Methodology

How PremarketPlay builds the daily Playbook, freezes it before the open and grades it after the close. Bias is an informational assessment of premarket conditions. PremarketPlay is premarket planning only, and Standardised Outcomes are not executed trades or investment returns.

1. ScanAbout 2,000 liquid U.S. stocks, premarket data from 04:00 ET (consolidated SIP feed)
2. QualifyPremarket gappers and Day-2 continuations that clear the gap, activity and range rules
3. BiasBullish or bearish from premarket behaviour. No clear bias, not published
4. CatalystCompany news checked and summarised in our own words, or “no material news found”
5. Freeze09:00 ET: the plan is sealed with a SHA-256 hash and can't be edited
6. EvaluateAfter the close every call is graded by a fixed, published method, failures included

1. Selecting Stocks in Play

Each night PremarketPlay builds a universe of liquid U.S. common stocks (NYSE, Nasdaq and NYSE American; no OTC, warrants, rights, units, preferreds or most exchange-traded products), filtered by price and by 30-day average volume and dollar volume. From 04:00 ET it records every 1-minute premarket bar from the consolidated (SIP) feed.

Two setups can qualify a stock:

• Gap (premarket gapper): today's premarket price is gapping away from the previous close, with enough premarket volume relative to the stock's normal daily volume and enough premarket range.
• Day-2 (continuation): yesterday the stock gapped and closed near the extreme of its range in the gap direction. It is dropped if today's premarket is moving against that direction.

Qualifiers are ranked by an internal Play Score that combines premarket activity, gap size, premarket range and catalyst relevance. At most ten are published. The published rank is a way to prioritise your preparation. It is not a predicted probability, and the score itself is never shown. On a quiet day the Playbook can be short or empty (“No Stocks in Play met the criteria today”) rather than padded.

All prices are split-adjusted (not dividend-adjusted), so the previous close matches what you see on a chart.

2. Bullish or bearish bias

Every published stock has a bullish or bearish bias. For a gapper the bias follows the direction of the premarket gap. For a Day-2 continuation it follows yesterday's gap and close. A stock that is in play but has no clear direction is not published, so every published call can be evaluated. If one stock qualifies both ways, today's premarket wins.

The bias describes premarket conditions. It is not an instruction to trade and has no entry or exit attached.

3. Market context

Risk-on when SPY and QQQ are both up 0.30% or more in the premarket, Risk-off when both are down 0.30% or more, otherwise Mixed, measured at the freeze. For volatility we show VIXY's premarket change: VIXY is an exchange-traded fund that tracks short-term VIX futures, so it usually rises when the market expects more volatility. It is not the VIX index itself, and it is shown for information only: it is not part of the Risk-on/Risk-off rule.

4. Catalysts

For each candidate PremarketPlay reads company news published since the previous close and before the freeze. A language model classifies it (Earnings, Guidance, FDA, Analyst, M&A, Product, Legal, Offering, Macro/Sector, Other News) and writes a short summary in our own words. Code checks the summary: no copied phrasing, no speculative words, and every number must appear in the source. The original source is linked. When no company-specific news exists, the Playbook says so in fixed text. Heavy volume alone is never presented as news. The language model never produces prices, levels or any other number in the Playbook.

5. The 09:00 ET freeze

• From 08:00 ET members see a provisional list, rebuilt every 5 minutes and again just after the 08:30 economic releases. Provisional rows can change. They are kept for audit but never published in the archive or evaluated.
• At 09:00 ET the official plan is rebuilt from scratch from the 04:00–08:59 bars, checked so that no input timestamped after 09:00 can enter it, and frozen. Its canonical form is sealed with a SHA-256 hash, and from then on the database refuses any change to it.
• A plan frozen more than 60 seconds late is labelled late with its real time, never backdated. If no plan is frozen before the open, the day shows “No Official Playbook”.
• 09:29 ET “Final PM levels” (premarket high, low and last price) are captured for reference. They are informational, labelled separately and never change the plan or its evaluation.
• Corrections are published as separate, timestamped notes showing the original value, the new value and the reason. The original stays visible.

6. After the Bell outcomes

Each day becomes public at its session close. About 15 minutes later every published stock is graded with SIP data: the official open, high, low and close come from the consolidated daily bar, and the order of events from 1-minute bars. Outcomes show Pending until then.

Open Move

O = official open, H/L = regular-session high/low, C = close, d = +1 for a bullish bias and −1 for a bearish bias.

• Bullish: favourable = max(0, H − O); adverse = max(0, O − L)
• Bearish: favourable = max(0, O − L); adverse = max(0, H − O)
• Percent = 100 × amount ÷ O (favourable shown positive, adverse negative)
• Bias-adjusted open → close % = 100 × d × (C − O) ÷ O

Standardised Outcome (five-minute reference)

The reference is the completed 09:30:00–09:34:59 candle, captured at 09:35. In mechanical terms it acts as an entry at the candle's close and a stop at its opposite extreme. These are reference points for measurement only.

• 09:35 Reference E = close of the 09:30–09:34 candle
• Opening-range risk boundary S = that candle's low (bullish) or high (bearish)
• Reference risk 1R = r = E − S (bullish) or S − E (bearish), in dollars
• Only prices from 09:35 to the close count. Hpost/Lpost = their high and low.
• Bullish: MFE = max(0, Hpost − E), MAE = max(0, E − Lpost). Bearish: MFE = max(0, E − Lpost), MAE = max(0, Hpost − E)
• MFE % = 100 × MFE ÷ E; MAE % = −100 × MAE ÷ E; MFE R = MFE ÷ r; MAE R = −MAE ÷ r

MFE and MAE are full post-reference session excursions, including moves after the bias was invalidated, and they are labelled that way. R values are shown rounded toward zero to two decimals, so 0.9995R shows as 0.99R and never as a touch that didn't happen. Illustration, not historical performance: bullish bias, E = $213.00, S = $211.20, r = $1.80, later high $219.30 → MFE = $6.30 = 2.96% = +3.50R.

First-touch classification

For each later price P, q = d × (P − E) ÷ r. We record the first time q reaches +1, +2 and −1 (inclusive; a gap through a level counts at the first price beyond it).

1. Bias Invalidated: −1R is reached before +1R. Final, even if the stock later moves strongly in the bias direction.
2. Strong Follow-Through: +1R before −1R, then +2R before any −1R.
3. Followed Through: +1R before −1R, but not +2R before −1R or the close.
4. Limited/Mixed: neither +1R nor −1R by the close.

When one 1-minute bar touches both +1R and −1R, that minute's consolidated trades decide the order. If they can't, the row is shown as Limited/Mixed — first-touch order unresolved and flagged. We never assume the favourable order.

Data exceptions

An outcome is Unavailable, with the reason shown, when: the reference risk is zero or negative; there is no official open or no reference candle; a halt covers the reference window or two or fewer of its five minutes traded; a corporate action makes the prices inconsistent with the frozen previous close (more than 3% apart); the session high/low/close is missing; or less than half of the minutes after 09:35 have data. Gaps of five minutes or more without trades are listed as possible halts. Pending and Unavailable are never filled in with made-up results.

Every call is shown in its original rank, including those that failed. Nothing is hidden, collapsed or reordered.

7. Data and versions

Market data: the consolidated SIP feed (all U.S. exchanges), 1-minute bars, split-adjusted. Times are U.S. Eastern. If source data is corrected later, an outcome may be recalculated. Each recalculation is a new, logged revision with its reason, and the frozen plan is never touched.

Any change to the freeze time, bias rules, outcome rules or data resolution creates a new methodology version. Each day's page shows the versions it used.