SMB Capital•Closeout Review

CRWD Closeout

Pullback to the 50 DMA — carried to the exit
Entry 2026-07-31 @ 191.21  •  Last share out 2026-08-14 @ 216.95
Exit method three-stage ladder — 2R2RWhat it isTwice what you risked. If 1R is $2, then 2R of profit is $4 a share.Why it matters hereIt is the first milestone in the trade — the point where the management rules change gear.For exampleBuy at $100 with a stop at $98. 1R is $2, so 2R is reached at $104. banker, 17-MA trailTrail (trailing stop)What it isAn exit that moves up as the stock moves up. You do not pick a selling price in advance. You follow a line on the chart and get out when price closes below it.Why it matters hereIt means never having to guess where the top is. You stay in until the trend actually breaks.For exampleTrailing the 17-day average with the line at $210: you hold. Tomorrow the line is $212 and price closes at $211 — below the line, so you sell., 5-DMA runner
Result on the ladder +5.34R  •  Actually banked +2.00R
Pairs with crwd-2026-07-31.html (entry plan — frozen)
01
The Verdict
CLOSED Last share out 2026-08-14. The trade is flat and this document is the record of the completed cycle — base, entry, trend, turn, exit.
PeakPeakWhat it isThe highest price the trade actually reached, as opposed to the objective, which is the level aimed for before entry.Why it matters hereThe gap between the two shows how much a trend gave beyond expectations — and how much a mechanical exit hands back.For exampleCRWD's objective was $217.50. It peaked at $225.53 — about 8 points beyond target — before turning. (close)
+10.61RR (and 1R)What it isR is the amount of money you are risking on the trade — the gap between the price you paid and your stop. Every result is then measured in Rs instead of dollars.Why it matters hereIt puts every trade on the same scale, so a $20 stock and a $200 stock can be compared. It also means the rules work whatever the size of the account.For exampleBuy at $100, stop at $98 → 1R = $2. Run to $106 and you made $6, which is +3R. Hit the stop and you lost $2, which is −1R.
225.53 on 2026-08-13
Ladder result
+5.34R
all three stages
All-out on the 5-DMA
+7.96R
single-rule benchmark
Actually banked
+2.00R
no hold rule existed

What the missing hold rule cost

The entry plan closed with an open item: “Not fixed — hold rule still requires advancement (write before entry, not in the moment).” It was never written, so the position was managed by feel and closed at 2R.

The plan recorded the miss as 4.63R, measured when CRWD sat at 212.64. The trend did not stop there — it ran to 225.53. The true cost of the unwritten rule was 8.61R on this single position.

The ladder now written into this document would have captured +5.34R of it — more than double what was taken, without a single discretionary decision.

02
The Full Arc
CRWD daily — base through exit
180 190 200 210 220 230 entry 191.21 stop 187.97 2R 197.68 ENTRY T2 OUT 211.55 PEAK +10.6R T3 OUT +8.0R 07/20 07/23 07/28 07/31 08/05 08/10 08/13
5-DMA
10-DMA
20-DMA
entry / stop / 2R
DATA Daily bars: karb via kti price daily CRWD. 30-min bars: massive.com 1-min archive (4/01–8/05) spliced to the live kti price minute feed (8/04–8/18), resampled to RTH 30-min. The overlap reconciles to 0.0000. CRWD split 4:1 on 2026-07-02 and the daily feed is not back-adjusted — pre-split bars are divided by 4 in the generator so the 20/50-DMA stays continuous. Every figure on this page is computed, not transcribed.
DateCloseR5-DMA10-DMA20-DMA
2026-07-31190.86-0.11R183.47186.21192.25entry
2026-08-03202.54+3.50R187.96186.62192.41
2026-08-04211.22+6.19R193.84188.62193.24
2026-08-05209.86+5.77R199.94190.77194.18T2 out
2026-08-06207.39+5.00R204.37193.17194.63
2026-08-07214.42+7.17R209.09196.28195.99
2026-08-10225.16+10.49R213.61200.78197.85
2026-08-11221.90+9.49R215.75204.80198.41
2026-08-12221.78+9.45R218.13209.03199.16
2026-08-13225.53+10.61R221.76213.07200.25peak
2026-08-14216.95+7.96R222.26 ✗215.68200.94T3 out
2026-08-17213.90+7.01R220.01 ✗216.81 ✗201.71
03
Where It Turned

The top — 2026-08-13

Closed 225.53, +10.61R. Tenth consecutive session above a rising 5-DMA. Nothing in the tape said sell on this bar; the line was still beneath price and still rising.

The intraday high came the next session at 227.50 (+11.22R), then reversed hard and closed near the low. That reversal bar is the whole signal.

The exit — 2026-08-14

Opened 226.64, ran to 227.50, sold off to close 216.95 — a full outside-down day that closed below the 5-DMA (222.26) for the first time since entry.

Last share out at 216.95, +7.96R. No discretion required.

Why the fast line was right here

The instinct is that a 5-DMA trail is too tight and shakes you out of good trends. On this trade it did the opposite: it held through ten straight sessionsSessionWhat it isOne trading day — 9:30am to 4:00pm New York time.Why it matters hereSeveral rules refer to what happens inside a single day, such as the 3:50 check just before the close.For example“Six sessions basing” means six trading days, not six calendar days. Weekends do not count. without a violation, then fired on the first real distribution bar — two sessions after the top. The slower lines (17, 20, 50-DMA) never triggered at all and are marked at +7.01R as of 2026-08-17, below what the 5-DMA locked in, and still exposed. Slower is not safer once the trend has turned.

04
The Exit Ladder
ASSUMPTIONS Two inputs were not settled when this was written and are stated here rather than hidden. Stop = the hold lowHold lowWhat it isThe lowest price the stock reached during the few bars just before you bought.Why it matters hereIt is where the stop goes. If price drops below the level it was holding, the reason for buying has failed, so there is no point staying in.For exampleCRWD held above $187.98 for several bars before breaking out. The entry was $191.21 and the stop went at $187.98. 187.975 (risk 3.235); the Pine v5.2 literal min(vlLo, lowest(low,3)) gives 187.18 / risk 4.030 and reprices the ladder to +4.28R. Weights = 30/50/20 as fractions of the original position — the only reading that sums to 100%.
STAGE 1 - the banker
    price tags 2R (197.68)  ->  sell T1 = 30% of the position
    T2 + T3 (70%) remain at the original stopStopWhat it isThe price at which you accept the trade was wrong and get out. You decide it before you buy, not after.Why it matters hereIt is the single thing that keeps a losing trade small. Without one, a bad trade has no natural end.For exampleBuy at $100, stop at $98. If the stock trades down to $98 you sell and lose $2 a share. That is the most you were ever willing to lose.

STAGE 2 - the trend trancheTrancheWhat it isA slice of the position that has its own exit rule. Rather than selling everything at one moment, the shares are split into pieces and each leaves at a different time.Why it matters hereIt solves a real problem: you cannot both bank a certain profit and stay in for a huge move with the same shares. Splitting lets you do both.For exampleBuy 300 shares and split them 30% / 20% / 50% — 90 shares, 60 shares and 150 shares, each following a different exit line.
    on a 30-min CLOSE above 2R  ->  T2 = 50% begins trailing
    exit T2 on the first 30-min close below the 17-period MA

STAGE 3 - the runner
    same arming bar  ->  T3 = 20% begins trailing
    exit T3 on the first DAILY closeClose  (and why it must be a close)What it isThe price at the very end of a bar. A 30-minute bar closes every half hour; a daily bar closes at 4pm; a weekly bar closes on Friday.Why it matters hereEvery rule here waits for a close. Prices poke above and below levels constantly and most of those pokes mean nothing. Waiting for the bar to finish filters out the noise.For exampleIf the stock dips under the line at 11am but finishes the day above it, nothing has happened. You do not act. below the 5-DMA

PROTECTING PROFITS - the pre-2R state
    price reaches 1R (194.45)  ->  T1's stop moves from the original stop
    to the 17-period MA trail on 30-min. T2 + T3 stay at the original stop.
    T1 then leaves via whichever comes first: the 2R target or that trail.
    NOT EXERCISED ON THIS TRADE - 1R and 2R were both tagged inside 90
    minutes on 08-03, so T1 never trailed. This clause is unvalidated here.

TRAIL SEMANTICS
    exit on a CLOSE below the line, on that line's own timeframe.
    intraday violations do not count. matches the call: "we're waiting
    for a close beneath the fiveday".
05
Ladder Result
WhenEvent
2026-08-03 09:301R tagged — T1 switches to the 17-MA trail
2026-08-03 11:002R tagged — T1 sold 197.68
2026-08-03 11:3030-min close above 2R — T2 arms the 17-MA, T3 arms the 5-DMA
2026-08-05 15:00T2 trailed out 211.55 — 17-MA at 212.05
2026-08-14T3 trailed out 216.95 — 5-DMA at 222.26
TrancheWeightPriceRContributionReasonWhen
T130%197.68+2.00R+0.600R2R target2026-08-03 11:00
T250%211.55+6.29R+3.144R30-min close < 17-MA (212.05)2026-08-05 15:00
T320%216.95+7.96R+1.591Rdaily close < 5-DMA (222.26)2026-08-14
NET100% +5.335R blended across the ladder

Against the benchmarks

+5.34R on the ladder  •  +2.00R flat at 2R (what was taken)  •  +7.96R all-out on the 5-DMA.

The ladder is a 167% improvement on what was actually banked. It also lands below the simplest possible rule — hold everything, exit on one daily close under the 5-DMA. The next section is why.

06
Where It Leaks

The weights and the speeds are inverted

A 17-period MA on 30-min bars is 1.3 trading days (17 × 30min ÷ 390min). That is roughly four times faster than the 5-DMA. So the ladder currently assigns the largest tranche the tightest leash:

TrancheWeightTrailEffective lookback
T130%fixed 2R—
T250%17-per / 30-min 1.3 days ← fastest
T320%5-DMA / daily5.0 days ← slowest

80% of the position was flat by 08-05 around 211, while the trend paid through 08-13 at 225.53. Only the 20% runner was still on for it.

T2 trail-speed sweep

30-min MA length≈ daysExitWhenT2 RNet 30/50/20
131.0211.552026-08-05 15:00+6.29R+5.335R
171.3211.552026-08-05 15:00+6.29R+5.335R
262.0209.872026-08-05 15:30+5.77R+5.075R
342.6202.292026-08-06 09:30+3.43R+3.904R
483.7202.292026-08-06 09:30+3.43R+3.904R
655.0220.092026-08-14 10:30+8.93R+6.654R
917.0218.472026-08-14 14:00+8.43R+6.404R

It is not monotonic. There is a dead zone around 34 periods (2.6 days) where the trail exits into the bottom of the 08-05/08-06 consolidation at 202.29 (+3.43R). Faster gets out above it; slower rides straight through. The middle is the worst of both.

One nuance worth keeping: at 65 periods — 5.0 days, the same lookback as the 5-DMA — T2 exits at 220.09, while T3's daily-close 5-DMA exits at 216.95. Same MA length, checked on 30-min closes instead of daily closes, got out roughly 3 points higher on the reversal day.

Weight sweep — T2 held at the spec 17-period trail

T1 / T2 / T3Net R
30 / 50 / 20+5.335R
30 / 35 / 35+5.585R
30 / 20 / 50+5.836R
20 / 20 / 60+6.432R
READ THIS This is one trade. Sweeping parameters against a single observation is curve-fitting, and no number in the two tables above is evidence. What the sweep establishes is the mechanism — the middle tranche carries the most size on the tightest stop — which is structural and will hold across trades. The specific lengths and the dead zone are properties of this tape until the protocol on the next page says otherwise.
07
Backtest Protocol

What has to be true before this is a validated play rather than one good chart.

Already built — reuse it

What has to be added

The dataset problem — read before running anything

The example set (ALOY, BTDR, BBIO, IBIT, JNJ, EBAY, CRWD — roughly 18 legs) is entirely composed of setups that worked. Thresholds fitted to winners alone will fit all 18 and predict nothing, because the same measurements on the failures were never observed.

ma_state.py already encodes the right discipline — it prints winners and losers and requires n ≥ 30 per bucket. Feeding it a winners-only set defeats it.

Required input: the pocket-pivot scanner history — every name that fired the setup, including the ones that failed. That is the missing half of the dataset and the gating item for validation.

08
Open Items
ItemStatusEffect
Stop convention — hold low vs the Pine literal min() unresolved reprices every R by ~25% (+5.34R vs +4.28R)
Engine ports v5.2; the Pine on disk is v6.0-swing drift backtest results do not transfer until re-ported
Loser set from the pocket-pivot scanner neededblocks the entire gate validation
Per-trade risk $ (separate from the daily loss cap) neededshare sizing; R math is unaffected
Tranche split — 30/50/20 vs 30/35/35 ambiguous+5.34R vs +5.59R
1R protecting-profits clause untestednever fired on this trade
Earnings policy for an open runner undecidedCRWD reports ~8/25
T2 trail speed — largest tranche on the fastest line hypothesismechanism is structural; length is not evidence
Exit fires without discretion settled—
Sell 30% at 2R settled—
Trail semantics — close below the line, own timeframe settled—
NOTE At entry on 2026-07-31, close was 190.86 while the 17-DMA sat at 191.76 and the 20-DMA at 192.25 — price was below both. A “must be above the 20-DMA” entry filter would have blocked this trade outright. A pullback-to-the-50 setup is below the faster lines by construction. Do not add that filter.

Plain-English Glossary

Every specialist term used on this page, explained from scratch. The same text pops up wherever a term appears — hover it, or tap it on a phone.
2RWhat it isTwice what you risked. If 1R is $2, then 2R of profit is $4 a share.Why it matters hereIt is the first milestone in the trade — the point where the management rules change gear.For exampleBuy at $100 with a stop at $98. 1R is $2, so 2R is reached at $104.
ATRWhat it isShort for Average True Range. It is simply how far a stock normally moves in one bar. A big number means a jumpy stock; a small number means a calm one.Why it matters hereIt lets us compare a $20 stock with a $200 stock fairly. “It moved $3” is meaningless on its own; “it moved one ATR” tells you whether that is normal for that stock.For exampleIf a stock's ATR is $2, a $2 move today is completely ordinary. A $6 move is three times normal.
Close (and why it must be a close)What it isThe price at the very end of a bar. A 30-minute bar closes every half hour; a daily bar closes at 4pm; a weekly bar closes on Friday.Why it matters hereEvery rule here waits for a close. Prices poke above and below levels constantly and most of those pokes mean nothing. Waiting for the bar to finish filters out the noise.For exampleIf the stock dips under the line at 11am but finishes the day above it, nothing has happened. You do not act.
Hold lowWhat it isThe lowest price the stock reached during the few bars just before you bought.Why it matters hereIt is where the stop goes. If price drops below the level it was holding, the reason for buying has failed, so there is no point staying in.For exampleCRWD held above $187.98 for several bars before breaking out. The entry was $191.21 and the stop went at $187.98.
PeakWhat it isThe highest price the trade actually reached, as opposed to the objective, which is the level aimed for before entry.Why it matters hereThe gap between the two shows how much a trend gave beyond expectations — and how much a mechanical exit hands back.For exampleCRWD's objective was $217.50. It peaked at $225.53 — about 8 points beyond target — before turning.
R (and 1R)What it isR is the amount of money you are risking on the trade — the gap between the price you paid and your stop. Every result is then measured in Rs instead of dollars.Why it matters hereIt puts every trade on the same scale, so a $20 stock and a $200 stock can be compared. It also means the rules work whatever the size of the account.For exampleBuy at $100, stop at $98 → 1R = $2. Run to $106 and you made $6, which is +3R. Hit the stop and you lost $2, which is −1R.
SessionWhat it isOne trading day — 9:30am to 4:00pm New York time.Why it matters hereSeveral rules refer to what happens inside a single day, such as the 3:50 check just before the close.For example“Six sessions basing” means six trading days, not six calendar days. Weekends do not count.
StopWhat it isThe price at which you accept the trade was wrong and get out. You decide it before you buy, not after.Why it matters hereIt is the single thing that keeps a losing trade small. Without one, a bad trade has no natural end.For exampleBuy at $100, stop at $98. If the stock trades down to $98 you sell and lose $2 a share. That is the most you were ever willing to lose.
Trail (trailing stop)What it isAn exit that moves up as the stock moves up. You do not pick a selling price in advance. You follow a line on the chart and get out when price closes below it.Why it matters hereIt means never having to guess where the top is. You stay in until the trend actually breaks.For exampleTrailing the 17-day average with the line at $210: you hold. Tomorrow the line is $212 and price closes at $211 — below the line, so you sell.
TrancheWhat it isA slice of the position that has its own exit rule. Rather than selling everything at one moment, the shares are split into pieces and each leaves at a different time.Why it matters hereIt solves a real problem: you cannot both bank a certain profit and stay in for a huge move with the same shares. Splitting lets you do both.For exampleBuy 300 shares and split them 30% / 20% / 50% — 90 shares, 60 shares and 150 shares, each following a different exit line.