Performance The model portfolio, measured.
Mosaic Mid-Horizon · Live since 2026-01-02
What the model portfolio has returned since it went live, against the index and against the way this kind of investor actually allocates, before and after a fee. The arithmetic behind every figure is at the bottom of the page.
Every figure on this page is as at 2026-09-28, on end of day prices. This is 8 months of history. That is short. I do not annualize it, and neither should you.
Since inception
- Total return
- +16.6%
- After the fee
- +16.0%
- SPY (TR), same window
- +12.7%
- 60/10/30 peer mix, same window
- +7.9%
- Worst drawdown
- -8.2%
- Annualized volatility
- 14.5%
- Beta to SPY (TR)
- 1.03
- Return per unit of risk
- 1.34
- SPY (TR) volatility
- 13.2%
- SPY (TR) worst drawdown
- -8.9%
- Worst complete month
- -4.7%
- Positive months
- 4/7
- Holdings
- 11
- Rebalances
- 1
Peak 2026-02-25 to trough 2026-03-30
correlation 0.94
SPY (TR) 1.12 on the same basis
2026-03
Since 2026-01-02
Return per unit of risk is the window's total return divided by the volatility realised in it, at a zero rate and not annualised, for the portfolio and the index alike. Beta is the slope of the portfolio's daily moves on the index's.
Cumulative return since inception
Model portfolioAfter the feeSPY (TR)60/10/30 peer mix
Built from 60% SPY, 10% AGG, 18% VNQ, 6% IBIT, 6% GLD, rebalanced to those weights every day.
The survey puts this investor at roughly 60% public equities, 10% bonds and cash, and 30% private and alternatives. Private equity, venture and private credit have no daily public price, so listed real estate, bitcoin and gold stand in for that third of the mix. It is a rough shape of how these portfolios are allocated, not a measurement of what they returned.
The fee on this page is not mine
The second line is the same portfolio charged 0.70% a year, which is the average fee the Long Angle 2026 survey found its members paying an adviser. I charge a subscription and I hold none of your capital, so this is not what StandPoint costs. It is there so the comparison is fair: if this book had been run for you at the going rate for having a book run for you, that is what would have been left.
Shorter windows
The chart above is always the whole record and has no control on it. These are here so a recent stretch that runs against the whole cannot hide inside it.
| Window | From | Portfolio | After fee | SPY (TR) | 60/10/30 peer mix |
|---|---|---|---|---|---|
| 1 month | 2026-08-28 | -0.3% | -0.4% | -0.5% | -1.8% |
| 3 months | 2026-06-26 | +3.9% | +3.7% | +5.0% | +3.3% |
| Since inception | 2026-01-02 | +16.6% | +16.0% | +12.7% | +7.9% |
The picks, restated
The single names are a different product from the portfolio above and are measured differently. What follows is the version I would want to see if I were reading someone else's record.
| The picks, restated | Positions | Average | Median | Win rate | Average hold |
|---|---|---|---|---|---|
| All positions | 255 | +4.5% | +5.2% | 65% | 15d |
| Closed | 251 | +4.6% | +5.3% | 65% | 16d |
| Open, marked | 4 | +0.3% | +1.1% | 50% | 6d |
Where the closed positions' return came from
The average closed position, +4.6% over 251 positions, split three ways over its hold.
- The market+0.9%
- The sector+0.8%
- The names themselves+2.9%
172 of 251 closed positions did better than their market and sector explain.
The market is the name's beta, as I measured it at entry, times SPY over the hold. The sector is the S&P names in the same GICS sector, equal weight, beyond the market. The rest is the name. Same caveat as the table: each position counts once.
Best single position +34.0%. Worst -21.7%.
Open positions are marked at their last close, 2026-09-28.
Each position counts once here, whatever size it was taken at, because I publish no sizing. So this is a record of decisions and not the return of a portfolio. The return of a portfolio is at the top of this page.
How these numbers are made
The portfolio series is computed once per trading day from the composition in force on that day, then compounded forward. It is not a look through at today's holdings priced backwards, which would credit names I bought last month for months I did not hold them.
Both comparisons run over the identical window as the portfolio, anchored on the same first day, using total return prices so dividends are in.
The net line accrues the fee on elapsed calendar days rather than on trading days, because a fee is charged over a weekend too. Nothing else is deducted: no tax, no slippage, no commission.
The picks figures include open positions marked to their last close. Closed only would let a book that sells its winners and holds its losers report a rising average the whole time it went backwards, and I would rather you were able to check that than be told not to worry about it.