Why c4 Was Parked: A Dispersion Strategy That Improved But Still Failed the Portfolio Gate

Repository reference: cutebacktests

Abstract

The c4 branch is one of the best public examples in this repository because it combines three distinct research realities. First, there was a real implementation problem that distorted the branch. Second, the branch became materially more interesting after that problem was repaired. Third, it still failed the portfolio gate and was parked.

Public strategy research often collapses those events into one vague story. This repo did not. As Episode 8 makes clear, the c4 branch improved after the repair, recovered 79 and 85 trade rows in follow-up work, and still ended with the explicit label park_c4.

The c4 result is best compared with Dispersion and Relative-Strength Backtests: Proxy Series and Data Limits, Backtesting Robustness, and Backtesting Data Quality Checklist. That framing keeps correlation regime, proxy construction, rolling window, split-adjusted bars, sample count, PBO, DSR, and drawdown in the evidence trail.

Question

The useful question is not whether c4 ever looked promising. It clearly did. The useful question is why promise was still not enough.

Most near-miss strategies do not die because they are obviously absurd. They die because they are not clearly additive enough once the full portfolio bar is applied. c4 is a textbook example of that process.

Method: Why c4 Was Evaluated as a Portfolio Candidate Rather Than as an Isolated Winner

The c4 family is a dispersion-relative breakout strategy. It asks whether the primary ticker is breaking its opening structure while outperforming a beta-adjusted proxy by a minimum edge. After the repair work, the branch then had to survive a more complete gate. It was not enough to be interesting on a rerun.

The admission bar, summarized in Episode 8, required:

• selection_status=feasible
• positive return
• trades_per_week >= 1.5
• orb_overlap_days >= 30
• c66_overlap_days >= 30
• offset_ratio_on_orb_down_days >= 0.5
• zero extra option attempts
• zero quote rejects

This is why the c4 story is so useful. It shows exactly what happens when a branch is judged as a potential portfolio member instead of as a chart pattern with a nice rerun.

Evidence / Results

The repo's summary now gives a balanced picture. In Toward The One Piece Of Sharpe, c4 is described as a promising research branch whose repaired stock stage restored 79 and 85 trade variants. That is the positive side.

The negative side is the final decision. By April 18, the repo's answer was still park_c4. The branch was not rejected because it was random junk. It was rejected because the combined admission conditions were still too demanding for what the repaired branch could deliver.

This is a much stronger public result than a simple win or loss. It tells the reader what improved, what remained weak, and which exact conditions still blocked admission.

What Worked

What worked was the debugging discipline. The repo did not mistake a launch or storage issue for a strategy conclusion. It repaired the path, reran the branch, and then judged the branch again. That prevented the wrong scientific story.

The branch itself also worked well enough to stay interesting. c4 is not in the same category as the failure-week cemetery of zero-trade or no-feasible-profile lanes. It became a real near-miss, which is why it remains one of the most educational parts of the portfolio journey.

What Failed

What failed was the combined gate. c4 accumulated several smaller deficits rather than one dramatic flaw. Feasibility was still demanding. Overlap and offset requirements were still meaningful. Option parity cleanliness still mattered. The branch did not become clearly additive enough relative to c66 and the broader portfolio goal.

That is an important negative result because many strategies die exactly this way. They do not collapse spectacularly. They remain too ambiguous to admit.

Takeaway

c4 was parked because the branch improved and still did not clear a portfolio-aware gate. That is one of the healthiest kinds of negative result to publish. It shows that the repo is not confusing post-bug recovery with promotion.

If you want the family context, Relative Strength Breakout Strategy: Testing Proxy-Based Intraday Breakouts With QQQ and DIA explains the logic underneath c4. If you want the QQQ-vs-SPY decomposition that made the dispersion branch more legible, Dispersion Trading Backtest: QQQ vs SPY and Why the Signal Was Not Symmetric is the next read. Join the research log to get the next backtest and failure report.

How the terminology applies

For Why c4 Was Parked: A Dispersion Strategy That Improved But Still Failed the Portfolio Gate, the backtesting workflow should treat Point-in-time contracts, Quote-aware fills, Reject reasons, Replay artifact, Cache key, and Signal timestamp as operational state rather than glossary decoration. That framing keeps the research claim causal: the strategy can only select instruments, prices, and labels that existed at the decision time.

A developer implementing this Validation idea should persist Look-ahead leakage, Walk-forward test, Slippage model, Same-bar fill, Promotion gate, and Options data API beside the result, instead of leaving those words in a term card. It also turns attractive performance into an auditable record where fills, skips, thresholds, and replay inputs can be challenged independently.

The review artifact for Why c4 Was Parked: A Dispersion Strategy That Improved But Still Failed the Portfolio Gate becomes more useful when OPRA-originating data, OCC option symbol, Bid/ask spread, Midpoint, Quote/trade condition, and Quote vs trade semantics appear in the same body of evidence as the selected rows. When a result is promoted, these fields should appear in the run manifest, rather than a prose summary or final equity curve.

In production notes for this backtesting workflow, REST snapshot, WebSocket stream, Entitlement gate, Quote freshness, Timestamp semantics, and Pagination cursor define the checks that decide whether the workflow is reproducible. The result is a backtest that can be rerun, compared across threshold families, and rejected when the evidence is not strong enough.

For Why c4 Was Parked: A Dispersion Strategy That Improved But Still Failed the Portfolio Gate, the practical acceptance test is simple: another developer should be able to read the body, identify the exact inputs, reproduce the request sequence, and explain the accepted and rejected rows without relying on the bottom terminology grid. If a phrase appears in the page vocabulary, it should correspond to a stored field, a validation check, a replay step, or an implementation decision in the backtesting workflow.

This is also the reason the article should not measure success only by the final chart, table, or headline metric. The better standard is whether the data path, timing model, entitlement state, and evidence trail survive review. When those pieces are written directly into the body, the terminology becomes part of the workflow readers can implement.

Parking c4 required comparable evidence

The c4 decision needed more than a return comparison. A dispersion sleeve has to preserve the primary ticker, proxy ticker, beta adjustment, signal timestamp, OHLCV aggregates, market session, and relative-strength threshold before it touches options. The option expression then needs point-in-time contracts, selected OCC option symbol, quote window, trade window, spread percent, quote freshness, and reject reasons.

That evidence matters because dispersion failures can hide in the join. The stock leg can look plausible while the option leg fails on NBBO width, no-bid exits, quote condition filters, or low open interest. The proxy can also introduce timestamp drift if the primary and benchmark aggregates are not aligned to the same session and bar boundary.

Parking a branch is easier to defend when the replay manifest shows those fields. It tells future researchers whether c4 was weak because the relative-strength signal was poor, because the option market could not express it cleanly, or because the sleeve overlapped too much with better candidates. Those are different reasons to stop work.