STUDY LOG

Every study. Including the ones that failed.

68 studies from the research record, each with its verdict in plain words: 3 adopted, 55 tested and closed. The ideas and the results are here; the exact tunings of the rules that survived are not. Back to For the Quants

3Adopted 2Still open 8Findings 55Tested and closed

Adopted · 3Passed its bar out of sample and runs in the engine.

Sep 30, 2026Peregrine

Holding XSP to settlementPasses. Letting XSP settle beats selling into the spread before the close.

Sep 18, 2026Peregrine

A RUT veto that only ever removes tradesAdopted on paper, September 18, 2026.

Sep 17, 2026Peregrine

An in-the-money strike, gated by dealer positioningAdopted for Peregrine.

Still open · 2Registered and being judged on new sessions only.

Sep 29, 2026Peregrine

The signed dealer pin, tested forwardPre-registered and judged on new sessions only.

Sep 26, 2026Peregrine

A language model as a second opinionPre-registered and running in shadow, judged only on sessions after its first verdict. It has no say in any trade.

Findings · 8Measurements that shape the work, not rules that trade.

Sep 22, 2026Peregrine

The level a trend call has to holdA broken swing is retested by the close 70% of the time, the same as any price at that distance. Distance matters; the level does not.

Sep 20, 2026Peregrine

Peregrine's entry model against the rule it replacedIt cuts drawdown and ties up less capital. It is not shown to add return.

Sep 17, 2026Research

Is the gamma pin a magnet?No. Price does not actually pin. The pin works as a tailwind, not a target.

Sep 17, 2026Peregrine

Legging into vertical spreadsThe 72% win rate is real. The downside protection is not: it is adverse selection.

Sep 17, 2026Research

What a win rate can and cannot tell youAt our sample sizes a gap under about five points cannot be resolved, which rules out most fine-tuning by win rate.

Sep 16, 2026Research

Base rates of the normal-variation-day frameworkIts claims about how far a day travels reproduce. Its claims about direction do not.

Sep 16, 2026Peregrine

Entering early versus waiting for a better setupEarly beats late decisively. The selection is real, but waiting for it costs more than it adds.

Sep 14, 2026Peregrine

How far a trend call runs against you firstThe median adverse move is 12 basis points in 30 minutes and 28 by the close. The opposite trend gets called before the close about 31% of the time.

Tested and closed · 55Failed its bar. Kept here so nobody re-runs it.

Sep 30, 2026Research

An options-implied probability cloudFails both pre-registered tests. The desk draws the plain implied cone.

Sep 30, 2026Peregrine

Dealer positioning as a correlate0 of 17 cells pass. Only the simple 09:35 read is useful.

Sep 30, 2026Research

Does S&P 500 breadth lead SPX?No. It moves with /ES and says nothing about the next move.

Sep 30, 2026Research

Does the local read move the probability cloud?No local state earns a drift or a width.

Sep 30, 2026Peregrine

Flows other money is forced to tradeBoth fail.

Sep 30, 2026Peregrine

Liquidity, implied correlation and systematic 0DTE flowNo signal.

Sep 30, 2026Peregrine

Outside positioning data: SqueezeMetrics, CFTC, retail attentionAll three fail.

Sep 29, 2026Peregrine

Overnight carry, opening option flow and blending the booksAll three fail.

Sep 29, 2026Peregrine

Pin distance against a yes-or-no pin checkBoth cells fail. The simple check stays.

Sep 29, 2026Peregrine

Signed dealer pin and drift clusteringNeither passes. One lead stays open in a forward test.

Sep 23, 2026Peregrine

Dealer quotes, overnight flow and all-day order flowNo cell passes.

Sep 23, 2026Peregrine

Opening-auction imbalanceFails every criterion.

Sep 23, 2026Peregrine

Peregrine: four more ideas, pre-registeredNo cell passes.

Sep 23, 2026Peregrine

Peregrine: the close and co-movementNo cell passes. Holding to the close and flat sizing stay.

Sep 23, 2026Peregrine

Positioning, orthogonal data and prediction marketsNo cell passes.

Sep 23, 2026Peregrine

Round SPX levels as barriersThey are not barriers.

Sep 22, 2026Peregrine

A volatility-scaled trend triggerBoth scaled versions fail all three adoption criteria.

Sep 22, 2026Peregrine

An NDX veto at 09:35Fails four of five adoption rules, and the trades it would remove are profitable.

Sep 22, 2026Peregrine

Peregrine's direction and entry: four pre-registered studiesNo cell passes. Nothing changes.

Sep 22, 2026Peregrine

Strike depth, futures and the settlement exitClosed. Each of them fails.

Sep 20, 2026Peregrine

Six tests on 1-minute SPXW quotesA dealer-flow gate, a 09:35 range forecast and taking half off at +100% all fail.

Sep 19, 2026Phoenix

A deliberate double butterfly, and a late-entry gateBoth rejected.

Sep 19, 2026Peregrine

Dealer vanna, charm and vomma0 of 24 pre-registered cells pass.

Sep 19, 2026Phoenix

Implied over forecast realized vol as a Phoenix filterFails. Positive in 2021 to 2023, flat in 2024 to 2026.

Sep 19, 2026Peregrine

Peregrine through a debit verticalFails. It gives up the edge it was meant to keep.

Sep 19, 2026Phoenix

Peregrine's read as a side check on PhoenixUntestable. The two disagree on 2.4% of fires, too few to judge.

Sep 19, 2026Phoenix

The 09:35 dealer pin on PhoenixFails all three pass conditions. It helps Peregrine, not Phoenix.

Sep 17, 2026Peregrine

Grading the pin by distanceRejected on all three pre-registered clauses. The walk-forward is why.

Sep 17, 2026Peregrine

The gamma flip levelDead. The wrong sign on win rate in every construction, with intervals spanning zero throughout.

Sep 16, 2026Peregrine

A faster trend-onset detectorIt agrees with the current read and arrives later, not earlier.

Sep 16, 2026Peregrine

A volume clock instead of a wall clockIt does not arrive early enough, and volume pace sorts nothing.

Sep 16, 2026Research

Auction theory: following or fading acceptanceClosed. Following acceptance is right 50 to 52% of the time at every threshold. Fading it needs a regime that held in two of the last six years.

Sep 16, 2026Research

Auction theory: rejection off the developing value areaClosed. The continuation branch produces no tradeable move at any threshold.

Sep 16, 2026Peregrine

Dealer gamma at the open as a predictor of the day's rangeRejected on every pre-registered criterion.

Sep 16, 2026Peregrine

Opening pace against the implied moveRejected, and so was the control that looked alive.

Sep 16, 2026Peregrine

Pre-open activity as a way to sort daysOvernight volume and overnight range separate nothing.

Sep 16, 2026Peregrine

Strike choice, short holds and narrow-range daysNothing shipped: no strike change, no exit change, no narrow-range tag.

Sep 16, 2026Phoenix

The developing volume profile as a Phoenix inputClosed as an anchor, a sizing input and a directional read.

Sep 16, 2026Peregrine

The opening range and profile nodes as a filterClosed. A narrow opening range does predict a bigger day, but the read arrives too late to act on.

Sep 16, 2026Peregrine

Volume confirmation as a signal gateNet negative on SPX. Not added.

Sep 13, 2026Peregrine

Market breadth on the gated bookNull. Every interval spans zero, on win rate and on dollars.

Sep 13, 2026Peregrine

Momentum slowing at the moment of the callRefuted by seven independent checks. Nothing shipped.

Sep 11, 2026Peregrine

Gamma walls and the pin as trend filtersNo filter passed, and the wall filters lost money out of sample.

Sep 11, 2026Peregrine

Overnight /ES momentum as a head startNot shipped. Every version failed at least three of five pre-registered tests.

Sep 10, 2026Peregrine

An earlier, range-based entryRejected on the quote archive we already had. No new data was needed to see it fail.

Sep 10, 2026Peregrine

Calling reversals inside chopKilled. A small move against the trend inside a choppy hour is just chop: the reversal side loses over the next 30 minutes.

Sep 10, 2026Peregrine

Five-minute swing structure as an entryRejected. Both versions lose money on a realistic short hold priced ask to bid.

Sep 10, 2026Peregrine

Order flow as confirmation of the callAll coins in the first hour. Cumulative volume delta does not tell a continuing move from an exhausted one.

Sep 10, 2026Peregrine

Two replacement entry rulesNeither passed its pre-registered bar. The production entry stays.

Sep 8, 2026Peregrine

Exiting on a reversal off the session extremeKilled. No version made money out of sample.

Sep 8, 2026Peregrine

Reading the trend from the opening runKilled. It beat a placebo and nothing else: four of five ship criteria failed.

Sep 4, 2026Peregrine

Entering on a weaker read, with a managed exitRejected as an entry, whether held or managed.

Sep 2, 2026Research

Chart patterns as a read58 classical patterns scored out of sample: every one a coin. None clears 50% at the low end of its range.

Sep 2, 2026Peregrine

Intraday implied-vol changes as a filterDead. The primary test failed and nothing shipped.

Sep 2, 2026Peregrine

Trading Peregrine's call through a butterflyRejected. Every butterfly version lost to simply holding the wings.

Each study was written down before its data ran: the question, the test and the bar to pass. Results are hypothetical backtests unless a line says otherwise. Past performance, hypothetical or otherwise, is not indicative of future results.

Try Signal free for 7 trading days.

$0 today. Cancel before the trial ends and you pay nothing.