Section 03 — Spare Parts

Prior IPs

Rejected, Shelved & Paused Strategies

Don't let prior work go to waste. Take apart failed IPs for signals, methods, and lessons.

The Rule

Do Not Re-Submit Unchanged Failures

Do not re-submit an unchanged failed IP. If you want to revive a shelved strategy, document what materially changed and why it addresses the original rejection reasons. Come prepared to explain the delta.

Rejected and shelved IPs are a resource — not just a record of failure. Before starting a new IP, review prior work in the same asset class or strategy family. You may find signal construction logic, data pipelines, or test frameworks that save significant time.

Strategy Graveyard

All prior IP reports and codebases are archived in OneDrive under AlgoGators / Research / Prior IPs /. Review before starting a new IP to avoid duplicating prior work.

Strategy Name Semester Status Rejection Reason What's Salvageable
[Strategy Name] [Semester] Rejected Sharpe below minimum; excessive parameter sensitivity in the signal lookback. Signal construction logic, data cleaning pipeline for this asset class.
[Strategy Name] [Semester] Shelved Data source too expensive / unreliable to maintain in production. The mean-reversion model itself; could be re-applied with alternative data.
[Strategy Name] [Semester] Rejected No economic rationale; pure in-sample curve fit, collapsed out-of-sample. Walk-forward testing framework written for this IP can be reused.
[Strategy Name] [Semester] Paused Member graduated before completing incubation. Promising concept not fully tested. Full codebase and report available in OneDrive. Worth revisiting.

How to Revive a Shelved IP

What You Must Document

  • Original rejection reason (verbatim from the prior review)
  • What materially changed in your revised proposal
  • Why those changes address the specific failure mode
  • New data or methodology that wasn't available before

What Doesn't Count as Material Change

  • Tweaking parameters until the backtest looks better
  • Extending the backtest window to include a favorable period
  • Changing the asset class without changing the signal logic
  • "We have a better understanding now" without specific evidence
← Research Hub Fund Performance →