Competitive Rotation: Why Portfolios Should Be Zero-Sum Tournaments

---

title: "Competitive Rotation: Why Portfolios Should Be Zero-Sum Tournaments" description: "Buy-and-hold assumes your best ideas stay your best ideas. They don't. Here's the rotation rule that swaps weak holdings for stronger candidates — but only when the edge is real." publishedAt: "2026-03-15" updatedAt: "2026-03-15" keywords: ["portfolio rotation strategy", "stock rotation", "active portfolio management", "position sizing"]

TL;DR. Competitive rotation swaps the weakest holding for the strongest available candidate when the score gap is at least 0.20. A position must be held at least 30 days before it can be rotated out. No more than 3 rotations per rebalance. If 75% of portfolio notional has already been touched by recent rotations, the engine pauses. Rotation is frozen entirely in CRISIS regime. Everything below is the reasoning behind those numbers.

The buy-and-hold failure mode

You bought X six months ago at a Tessera Rating of 0.85. It was the best idea on the board that week. Six months later, the signal for X has decayed — earnings quality slipped, the sector re-rated, the P/E discount closed. X is now sitting at 0.62. Meanwhile, Y has appeared in the screen at 0.91. Y was not available when you bought X. Y is available now.

In a pure buy-and-hold framework, you do nothing. You sit in X until a stop-loss fires or the thesis breaks so badly the signal goes negative. The opportunity cost of holding 0.62 while 0.91 is available never shows up on your P&L statement as a loss, but it is a loss. It is the return you did not earn.

Rotation asks a simple question on every rebalance: is each position still earning its slot?

Rotation as a continuous tournament

Think of the portfolio as a 20-seat tournament. Every seat is occupied by the best player we could find at the time it was filled. New players arrive every week — the screening-service discovers fresh candidates from the full investable universe on each rebalance cycle. If a new player is meaningfully stronger than the weakest incumbent, the incumbent loses its seat.

This is not "sell on price drop." Price drops trigger stops, which are a separate mechanism. Rotation triggers on relative signal strength: we sell not because X got worse, but because Y got better enough to justify the swap.

The distinction matters. A stop-loss asks, "has this position broken?" Rotation asks, "can we do better with this capital?" They are complementary, not redundant.

Tessera's rotation rules (exact thresholds)

| Rule | Value | Why | | --- | --- | --- | | Score gap threshold | 0.20 | Below this, the difference is within noise of the scoring model | | Min holding days | 30 | Avoids churn on week-to-week signal wobble | | Max rotations per rebalance | 3 | Prevents turnover spirals in volatile weeks | | Capacity cap | 75% | If recent churn has already touched most of the book, pause and let it settle | | CRISIS regime | Frozen | Don't swap losers for "best losers" in a broad bear market |

Score gap of 0.20. The Tessera Rating aggregates 24 quality factors plus the RELATIVE P/E signal into a score between roughly 0 and 1. Based on backtest grid search, gaps smaller than ~0.20 between incumbent and candidate produced negative net alpha after transaction costs — the signal difference was not large enough to survive the round-trip cost plus the uncertainty in the scores themselves. Gaps above 0.20 produced persistent alpha. So 0.20 is where the signal-to-noise ratio crosses a useful threshold.

Minimum 30-day holding period. Weekly rebalance produces weekly score updates, and scores wobble. A position that screens 0.78 this week might be 0.81 next week and 0.75 the week after without anything fundamentally changing. The 30-day minimum holding period ensures we are rotating on persistent signal changes, not transient noise. It also reduces pathological cases where a freshly bought position immediately gets rotated out on the next rebalance because a slightly better candidate appeared.

Maximum 3 rotations per rebalance. In volatile markets, many candidates appear strong simultaneously — usually because the regime has shifted and the signal model is recalibrating. Without a cap, the portfolio could rotate 8-10 positions in a single week, generating enormous turnover on what might turn out to be a spurious regime shift. Capping at 3 means the portfolio evolves, not thrashes.

75% capacity cap. This is a cumulative check over the trailing window of rotations. If the rotations over recent rebalances have already touched 75% of portfolio notional, further rotations are paused until the churn settles. This is a brake on the "everything looks attractive in new regime, rotate everything" failure mode.

CRISIS regime freeze. In CRISIS regime (breadth below 25%), rotation is disabled entirely. More on this below.

The tradeoffs at looser or tighter thresholds

A score gap of 0.20 is a calibration decision, not a law of markets. It is worth articulating what happens on either side.

If the threshold were 0.10, rotations would be more frequent. You would capture more small edges, but you would also rotate on positions that later turn out to have been equivalent — the difference was within noise — and you would pay the round-trip transaction cost on every swap. In backtesting, this produced higher turnover without meaningfully higher returns, and lower risk-adjusted performance after costs.

If the threshold were 0.30, rotations would be rare. You would miss real, persistent edges. The portfolio would drift toward buy-and-hold and forfeit the rotation premium entirely.

0.20 is where the backtest grid search settled. It is not magic. It is the best parameter we could identify on historical data. Live results will differ from backtested results, which is why we are explicit about survivorship bias, look-ahead, and other backtest caveats.

The calibration mindset: these parameters are measured, not chosen. They reflect what worked on historical data with honest out-of-sample discipline. When we ship a parameter change, it is because new data has moved the optimum, not because it felt right.

Transaction cost consideration

Rough math on a 20-stock portfolio. If rotation produces ~20 swaps per year (roughly 1x portfolio turnover from rotation alone, not counting stop-outs and rebalances), and each round-trip costs ~0.05% on liquid US equities (tight spread plus minimal commission), that is roughly 5 basis points of drag per year on the total portfolio.

5 bps is small but not zero. For rotation to earn its keep, it must produce more than 5 bps of alpha annually. In backtesting, rotation has produced considerably more than that — typically 100-300 bps over buy-and-hold variants of the same strategy — but backtests overstate live performance, so we keep the cost discipline visible.

If the portfolio shifts toward less liquid names, or if spreads widen in a stressed market, the cost math changes. The 3-rotations-per-rebalance cap plus the 75% capacity cap are both partial mitigations for cost-blowup scenarios.

Regime interaction

In CRISIS regime, rotation is frozen entirely. This is a deliberate asymmetric rule.

The reasoning: in a broad drawdown where breadth is below 25%, most stocks are declining together. Rotating from one declining stock to another declining stock is theater — you are paying transaction costs to swap a bad holding for a marginally-less-bad holding, with no real improvement in risk or return. The correlation of losses is too high for relative-ranking to add value.

Better to hold, let stops do their work, and let the cash that accumulates from stopped-out positions simply sit. Cash is a position. In CRISIS, it is often the right one.

The full regime logic is covered in Regime-Aware Investing.

Failure modes (honest caveats)

Rotation is not a free edge. Specific ways it can fail:

• Correlated positions. Rotating within the same sector does not diversify. If the portfolio is already heavy in one sector and the incoming candidate is also in that sector, the rotation increases concentration rather than reducing it. Sector caps at the portfolio level partially address this, but the rotation rule itself does not directly check correlation.
• Momentum chase risk. The Tessera Rating includes factors correlated with recent price action. In strong trending markets, rotation can systematically buy recent winners near their tops. The quality factors (earnings quality, balance sheet strength, cash flow durability) partially counteract this, but the risk is real in late-cycle momentum markets.
• Data-mining risk. The 0.20 threshold, the 30-day minimum hold, the 3-rotation cap — all were calibrated on backtest data. They may be slightly overfit to the period they were calibrated on. We monitor live performance against backtest expectation and will recalibrate if the live distribution drifts.
• Tax drag. In a taxable account, rotation creates taxable events. US short-term capital gains are taxed as ordinary income — typically 22-37% federal for US individual investors. A strategy that produces 200 bps of alpha before tax can produce 80 bps after tax in a high-turnover configuration. Rotation is most efficient in tax-advantaged accounts (IRAs, 401(k)s). This is not investment advice, and tax treatment varies by jurisdiction and individual situation.

Related reading

• Regime-Aware Investing
• Quality Screening Factors
• Position Sizing and Stop Losses