ES / MES · Sep 2016 – Aug 2026 · 120 months · P1 method books May 2019 – Aug 2026
Market Regimes
What a trading regime is, how to tell which one you are in without hindsight, what a P1 trade looks like in each, and ten years of the S&P futures sorted month by month.
What a regime is
A regime is a stretch of time where the market's behaviour stays roughly the same: how far it moves in a day, whether moves follow through or snap back, and which way it leans. Price is the output. The regime is the machine that produced it.
It matters because every method is a bet on behaviour. An opening-range breakout needs follow-through. A mean-reversion fade needs snap-back. Neither is good or bad in general. Each is good in some regimes and bleeds in others, and the same setup that made $2,400 in one month can lose $800 in the next without anything being wrong with the rules.
Regimes are not observable directly. You infer them from measurements. Every label on this page is a rule that can be stated and defended, not a law. Move the thresholds and a few edge months flip. The eras do not.
The three axes, and how many types that makes
Direction
monthly return · close vs 200-day average
Up: month return above +2.5%
Down: below −2.5%
Flat: in between
For the P1 methods this axis turns out not to matter.
Volatility
20-day realized vol, annualized · ATR as % of price
Quiet: under 10.2%
Normal: 10.2% – 15.5%
Turbulent: over 15.5%
Cuts are tertiles of ES 2011–2026. Turbulent means daily ranges roughly 1.5 to 2× a normal month. This is the axis that decides whether P1 wins or bleeds.
Trendiness
ADX(14) on daily bars · efficiency ratio
Trending: ADX above 25.8
Mixed: 18.5 – 25.8
Choppy: below 18.5
Independent of direction. A bear can trend, a bull can chop. Turbulent is not the same as choppy: of 32 turbulent months since 2019, 13 were trending.
Direction × volatility at three levels each gives nine cells. That is the grid used everywhere on these pages. Each cell shows what P1 Runner M3 averaged per trade in months of that type, May 2019 to Aug 2026, 532 trades.
Quiet
Normal
Turbulent
Up
Melt-upGrinding higher, tiny ranges. Breakouts stall short of target.+0.06 R / trade · 49
Healthy bullClean breaks with follow-through. The best cell for P1.+0.31 R / trade · 75
V-recoveryViolent up-moves off a low. Stops hit by noise.−0.23 R / trade · 108
Flat
CoilingCompression. Small R, but clean.+0.23 R / trade · 50
RotationTwo-way trade, decent ranges. Most common cell.+0.15 R / trade · 98
WhipsawBig ranges going nowhere.+0.15 R / trade · 55
Down
Slow bleedRare in index futures.+1.05 R · 4 trades, too few
CorrectionOrderly pullback inside a bull. P1 does fine.+0.16 R / trade · 37
Bear / crashTrend down with expanding ranges.−0.11 R / trade · 53
Read it by column, not by row. Down months are fine. Big, messy months are not. The other four P1 books (Master M2, Fast reBreak, Fast no-rearm, Fast RSI-off) show the same shape, with turbulent months around −0.15 R per trade each.
How to notice a regime in real time
Every regime measure is a rolling window, so it only confirms a change after the window fills. Tested against ten clean shifts into turbulence since 2019, this is how long each common detector took and how often it cried wolf.
Detector
Median lag (trading days)
False alarms per year
Verdict
20-day realized vol above 15.5%
19
2.9
Reliable, late. The right basis for a pause rule.
5-day realized vol above 15.5%
14
11.8
Faster, three to four times the false alarms.
3-day ATR above 1.3× 20-day ATR
10
9.6
Middle of the road, still noisy.
Today's range above 1.5× 20-day ATR
4
18.9
Fastest by far. Fires every 2–3 weeks for nothing.
Close below the prior 20-day low
18
5.4
Direction, not volatility. Misses shifts that happen above it.
So use two speeds:
Slow state, once a week. Compute 20-day realized vol, ADX(14), and close versus the 200-day. Place the month on the grid above. This tells you what to trade and how to read your P&L. Accept that it is about three weeks stale.
Fast tripwire, every evening. Today's range against the 20-day ATR, and the 3-day ATR against the 20-day. One spike means nothing. Two spikes in five days, or a spike with a close through the 20-day low or high, means the slow state is probably about to flip.
Sizing reacts daily, the label does not. Stop distance and contract count key off current ATR every day. Regime labels decide what to trade; ATR decides how much.
What it looks like on the tape before the numbers confirm it: overnight ranges widen, the opening range stops resolving cleanly, gaps stop filling, and the first pullback after a breakout runs past the entry instead of holding.
The 2022 trap. The one true bear in the window stayed turbulent for 107 straight trading days. Fast detectors keep firing inside a long regime, so once the slow state says turbulent, stop treating each spike as news and wait for the calm confirmation, which lags by about two weeks.
What a P1 trade looks like in each regime
Real trades from the P1 Runner M3 book on MES (sim fills, $0.74 per side commission). Each pair shows the kind of trade the regime produces, not a cherry-picked best case: one is the typical outcome, the other shows what the regime does to the winners.
Healthy bull · up + normal · +0.31 R / trade
The regime P1 is built for
2024-02-12 10:35 ET · long 5045.75 · stop 5043.79 (2 pt) · trailed out 5061.55 · +7.9 R · MFE 10.6 R
A tight opening range, a clean break, and a trend day that kept giving. A 2-point stop against a 16-point run is what normal volatility with follow-through looks like.
Healthy bull · the typical trade
Most trades are small losses even here
2024-09-04 15:16 ET · short 5520.75 · stop 5529.57 · stopped −1.0 R · MFE 0.5 R
Win rate in this book is 35%. The cell is profitable because the winners run, not because the losers are rare.
Rotation · flat + normal · +0.15 R / trade
Two-way trade, occasional runner
2026-02-11 09:45 ET · short 6999.75 · stop 7007.11 (7 pt) · trailed out 6950.31 · +6.7 R
No monthly direction, but individual days still trend. The stop is wider because ranges are bigger, so the same 50-point move is a smaller R.
Rotation · the typical trade
Stopped at −1 R
2025-11-26 10:01 ET · short 6803.00 · stop 6810.49 · stopped −1.0 R · MFE 0.5 R
Half a point of favourable excursion, then the stop. This is what "rotation" means at the trade level.
V-recovery · up + turbulent · −0.23 R / trade
The regime that bleeds
2022-03-04 09:58 ET · short 4316.00 · stop 4323.96 (8 pt) · stopped −1.0 R · MFE 0.0 R
Zero favourable excursion: the break reversed on the next bar. In turbulent months this happens often enough that the occasional +5 R does not cover it.
Bear / crash · down + turbulent · −0.11 R / trade
Even the winners are small
2020-02-04 10:13 ET · long 3289.25 · stop 3285.38 · trailed out 3299.95 · +2.7 R (best trade of the cell)
The best trade in 53 attempts was +2.7 R. Ranges are huge, so stops are wide and the trail gives back most of the move. Direction is not the problem; the size of the noise is.
Coiling · flat + quiet · +0.23 R / trade
Small stops, rare but big payoffs
2021-11-09 09:40 ET · short 4697.00 · stop 4699.24 (2 pt) · trailed out 4674.87 · +9.8 R
Quiet months compress the stop to a couple of points. When a compression breaks, the R multiple is enormous. It does not happen often.
Correction · down + normal · +0.16 R / trade
Down is fine when ranges are normal
2023-10-10 09:55 ET · long 4378.75 · stop 4374.42 · trailed out 4411.91 · +7.6 R
A long in a down month. The method does not care which way the month leans, only whether the day follows through.
Ten years, month by month
ES futures (the full-size contract, same index as MES; MES only exists from May 2019) from September 2016 to August 2026. Fill is the volatility tier, the glyph is direction (▲ up, ▼ down, ▬ flat), the small number is the month's return, the bar repeats direction. Hover a month for its numbers.
QuietNormalTurbulent▲ Up▬ Flat▼ Down
ES month-end closeQuietNormalTurbulentTurbulent month
Data. Regime metrics from the ES master (Databento ES.v.0, Jun 2010 – 14 Aug 2026 for the decade timeline; the last month is partial). Trade examples and cell averages from the MES P1 Runner M3 book, May 2019 – Aug 2026, engine fills (touch fills, no slippage). MES and ES agree within a tick on 98.8% of shared bars.
Thresholds are choices. Tertiles of 2011–2026 ES for volatility and ADX, ±2.5% for direction. Edge months flip if you move them; the eras do not.
Monthly labels hide intra-month breaks. April 2025 held a crash and a full recovery and gets one label.
Sim is not live. Every R figure on this page is a simulation result.
Built 2026-09-04 from data/es-1m.npz and db/backtests.db (run 374, variant 18438). Thresholds: vol tertiles 10.2 / 15.5, ADX tertiles 18.5 / 25.8, direction ±2.5%.