R124/R130 Bull Seagull: A New Paper-Promotable Model Family

Repository reference: cutebacktests

Abstract

The May 15 result finally cleared the bar that the earlier May passes did not. The new R124/R130 family is paper-promotable as a P0/P1 candidate under the repository framework.

The selected family is r124_r130_causal160_first8d_qqq_bull_seagull_slots: a QQQ bull seagull / protected risk-reversal structure expressed as eight deterministic slot lanes. It is not live-capital approved, and it was initially dry-run only. But as a model-family result, it is the strongest new paper candidate we have.

Question

The question was whether the search had finally found a new options-only family with good Sharpe, DSR, PBO, quote cleanliness, and executable paper runtime support.

On May 15, the answer became yes for P0/P1 paper candidacy. The broader point is that paper trading needs stricter integrated tooling before it should be treated as a clean deployment signal.

Method

R124 added causal public-state filters to the quote-native seagull/throttle source rows. The important public features included prior risk-on breadth, prior IWM/SPY relative strength, cross-asset dispersion, core breadth, semis leadership, and market-regime labels.

R130 then decomposed the chosen public-state lead into slot variants and cumulative first1d through first8d books. The selected object was the cumulative first-eight-day expression, rendered as eight independent dry-run slot lanes for paper confirmation.

That decomposition matters. A multi-leg options structure can look attractive at the combined level while hiding unstable behavior in the individual trade sequence. Splitting the candidate into deterministic slots lets the research process ask a more precise question: is the edge coming from one lucky entry, or does the cumulative first-N expression keep improving as the controlled slots are added?

In this case, the strongest evidence came from the cumulative first8d book rather than from a single isolated slot. That is important because it means the candidate was not promoted as a one-trade anecdote. It was promoted as a structured family whose selected expression improved the active research book under the relevant checks.

Strategy Shape

A bull seagull is a defined-risk, multi-leg options structure. In plain language, it is designed to express bullish participation while using additional option legs to reshape the payoff. The point is not simply to buy a call and hope the underlying rises. The point is to express a directional view with a more controlled payoff profile than a naked single-leg position.

That is why this family is materially different from many earlier candidates. A single-leg option strategy can be easier to backtest and easier to paper trade, but it can also overstate practical edge when premium decay, spread width, and entry timing are not handled carefully. A multi-leg structure forces the research engine to evaluate the whole package: all legs, all quotes, the net entry price, the exit value, and the combined risk.

The selected family is also not a generic "always bullish" rule. The causal state gate matters. It requires a risk-on backdrop and a minimum relative-strength condition before the candidate is eligible. That keeps the strategy from becoming an unconditional QQQ options bet and makes the paper candidate easier to reason about.

Evidence

The selected family:

• Family: r124_r130_causal160_first8d_qqq_bull_seagull_slots
• Candidate: r130_r124_slot__causal_160_first8d
• Structure: QQQ bull seagull / protected risk reversal
• Causal gate: prior_iwm_spy_rel_5d >= -0.01326065318 and prior_risk_on_breadth_5d >= 0
• Research weight: 0.35x

The weight is intentionally modest. Paper promotion does not mean a strategy deserves a full-size allocation. It means the strategy has enough evidence to deserve controlled confirmation. A 0.35x research weight keeps the candidate visible while respecting the fact that it still has to prove its behavior outside the backtest artifact.

The research gates were green:

The book CSCV diagnostics also passed:

The execution gates were clean: quote fill rate 1.0, fallback-last count 0, and stock row count 0. Dry-run validation passed for the slot contracts, and the contracts failed closed when the day's causal predicate was not active.

How To Read These Results

The headline result is not raw PnL. Raw PnL was positive, but the more important evidence is the combination of additivity, robustness, and execution cleanliness.

Daily Sharpe delta and weekly Sharpe delta ask whether the candidate improves the active book at different time resolutions. That matters because a strategy can look useful day by day while adding unstable weekly behavior, or it can smooth weekly returns while hiding ugly daily concentration. This candidate improved both.

Incremental PnL asks whether the candidate actually adds something after the existing book is already present. That is a harder test than evaluating the strategy alone. The goal is not to find the most attractive standalone curve. The goal is to find a sleeve that improves the combined system.

Residual top-3 day share checks concentration. A strategy that gets most of its contribution from one or two days is difficult to trust in paper trading because live observation may spend weeks waiting for another rare event. The selected row was still concentrated enough to watch carefully, but it was materially better than many rejected near misses.

PBO and DSR address selection pressure. PBO asks how often the apparent winner fails when the cross-validation split changes. DSR asks whether the selected Sharpe remains meaningful after accounting for the search process. The promotion-scope PBO of 0.002 and selected DSR of 0.9999998518 are the clearest robustness evidence in the table.

The quote and fallback gates are equally important. A paper candidate should not depend on optimistic last-price fills or stock-proxy rows. The selected family cleared the quote-fill, no-fallback, and options-only gates, which is why it could move beyond research-only status.

What Worked

The family combined several things that rarely line up:

• a new option structure rather than another single-leg variation;
• public-state causal gates that can be produced before the session;
• meaningful PBO/DSR diagnostics across promotion, scout, and all scopes;
• no stock rows and no fallback-last fills;
• deterministic paper slot contracts;
• tested multi-leg entry, exit, and combo-aware reconciliation support.

That is why the completion audit marked the goal goal_complete for paper-promotable P0/P1 candidacy.

Why It Beat The Prior Near Misses

The earlier May search found plenty of rows that were interesting but not promotable. Some had positive standalone PnL but failed residual additivity. Some had too few unique residual days. Some looked strong until the robustness checks changed the selection pressure. Others were too hard to express cleanly in paper trading.

R124/R130 cleared more of the stack at once:

• it used a distinct multi-leg option expression;
• it kept the traded object options-only;
• it relied on causal public-state features instead of future labels;
• it improved the active book on daily and weekly measures;
• it passed meaningful PBO/DSR checks;
• it had clean quote and fallback evidence;
• it had a deterministic dry-run expression suitable for controlled paper confirmation.

That combination is what makes the result promotable. It is not that the family had one impressive number. It is that fewer parts of the evidence contradicted each other.

Takeaway

R124/R130 is the first new family in this pass that deserves the phrase paper-promotable. It passed the research gates, the book CSCV gates, the quote/fallback/stock-row gates, and the P0 dry-run execution gates.

But paper trading is still a validation surface, not a victory lap. The model family is strong enough to promote into controlled paper confirmation. The next step is making the integrated paper module strict enough that paper results are easier to trust and easier to debug.