One strategy is a bet. Three uncorrelated ones are a portfolio.
Every mechanical strategy has weather it hates, and sitting through a 59% drawdown in one sleeve is how people quit. The model portfolio holds three families that make money differently, at equal weight, rebalanced quarterly. It is the same portfolio for everyone who runs it.
| Sleeve | Family | Weight | CAGR | Sortino | Max DD |
|---|---|---|---|---|---|
Leaders | Trend | 33.3% | 28.2% | 2.24 | −30.2% |
Topline | Fundamental | 33.3% | 34.9% | 3.39 | −26.0% |
Metronome | Accumulation | 33.3% | 42.6% | 2.41 | −59.2% |
Model portfolio All three, equal weight, rebalanced quarterly | Blend | 100% | 40.6% | 3.76 | −30.7% |
S&P 500 Benchmark, same window | Index | — | 14.7% | 1.39 | −28.7% |
The blend’s Sortino of 3.76 is higher than any sleeve it is made of, which is the whole argument for holding all three. It also cuts the worst sleeve’s −59.2% drawdown to −30.7%. Read the next table before trusting either figure too far.
Every figure above is computed on quarter-end marks, and that flatters the risk numbers. Here is by how much.
| Marked | Observations | CAGR | Sortino | Max DD |
|---|---|---|---|---|
| quarterly | 38 | 37.9% | 2.49 | −39.5% |
| monthly | 112 | 39.3% | 2.10 | −40.7% |
| weekly | 119 | 276.8% | 3.49 | −40.8% |
The same two sleeves, the same window, and only the mark frequency changes. Return barely moves. Sortino falls -40% and the drawdown deepens by 1.4 points, because a quarter-end mark cannot see a fall that happened and recovered inside the quarter. A sleeve down 30% in February and back by 31 March contributes nothing to downside deviation.
We publish the quarterly figure because revGro genuinely rebalances quarterly, and forcing it to monthly marks would forward-fill flat months and inflate its Sortino to nonsense. So the quarterly number is the only honestly computable one for this blend. Treat the published Sortino as roughly a quarter too high and the drawdown as a floor.
| Leaders | Topline | Metronome | |
|---|---|---|---|
| Leaders | 1.00 | -0.24 | 0.68 |
| Topline | -0.24 | 1.00 | -0.26 |
| Metronome | 0.68 | -0.26 | 1.00 |
Read this honestly. Momentum and Cadence run positively correlated — both are long US growth in different clothing, and they will hurt at the same time. revGro is the only genuine diversifier, negative against both. So this is three sleeves and closer to two independent bets, which is the honest way to describe it.
How the returns are computed
- Every figure is a hypothetical backtest. Each sleeve is the simulated equity curve of its fixed ruleset, not a brokerage account, and almost all of the record was computed after the fact.
- The three sleeves. Leaders is a relative-strength screen of large US companies. Topline is a quarterly fundamental screen of large US equities, stock only. Metronome holds a leveraged Nasdaq ETF alongside cash, with no new money added. Each strategy’s page describes it at the same level.
- Equal weight, rebalanced quarterly. Each quarter’s portfolio return is the plain average of the three sleeves’ returns for that quarter, which is what resetting each to one third at every quarter end produces. The chart compounds those returns from $100,000.
- Quarter-end marks. Each sleeve is valued on its last mark in each calendar quarter, and only quarters where all three sleeves and the S&P 500 have a mark are used: 2017-03-31 to 2026-12-31, 39 quarters. CAGR, Sortino, Sharpe and drawdown are all computed on those quarterly returns.
- The latest quarter is usually unfinished. The last mark is 2026-10-01. The return from the previous quarter end to that date is counted as a full quarter, which slightly overstates CAGR when the quarter has gone well and understates it when it has not.
- Costs. Leaders charges 10 basis points each side of every trade and Metronome 5 basis points. Topline models no trading costs. Nothing is charged for the quarterly rebalance between sleeves, and no taxes, slippage or fees are modelled.
- Benchmark. The S&P 500 is SPY with dividends reinvested, measured on the same quarter-end marks over the same window.
What is and is not in this
- Three sleeves, not four. A mean-reversion sleeve was removed: it scored a Sortino of 0.58 on weekly marks over eleven years and ran 0.83 correlated with Cadence, so it was a second bet on the same asset wearing a different name.
- The hedged contribution strategy is deliberately excluded. Of its terminal value, a known share is deposited cash. Equal-weighting its return against a lump-sum sleeve would count cashflow as performance. It has its own page.
- A trend-following sleeve was removed entirely after it returned less than the index over the same window while taking more than twice the drawdown.
- Equal weight is a choice, not an optimum. It is published in advance and identical for everyone precisely so that it is not fitted. An optimised weighting would score better here and generalise worse.
- One historical window. Backtested, hypothetical, with the benefit of hindsight. No costs, taxes or slippage are modelled at the blend level, and quarterly rebalancing across three sleeves would generate short-term gains in a taxable account.