15.11 Playbook 11: an MCX crude-mini trend trade sized for the gap (CRUDEOILM, 16 October – 2 November 2026)

15.11.1Prerequisites and what this playbook teaches#

Read first (essential) M10 · 10.14–10.15 (MCX costs, tax, contract choice, margins, the evening session, expiry), M10 · 10.10 (the contracts), M10 · 10.8 (crude oil), M03 · 3.1.1 (how many trades distinguish edge from noise) and 3.9 (sizing), M04A · 4A.10 (stops are not moved), M15 · 15.9.2 (gap-adjusted sizing (15.9.1)). For depth: M10 · 10.3 and Problem 10.23 (roll yield and roll gaps in continuous series), M10 · 10.11.1 (negative WTI).

What it teaches

1. How to trade an MCX contract when MCX's own price history is not freely available. Crude is the cleanest case: MCX crude's final settlement price is the NYMEX WTI settlement times the RBI reference rate (M10 · 10.10), so WTI times USDINR is a close proxy for the MCX price.
2. Why the contract you trade is not the continuous series you test. Rolls change the answer.
3. How a ₹10 lakh account chooses between CRUDEOIL and CRUDEOILM, and why gold mini does not fit it at all.
4. What a trend-following rule's losing streak looks like from the inside: two full stop-outs in two weeks, one of them after the position had been up more than 2R.
5. The MCX cost line (CTT, MCX charges) and the overnight and weekend gap that a stop cannot catch.

The trades below are paper trades on a proxy that lost about 1% of capital. They are an educational reconstruction on public data, not a recommendation to buy or sell crude oil or any contract.

Instrument and proxy

• The contract. CRUDEOILM, MCX's crude oil mini future: 10 barrels a lot, quoted in rupees per barrel, tick ₹1 (₹10 a lot), cash-settled on NYMEX WTI. Reported to be still listed after its relaunch at 10 bbl in March 2023 [S]. MCX's November 2026 contract expires about one business day before NYMEX's December contract (M10 · 10.15.3), around 19 Nov 2026; check the date in MCX's expiry calendar.
• The proxy price. $P_t = W_t X_t$, where $W_t$ is the Yahoo daily bar of NYMEX's December 2026 WTI contract (CLZ26.NYM) and $X_t$ Yahoo's USDINR close (INR=X), fetched 6 Nov 2026. The proxy is good at the close: MCX closes at 23:30 IST (23:55 from 2 Nov), close to NYMEX's settlement at 14:30 New York time.
• Where the proxy is weak. It is worse for the open, because Yahoo's daily open is the CME Globex reopening at 18:00 New York time, about five hours before MCX's 9:00 IST open. It is also worse for the high and low, because NYMEX trades almost round the clock and an extreme printed while MCX is shut never reaches an MCX stop. Treat every intraday fill below as approximate.

Why not gold mini? On 6 Nov 2026 the duty-paid gold parity was ₹1,41,448 per 10 g, and its 20-day mean absolute daily change was ₹1,764.7. A 2N stop on one GOLDM lot (10 units of 10 g) risks $2 \times 1{,}764.7 \times 10 = \text{₹}35{,}294$, 3.5% of a ₹10 lakh account. The 1% rule allows zero lots. GOLDTEN (10 g; M10 · 10.10) would allow two. Contract size, not analysis, decides what a small account can trade.

15.11.2The setup as testable rules#

The rule is the entry and exit of the Turtle traders' "System 1", as described by Faith (2007, Way of the Turtle), adapted to closing prices and to the course's 1% risk limit. The parameters are the published ones, not optimised here.

1. Data. Signals use the rupee proxy of the contract traded. For the base rates, which need 2021–2026 and therefore a continuous series, the proxy is Yahoo's front-month CL=F times USDINR, back-adjusted at each NYMEX expiry: the gap between the old contract's last close and the new contract's first open is removed (M10 Problem 10.23). Data from 2020 are used only as warm-up. The negative settlement of April 2020 (M10 · 10.11.1) makes a rupee-sized rule ill-defined.
2. Volatility. $N_t$ is the mean absolute close-to-close change of $P$ over the last 20 sessions, in ₹/bbl. Faith's $N$ is an average true range; closes are used here because the proxy's highs and lows are the weakest part of it.
3. Signal (after the MCX close). Long if $P_t$ closes above the highest close of the previous 20 sessions; short if below the lowest.
4. Entry. At the next MCX morning open (proxy: the next Yahoo open times $X_t$).
5. Stop. $2N$ from entry, as a resting stop-loss order. If the market opens beyond the stop, the fill is the open.
6. Exit. A close below the lowest close of the previous 10 sessions (for a long) exits at the next open.
7. Expiry. Never hold into the last two sessions before MCX expiry: roll to the next month two sessions before expiry, paying a second round trip. Crude is cash-settled, so there is no tender risk; the rule avoids settlement-day risk (M10 · 10.15.3).
8. Size. Fixed-fractional with the gap cap of (15.9.1), per lot of 10 bbl: with $d = 2N \times 10$ the risk per lot, $P$ the entry price and $g_{99}$ the 99th percentile of adverse overnight gaps $1 - O_t/C_{t-1}$ over the previous five years (roll days excluded),
   $$Q = \min\left(\left\lfloor\frac{fE}{d}\right\rfloor,\ \left\lfloor\frac{1.5\,fE}{d + 10\,g_{99}P}\right\rfloor\right),\qquad f = 1\%,\ E = \text{₹}10{,}00{,}000. \tag{15.11.1}$$
9. Re-entry. After an exit, a new signal is a new trade.

15.11.3The scan and the episode#

Base rates first (continuous back-adjusted proxy, entries 6 Jan 2021 to exits 3 Nov 2026, both directions, costs of 15.11.4 including one extra round trip per roll):

	Value
Trades	73 (12.5 a year)
Win rate	26.0%
Average win / average loss	+2.88R / −0.92R
Mean / standard deviation (net)	+0.069R / 2.17R ($t = 0.27$)
Break-even win rate	24.2%
Longs / shorts	36 trades, +0.65R / 37 trades, −0.50R
Three best trades	+24.65R, against a total of +5.0R; without them the mean is −0.28R
Longest losing streak	14 trades
By year of entry (total R)	2021 +18.4; 2022 −7.9; 2023 −1.3; 2024 −2.8; 2025 −14.3; 2026 +12.9

This is the shape of trend following: most trades lose a little, a few win a lot, and the record is decided by those few. Over six years it is indistinguishable from zero.

The episode, chosen by the rule The MCX November contract became the one to trade when the October contract expired, around 19 Oct 2026. On the December-WTI proxy, the rule signalled long twice in that contract's life before 6 Nov:

Signal day	Close $P_t$	Previous 20-day high	$N_t$	Notes
Fri 16 Oct 2026	₹9,359.85	₹9,182.10	₹136.04	WTI Dec $99.02, USDINR 94.52
Thu 22 Oct 2026	₹9,381.24	₹9,359.85	₹118.94	after the first trade was stopped

Calendar inside the window (M10 · 10.15.2):

• EIA weekly reports on Wednesdays 21 and 28 Oct, at 20:00 IST, inside MCX hours.
• The FOMC decision on 28 Oct at 14:00 New York time, which is 23:30 IST, the moment MCX closed.
• US daylight saving ended on 1 Nov, and MCX's close moved to 23:55 from 2 Nov.
• An OPEC+ meeting on Sunday 1 Nov 2026 held December production targets at September levels (OPEC press release of 1 Nov 2026), so news arrived while MCX was shut.

15.11.4The plan before entry#

Written on the evening of 16 Oct 2026 after the MCX close.

Thesis A 20-day closing high in crude, after a four-week range, is the start of a trend more often than chance would explain in the instruments the Turtles traded. Whether that holds in crude in 2021–2026 is exactly what the base rates above fail to show. This is a probationary trade (M14 · 14.10.3), taken to practise the process and compare live behaviour with the backtest, not because an edge has been demonstrated.

Entry and stop Buy at the 19 Oct open; proxy fill ₹9,406.59 (WTI Dec $99.04 × 94.98). Stop at $9{,}406.59 - 2 \times 136.04 = \text{₹}9{,}134.51$, entered as a stop-loss order immediately after the fill.

Size by (15.11.1)

• Risk per lot $d = 2 \times 136.04 \times 10 = \text{₹}2{,}720.85$, so the risk term allows $\lfloor 10{,}000/2{,}720.85\rfloor = 3$ lots.
• $g_{99} = 3.41\%$ (adverse proxy gaps over five years; the largest was 11.3% on 10 Mar 2026, during the Hormuz shock of M10 · 10.11.5). The gap term allows $\lfloor 15{,}000/(2{,}720.85 + 10 \times 0.0341 \times 9{,}406.59)\rfloor = \lfloor 2.53 \rfloor = 2$ lots.
• So 2 lots (20 bbl, notional ₹1,88,132), with a planned loss at the stop of ₹5,441.70 (0.54% of capital). The gap term binds.
• One CRUDEOIL lot (100 bbl) would have risked ₹27,208, 2.7% of the account.

Margin At MCX's minimum initial margin for crude of 30% (from 18 May 2026 [S]; M10 · 10.15.1), 2 lots need about ₹56,440, 5.6% of the account. Keep at least as much again free for a margin increase.

Costs (M10 (10.28), rates as of 6 Nov 2026) Brokerage ₹20 an order (illustrative), MCX ₹2.10 per lakh each side, SEBI ₹10 per crore, stamp 0.002% on the buy, CTT 0.01% on the sale, GST 18%. For this round trip: ₹78.86, of which brokerage ₹40.00, CTT ₹18.27, MCX charge ₹7.79, GST ₹8.67, stamp ₹3.76, SEBI ₹0.37. That is 0.014R.

Expiry Exit or roll by 17 Nov 2026 (two sessions before the expected 19 Nov expiry; confirm on MCX).

Tax Non-speculative business income (M10 · 10.14.2); a loss can be set off against other non-salary income or carried forward 8 years.

15.11.5Day-by-day management#

Proxy bars, ₹/bbl (open / low / close; open and low approximate, see 15.11.1). Lot = 10 bbl; 2 lots held.

Date	O / L / C	Open P&L (2 lots)	Action and notes
Fri 16 Oct	— / — / 9,359.85	—	Signal after the close. Plan written.
Mon 19 Oct	9,406.59 / 9,186.24 / 9,359.10	−₹950	Bought 2 lots at the open; stop at 9,134.51. The MCX October contract expired around today.
Tue 20 Oct	9,332.17 / 9,048.14 / 9,292.27	—	Proxy low 9,048 below the stop. Stopped at 9,134.51: −₹5,441.70 gross, −₹5,520.56 net (−1.01R).
Thu 22 Oct	— / — / 9,381.24	—	New 20-day closing high. Second signal; $N = 118.94$. Plan: stop $2N$, 2 lots (risk term 4, gap term 2).
Fri 23 Oct	9,385.89 / 9,361.26 / 9,780.84	+₹7,899	Bought 2 lots at 9,385.89; stop 9,148.01 (risk ₹4,757.60, 0.48% of capital).
Mon 26 Oct	9,741.59 / 9,525.13 / 9,660.89	+₹5,500
Tue 27 Oct	9,641.49 / 9,465.40 / 9,509.18	+₹2,466
Wed 28 Oct	9,533.96 / 9,493.99 / 9,782.37	+₹7,930	EIA at 20:00 IST; FOMC statement at the 23:30 close.
Thu 29 Oct	9,707.56 / 9,500.96 / 9,711.37	+₹6,510
Fri 30 Oct	9,781.67 / 9,746.46 / 9,907.27	+₹10,428 (+2.19R)	Best close. The 10-day exit level is 9,201.70 and the stop 9,148.01; the rule does not tighten either. Position held over the weekend.
Mon 2 Nov	9,717.32 / 9,030.21 / 9,154.54	—	Proxy open −1.9% after the OPEC+ weekend; the fall continued in the session through the stop. Stopped at 9,148.01: −₹4,757.60 gross, −₹4,836.54 net (−1.02R).

Crude kept falling after the exit: on Wednesday 4 Nov, the EIA day of Playbook 9, the proxy closed at ₹8,349.23, 8.7% below the stop.

15.11.6Exit and result#

Trade	Entry	Exit	Gross ₹	Costs ₹	Net ₹	Net R
1 (signal 16 Oct)	19 Oct, 9,406.59 × 20 bbl	20 Oct, stop 9,134.51	−5,441.70	78.86	−5,520.56	−1.01
2 (signal 22 Oct)	23 Oct, 9,385.89 × 20 bbl	2 Nov, stop 9,148.01	−4,757.60	78.87	−4,836.54	−1.02
Total			−10,199.30	157.73	−10,357.10	−1.04% of capital

The costs were 1.5% of the loss. On an NSE equity future of the same size the STT alone would have been about ₹91 a trade (0.05% of the sale), more than the whole MCX cost line of ₹79. CTT is the reason MCX costs are low. The loss is non-speculative business income (M10 · 10.14.2).

15.11.7Post-mortem#

Was the process followed? Yes:

• size from the fill, with the gap term binding both times;
• stops entered at once and never moved (M04A · 4A.10), including on 30 Oct, when the trade was up 2.19R and the stop was still 6.4N below the close;
• re-entry on the second signal, as the rule says.

Each loss was the planned loss plus about 1.5% in costs, and each was about half of 1% of capital because the gap cap halved the size.

Was the rule wrong to give back 2.19R? That is the rule. System 1 exits on a 10-day low, not on a trailing stop, and its average winner (+2.88R) is paid for by trades like this one. A rule that locked in profits after +2R would be a different rule. Its base rates must be computed over all 73 trades before anyone adopts it; picking it because of this episode is the selection error of M03 · 3.1.1.

The roll lesson The base-rate backtest uses the continuous front-month series. For the 16 Oct signal, that series held the November NYMEX contract until its expiry on 20 Oct.

• On 19–20 Oct the November contract traded $3–4 above December (CL=F closes $103.66 and $101.79 against CLZ26's $98.54 and $97.82).
• In the backtest the same signal entered at ₹9,294.79 (back-adjusted), was never stopped, and exited on the 10-day rule at the 3 Nov open for −0.54R net.
• In the contract an MCX trader actually held, it was stopped on day two for −1.01R.

Neither number is wrong; they answer different questions. A backtest on a continuous series answers "what would a rolled front-month position have done?", which is close to but not the same as the trade you place. Problem 15.22 works through it.

How many trades would be needed? With the base rates $E = 0.069$R and $s = 2.17$R, (3.2a) gives $N = (2s/E)^2 \approx 4{,}000$ trades for $t = 2$, and (3.2b) about 8,100 for 80% power. At 12.5 trades a year that is over three centuries. As with Playbook 8, the rule is a hypothesis, not a strategy. The defensible reason to run it is diversification, not proof of edge: across many markets, trend-following returns have been reported to be positive and weakly correlated with equities (Hurst, Ooi and Pedersen 2017, A Century of Evidence on Trend-Following Investing). One market over six years cannot test that claim.

What the proxy hides

1. The open. The proxy open is five hours earlier than MCX's 9:00 open. On 2 Nov the proxy opened at ₹9,717; the actual MCX open would have reflected five more hours of trading.
2. The low. The 20 Oct stop-out rests on a proxy low (₹9,048) that may have printed while MCX was shut. If it did, an MCX stop at ₹9,134.51 would not have been hit that day. The 2 Nov fill at ₹9,148.01 rests on the same assumption. Conclusions about intraday stops are the most fragile part of any test on daily proxy data.
3. USDINR timing. One closing rate is applied to all four prices of the day.

Anyone running this rule live should log MCX's own fills and compare them with the proxy for the first 30 trades before trusting the proxy backtest.

What the episode does teach The MCX execution chain works as M10 · 10.14–10.15 describes:

• a mini contract that fits a 1% risk budget;
• margin of about 6% of the account;
• a cost line dominated by brokerage and CTT;
• an evening session that contains the EIA release and ends at the FOMC statement;
• a weekend gap after an OPEC+ meeting that no stop could have prevented.

The gap cap is what kept two full stop-outs to about 1% of capital in total.

15.11.8Do it in the simulator#

The simulator trades NSE cash and F&O only. It has no MCX contracts, so this playbook is run as a paper trade in the journal.

1. Scan. After each MCX close, compute the proxy $P_t = W_t X_t$ for the MCX contract you would trade (the NYMEX contract one month later; check MCX's expiry calendar), its 20-day high and low closes and $N_t$. Use the lab below or your own script. Record the values in the journal under Journal → New entry, whether or not there is a signal.
2. Plan. On a signal, write the plan of 15.11.4: entry estimate, stop at $2N$, $g_{99}$, both terms of (15.11.1) and which binds, margin at the current MCX rate, the expiry and roll date, and the US releases and OPEC+ dates inside the likely holding period (M10 · 10.15.2).
3. Fill. Next morning, record the MCX open from the exchange's website as your fill, and the proxy open beside it.
4. Manage. Each evening, check the stop against MCX's actual low and the 10-day exit against the close. Do not move the stop.
5. Review. At exit, compute net R with the cost line of 15.11.4 and add the trade to your own base-rate table. Note the difference between MCX's prices and the proxy's on entry, exit and stop days.

There is no MCX page on the course's live-data site. The NSE alternatives for commodity exposure are the gold and silver ETFs at https://oicharts.com/etfs/gold and https://oicharts.com/etfs/silver (Playbook 8).