exactly why we paper-test. The tilt has to actually beat VT on paper before real money follows it — not just look good in a backtest.
3. Keep the tilt a minority (the 20–25% sleeve), so a decade of lag dents you but never sinks you.
4. The real growth lever is contributions and time, not the tilt.

The tilt isn't a money machine. It's a small, patient, uncertain bet — and the most valuable thing we get from testing it is understanding, which is the whole point of this trial.

---

Next: I can show you what the paper trial itself would look like — how we'd run the tilt against a plain-VT benchmark and actually see, month by month, whether it's earning its keep.