Gap Up Failure Fade Backtest: The Difference Between Intuition and Evidence

Repository reference: cutebacktests

Abstract

Gap-up failure fade is the kind of setup many discretionary traders find immediately attractive. Overnight enthusiasm fades, the market fails to reclaim VWAP, the bounce stalls, and the short side looks clean. The story is coherent. The challenge is whether that story survives once the pattern is converted into a repeatable strategy.

This repository's c32 branch answers that question more honestly than most public writeups. As described in Episode 7, c32 looked for a gap-up session that failed to reclaim VWAP and then shorted the continuation of that failed bounce. The quality version made the gap threshold larger and shortened the deadline. The result was still no_feasible_profile, with failed DSR, Sharpe, Sortino, and trades-per-week requirements.

Question

The practical question is not whether failed bounces exist. The market produces them constantly. The useful question is whether the failed-bounce pattern is robust enough to be a strategy rather than a memorable chart anecdote.

That is the distinction the c32 branch makes possible. Once the gap size, reclaim behavior, and VWAP timing deadline are specified, the setup becomes something that can disappoint you honestly.

Method: How the Gap Up Failure Fade Backtest Was Structured

The c32 branch defined a fairly intuitive reversal archetype. A stock gaps up into the open, cannot reclaim VWAP cleanly, and then continues lower after the bounce fails. The quality version then makes the setup more selective by requiring a larger gap and a shorter deadline.

This is a good structure for testing because it turns intuition into constraints. The strategy is no longer "short failed strength." It becomes a measurable combination of overnight gap, intraday reclaim failure, and time-limited continuation.

Evidence / Results

The outcome in Episode 7 is clear:

• lane: c32 gap-failure fade
• result: no_feasible_profile
• main blockers: failed DSR, Sharpe, Sortino, and trades/week

That means the branch could tell a convincing market story more easily than it could produce robust out-of-sample evidence. This is exactly the sort of negative result that should be public because it is so easy to over-believe the pattern.

What Worked

What worked was the conversion of narrative into testable logic. The repo did not let the setup stay at the level of "this often happens." It defined the conditions tightly enough to let the evidence push back.

That alone is valuable. A failure under explicit structure teaches more than a success under a vague idea.

What Failed

What failed was the evidence bar. The branch could generate a plausible narrative and still could not satisfy the combination of robustness and activity needed to survive. This is a common fate for reversal ideas built from visually appealing post-open behavior. The pictures are easier to remember than the sample statistics.

The c32 result is also a reminder that tightening a setup does not always create a stronger strategy. Sometimes it creates a cleaner story and a weaker sample at the same time.

Takeaway

The gap-up failure fade backtest shows the difference between intuition and evidence very clearly. The intuition was good enough to deserve a test. The evidence was not good enough to earn continuation.

If you want the broader negative-results context, Failed Trading Strategies: 7 Ideas We Tested So You Do Not Have To is the main companion. If you want the continuation-side failure instead of the reversal-side failure, Gap Reclaim Strategy Backtest: Why a Good Chart Pattern Failed the Data is the natural pair. Join the research log to get the next backtest and failure report.