Relative Strength Breakout Strategy: Testing Proxy-Based Intraday Breakouts With QQQ and DIA

Repository reference: cutebacktests

Abstract

Relative-strength breakout strategies are often described too loosely. A trader says one symbol is stronger than the market and then treats that statement as a signal. In a research setting, that is not enough. The proxy has to be defined, the comparison has to be normalized, and the breakout still has to prove itself quickly enough to matter.

The c4 family in this repository is useful because it formalized that problem. As summarized in Episode 8, the stock-level logic required the primary ticker to break its opening structure while outperforming a beta-adjusted proxy by a minimum relative-strength edge. The guard_v3 version tightened that edge, capped the proxy's own movement more aggressively, added a beta-shock veto, and forced the trade to prove itself quickly. Later repair work even introduced an SPY-to-DIA proxy override so the SPY sleeve could be benchmarked against a broader market proxy without disturbing the default map elsewhere.

Question

The practical question is not whether relative strength matters. It is how to define a relative-strength breakout strategy tightly enough that it can be tested and falsified.

That is exactly why the c4 branch is interesting. It takes a familiar discretionary idea and turns it into a code-faithful object. The primary symbol must be stronger than its proxy by a specified edge, the proxy's own move must stay controlled, and the breakout has to validate quickly instead of wandering around for a long time.

Method: How a Relative Strength Breakout Strategy Was Built Here

The c4 family uses a proxy-relative framing rather than a standalone breakout framing. The strategy does not ask only whether a ticker is breaking its opening structure. It asks whether that move is happening while the ticker is outperforming a chosen proxy after adjusting for beta and while the proxy itself is not exhibiting a movement pattern that invalidates the comparison.

This produces several useful controls:

• a minimum relative-strength edge versus the proxy
• caps on proxy movement so the comparison remains interpretable
• a beta-shock veto
• time pressure so the trade must prove itself quickly

In the SPY repair path, the repo added an SPY-to-DIA override. That is an especially useful example because it shows the proxy is not a decorative input. Changing the proxy can change what the branch is actually testing.

Evidence / Results

The c4 branch did not end as a winner, but it did become more interesting after repair. In Toward The One Piece Of Sharpe, the repaired stock stage restored realistic trade counts and later follow-ups reached 79 and 85 trade variants. That is an important result because it showed the earlier zero-trade collapse was not a valid scientific conclusion. Part of it was a launch bug around an empty writable DuckDB.

Even so, the branch was eventually parked. That outcome makes the strategy more educational, not less. It means the proxy-relative framing produced something real enough to study and still not strong enough to admit.

What Worked

What worked was the research framing itself. The c4 family is one of the repo's strongest examples of turning a discretionary concept into a measurable strategy. It asked a more sophisticated question than "did the symbol break out?" It asked whether the symbol broke out while outperforming a relevant proxy under a controlled relative-value regime.

The repair work also worked in an important sense. It prevented the wrong conclusion. Without fixing the launch issue and the proxy/path details, the branch might have been dismissed as dead when it was actually a near-miss.

What Failed

What failed was admission. The proxy-relative idea improved after repair and still did not clear the portfolio gate. That is an important negative result because it stops the public story from turning into "the bug fix rescued the strategy." The bug fix rescued the measurement. It did not automatically create a promotable branch.

The dispersion family also exposed that relative strength was not evenly distributed across sleeves. The QQQ-specific descendants were more compelling than SPY. That further shows why proxy selection and symbol decomposition matter. The family was not one homogeneous edge.

Takeaway

A relative strength breakout strategy is only meaningful if the proxy relationship is defined tightly enough to be tested. The c4 family is a good case study because it built that structure explicitly, learned something real from the repairs, and still failed the final admission bar.

If you want the symbol-level dispersion result, Dispersion Trading Backtest: QQQ vs SPY and Why the Signal Was Not Symmetric is the best companion. If you want the admission decision itself, Why c4 Was Parked: A Dispersion Strategy That Improved But Still Failed the Portfolio Gate carries the story into portfolio context. Join the research log to get the next backtest and failure report.