Day trading market conditions: measuring hot and cold small-cap markets
How many gappers there are, how much they trade and how many hold their gains change from week to week. Measure the regime with numbers known before the session, then compare your base rates within each regime.
Start in plain English
Small-cap traders talk about "hot" and "cold" markets: weeks when many stocks gap and run, and weeks when few do and most moves fade. This lesson shows how to turn that feeling into measurements you can record each morning, and why it matters: a base rate you measured in one kind of week may not apply in another. For a scalper, the same setup can look like an edge in a busy month and a cost in a quiet one.
The core idea: a market regime is a set of conditions you measure before trading, and your base rates belong to the regime they were measured in. The measurement needs to be knowable before the session. "This was a hot week", decided on Friday afternoon, is a description of the week, not a condition you could have used on Monday.
Two paths from the same start
All trades, counts and results in this section are Illustrative.
The plan, the same in both weeks. One written setup definition (for example, a 5-minute opening-range breakout on stocks gapping 20% or more), stop at the range low, target 2R, time stop 10 minutes, $50 risk per trade. Regime measure, recorded before each session: the average number of US stocks gapping 20% or more over the previous five sessions, from a daily gapper list.
Path AA hot week
The five-session average going into the week is 14 gappers a day. The setup triggers 38 times. Average result +0.28R per trade after costs, +10.6R in total (+$532).
A trader who measured only this week would write down a strong edge.
Path BA cold week
Three weeks later the five-session average is 5 gappers a day. The same setup triggers 12 times. Average result −0.35R per trade, −4.2R in total (−$210).
Nothing in the rules changed. Fewer candidates, thinner follow-through and the same costs per trade.
Path CAn average week
Between the two, the average is 9 a day. The setup triggers 22 times for +0.02R per trade, +0.4R in total (+$22).
Review note: "Plan followed in all three weeks; regime measured before each session; outcome A, B or C." One week of 12 to 38 trades is far too few to say the setup works in hot weeks and fails in cold ones; it says only that the question is worth counting. The rest of this lesson shows what to measure, what FennEdge's own data shows about month-to-month change, and how to keep regime labels honest.
What a regime measure looks like
A useful regime measure is defined in advance, available before the open, and counted the same way every day. Candidates for small caps:
- Gapper count. How many US stocks met a fixed gap definition yesterday, or on average over the last five sessions. Known before today's open.
- Today's premarket count. How many are gapping by 09:00 ET under the same definition. Known at 09:00, so usable for decisions after it.
- Small-cap volume. Total or median small-cap dollar volume over recent sessions.
- How many gappers held their gains. Share of yesterday's gappers that closed at or above their open. Known after yesterday's close, so it is usable today but not for yesterday.
- Halt count. Volatility halts in small caps over recent sessions.
Measures that are not usable before the session: today's count of stocks that "ran", today's share that closed green, and any label decided after seeing your own P&L. Write the measure, its source and its threshold into your plan once, and keep them fixed for at least a few weeks; a measure that changes whenever it disagrees with your feeling is no longer a measure.
What FennEdge's archive shows by month
The FennEdge Gapper HOD study measured when US stocks gapping 20% or more printed their 04:00–20:00 ET high. Its month split shows how much one measured quantity moved across a single quarter.
Two readings, both FennEdge computations from the Gapper HOD study data, not printed in the study (sessions 2026-07-01 to 2026-09-30; FennEdge archive; as of 2026-10-02):
- Count per session. Dividing by the number of exchange sessions in each month (22 in July, 21 in August, 21 in September, from the exchange calendar; the study does not publish monthly session counts) gives about 8.9, 12.9 and 9.1 qualifying gappers per session. August was the busiest month by this measure.
- How large the September shift is. The difference between August's 56.3% (n = 270) and September's 47.9% (n = 192) is 8.4 points (sessions 2026-08-03 to 2026-09-30; FennEdge archive; as of 2026-10-02; FennEdge computation). A rough 95% range for that difference runs from about −1 to +18 points (normal approximation). Observations on the same day are not independent, so the honest range is wider still. Three months are three data points about regime, not a trend.
The study itself warns that market conditions, news flow and the mix of tickers can shift these shares, and that it covers one stretch of 66 sessions.
Updating base rates by regime
Once each session carries a regime label, decided before the open, you can split your own log:
- Fix the label rule in advance. For example: "hot" if the five-session gapper average is 12 or more, "cold" if 6 or fewer, "normal" otherwise.
- Label every session before trading, including days you did not trade.
- Compute expectancy per regime in R after costs, with the trade count next to it.
- Compare only when each regime has enough trades. A split cuts your sample into smaller pieces, each with a wider interval.
A useful outcome of this exercise is often not "trade only in hot regimes" but "my trade count rises in hot regimes and my cost per trade does not fall". Overtrading looks at that side.
Observe vs. infer
A chat message on a Thursday reads: "Market's dead this week, nothing works, sitting out." Separate the measurable parts.
| What could be measured (observed) | What is inferred | Why it is not established |
|---|---|---|
| Gappers per day this week vs. the last five sessions | "The market is dead" | "Dead" needs a definition and a comparison; one week can be noise around a normal level |
| Your own R per trade this week | "Nothing works" | A handful of trades cannot separate a regime change from an ordinary losing streak |
| Halts and volume in your names | "This is how it stays now" | Regimes change; three months of FennEdge data shows a measure moving by several points month to month |
Sitting out can be a sound choice. The point is to make it from a measure written in advance, not from the last three trades.
What shifts the odds
These are hypotheses to record, not findings; no FennEdge study yet links regime measures to setup outcomes:
- Gapper count. More candidates may raise both opportunity and the temptation to trade lower-quality setups.
- Share of gappers holding gains. May relate to how often breakouts follow through.
- Timing of highs. In the HOD study, the premarket share of highs moved from 56.3% in August (n = 270) to 47.9% in September (n = 192; sessions 2026-08-03 to 2026-09-30; FennEdge archive; as of 2026-10-02); setups that depend on early highs may behave differently when that share shifts.
- Index and small-cap breadth. Broad market direction and small-cap breadth are separate measures; record both if you use either.
- Calendar effects. Holiday weeks and early-close days thin out volume.
Cost of being wrong
Two errors cost money here, using the Illustrative weeks above at $50 risk per trade:
| Error | What happens | Cost |
|---|---|---|
| Treating a hot-week base rate as permanent | Size up after Path A, then meet Path B at double size | 12 trades × −0.35R × $100 = −$420 |
| Treating one cold week as a regime change | Stop trading a setup after Path B and miss weeks like Path A | Forgone +$532 in a week like Path A, if the edge is real |
| Labelling regimes after the fact | "Hot" and "cold" chosen from your own P&L | The split looks clean and cannot be traded |
How many trades before this means anything
A regime split divides your sample. If you have 200 trades and they fall 90 / 70 / 40 across hot, normal and cold, the cold-regime estimate rests on 40 trades. With a per-trade standard deviation of about 1.2R, a rough 95% range for its average is ±0.37R (Illustrative, normal approximation), which is wider than most differences you are trying to see. Count weeks as well as trades: 40 trades from two cold weeks are two observations of "cold", not forty. How many trades before your numbers mean anything covers why correlated trades count for less.
What this lesson does not claim
- It does not claim any setup works in hot markets or fails in cold ones. The weekly results are Illustrative.
- The HOD monthly split describes the timing of highs over one quarter. It is not a regime model and not a forecast.
- A regime label is a condition to record, not a signal to trade or not trade.
- The per-session counts are our arithmetic from published figures and an exchange calendar, not numbers published in the study.
Go deeper
- FennEdge first.
- Top Gappers Calendar: past sessions' gappers. Guests, Free and Plus see the latest session at the default filters (top 20 rows, daily volume of at least 1 million shares); full history is Pro and Max. It is historical daily data, not real-time.
- Daily Top Gappers recaps (public, from 2026-10-01) list the top 10 and state how many symbols met the default filters each session, a ready-made gapper count for your regime sheet.
- Gapper HOD Timeline: when gappers printed their highs (guests, Free and Plus: latest completed session; earlier sessions need Pro or Max).
- Small Cap Volume: small-cap volume context (Pro and Max; guests, Free and Plus see an overview only). Tool access in this list is as of 2026-10-05.
- What time do small-cap gappers hit high of day?: the monthly split above, with its caveats.
- Time of day. Best time to day trade small caps is the within-day version of the same question.
- Backtests and regimes. Report backtest results by month; see Backtesting intraday ideas.
- Playbooks. Building a playbook records the regime a base rate came from, and the 100-trade edge review splits results by regime.
- Regime models. Hamilton (1989) introduced a statistical model where an economic series switches between regimes; trading regimes are a much looser use of the idea.
- Gap basics. US Day Trading Fundamentals covers gappers and gap and go.
Exercises
Five exercises follow this lesson: three concept checks (a knowable regime label; what the monthly split does and does not show; splitting a small sample), one scenario (reading a hot week and a cold week), and one practice task. Exercises, answers and progress are available after signing in.
Practice task: track a regime measure for 10 sessions. This is the fifth exercise (ungraded). Before the open on each of 10 sessions, record the previous session's count of gappers that met one fixed definition, from the Top Gappers Calendar (on the public view for guests, Free and Plus, count rows at 20% or more among the top 20, and write "20+" if all qualify). If you use the public daily recaps instead, they list only the top 10: count among those 10, write "10+" if all qualify, set your regime thresholds below 10 (a capped count cannot average more than 10), and do not mix the two sources in one log. Add a five-session average once you have five days, and the regime label from a rule you wrote before day one. Keep it in a separate sheet: this is a market log, not a trade log. If you also practise trades in those sessions, log them in the practice log (CSV) with the label set before the open in the regime column (for example hot) and the measure in conditions (for example gappers_5d=13.4). Keep both separate from the Trading Journal, which holds only real broker-synced trades.
Primary sources
- Hamilton, J. D. (1989). A New Approach to the Economic Analysis of Nonstationary Time Series and the Business Cycle. Econometrica 57(2).
- NYSE. Holidays and trading hours.
- FINRA. Frequent Intraday Trading: Understanding the Basics.
Sources checked 2026-10-05. The weekly trades, counts, regime labels and costs are fictional and Illustrative. Measured figures come from the FennEdge archive as stated. This lesson is education, not a recommendation to buy or sell any security. Past patterns do not indicate future results.