Targets, Position Management, and the 0DTE Overlay

The stop is the price at which you are wrong. The target is the price at which you are right, and it has to exist before you enter — written down, with the first limit order resting at the moment the position opens. Not because discipline is a virtue in itself, but because the impulse is over in about a minute and you will not make a good exit decision inside it. A target chosen while watching an open P&L is chosen by the part of you that holds winners too long and cuts losers too small.

Targets are the obstacle list, read a second time

Lesson 9 had you inventory everything between your trigger and the horizon to answer one question: can I trade this? The same inventory answers a different one here: where do I get out? Three destinations, derived at arming time alongside the stop.

T1  first micro liquidity     nearest 1m swing high, the opening-range high,
                              a round number within reach

T2  first real structure      session high, prior-day close or high, a 5m swing,
                              the opposing edge of a fair value gap, VWAP

T3  expansion                 the next major structural level; a measured move
                              only when nothing structural exists above

T1 carries the most weight, and the reason sits in the data rather than the theory. Across 489 breakout attempts on 490 SPY sessions, 75% closed back inside the range they broke. Among those that failed, the median best price reached was still 0.16W at three minutes and 0.24W at ten, where W is the width of the broken range — a median of 1.00 point on SPY.

That is the argument for a conservative T1: it is the only target that fires on the modal outcome. Set it to fill, not to maximise — the maximising happens at T2 and T3, on whatever survives.

Structure targets beat fixed percentages for a mechanical reason. "Take profit at plus thirty per cent" ignores where the move can go, does not adapt to the day's volatility, and has no relationship to your stop — so your risk-to-reward varies at random from trade to trade and your expectancy becomes unmeasurable. The correct order runs the other way: find T1 and T2 on the underlying, convert them to expected premium through delta, and use the premium figure as a sanity check rather than as the trigger.

Risk budget first, contract count second

The ordering is the whole discipline. You do not decide how many contracts to buy. You decide how much you are willing to lose, and the count falls out.

1  risk_budget       = account equity × risk per trade
2  stop_distance     = | entry − structural stop |
3  risk_per_contract = stop_distance × delta × 100
4  budget_contracts  = floor(risk_budget / risk_per_contract)
5  contracts         = min(budget, hard cap, liquidity cap, concentration cap)

Step five surprises people. A tight structural stop on a near-the-money contract produces a small premium risk per contract, so the budget will support a position far larger than you should hold. On that kind of setup the hard cap binds, not the budget — know which of the two is doing the work.

Harvest, protect, participate

Three phases, and every playbook is an implementation of them.

Harvest sells into the impulse, before resolution. It needs no evidence beyond the move itself, and it fires on winners and failures alike. Protect sells again once the impulse ends and acceptance starts forming, which cuts exposure heading into the retest — the highest-uncertainty moment left. Participate is whatever remains, trailed by structure, and it is where the occasional outsized result comes from.

Phase          Timing                Evidence needed        Fires
Harvest        30–90s after break    none — the impulse     most attempts
Protect        2–5 min               acceptance forming     some
Participate    5–20 min              structure above break  few

Read the last column. The phase needing the least evidence fires most often. That is the design: revenue is front-loaded onto the highest-probability segment of the sequence, and the speculative part is funded by it.

Two implementations cover most accounts.

2 CONTRACTS
enter 2, T1 limit placed with the entry
T1 fills            → sell 1, stop up to the breakout structure low   HARVEST
acceptance forms    → hold 1, trail below each higher low             PARTICIPATE
rejection, any time → exit both

4 CONTRACTS — 25 / 25 / 50
enter 4, T1 limit placed with the entry
T1 fills            → sell 1                                          HARVEST
T2 or extension     → sell 1, stop up to the breakout structure low   PROTECT
retest holds        → hold 2, stop below the retest low               PARTICIPATE
new high            → sell 1, trail the last contract
rejection, or 12 minutes with no new extreme → exit the remainder

The four-contract version earns its complexity through the protect phase. Once two contracts are out, the trade is usually profitable whatever the last two do, and a position you cannot lose money on is one you can leave alone through the retest — the hardest thing in this method. Throughout, the stop moves up or it stays: never down, and never to entry, which sits inside the zone where a normal retest reloads.

We have not backtested these schedules. The phase model is order-flow reasoning built on the excursion numbers above, not a measured result with a win rate attached.

One contract cannot execute this method at all

With one unit you have two states, in and out — every decision binary, irreversible, and taken at maximum uncertainty. With four you have five states and four decisions, each made with more evidence than the last.

That is not a comfort argument, it is structural. With a single unit you can harvest or you can participate, never both — and this method's revenue comes from doing both: the harvest collects from the majority of attempts that fail, the runner from the minority that work. Remove the ability to divide the position and what remains is an all-or-nothing trade present in every fake breakout, taking the full loss on three attempts in four.

What scaling out actually costs

Say this plainly, because most teaching on the subject is not honest about it: scaling out lowers expectancy whenever the move continues. Every contract sold at T1 is a contract not sold at T3, and on the trades that run, all-in and all-out beats every scale-out schedule by construction. Anyone presenting partial exits as free is selling something.

What you buy with that cost is being paid on the attempts that fail, and in our sample that is 75% of them.

Which is correct depends on the distribution you are trading. If failures were rare and winners long, all-out would be right. Here the failures are the majority and their early excursion is real — 0.16W at three minutes, 0.24W at ten — so the harvest collects a payment that exists on most attempts. It is a deliberate exchange of right tail for frequency, worth making only because that is the shape of this distribution.

The option is the vehicle; the underlying is the signal

The underlying tells you whether there is a trade, which way, where the stop sits and where the targets are. The contract answers one gating question: given the move I expect, does this thing convert it into enough money to justify the risk? Sometimes the answer is no, and you skip a setup that passed every other test. Options analysis never generates a signal — cheapness is not an edge, it is a description of how unlikely the move is.

Delta is the exchange rate — premium moves roughly by the underlying move times delta times one hundred — and it turns a structural target into a premium expectation. Gamma is why the impulse matters far more on a same-day option than on shares: delta rises as the contract goes into the money, so the second half of a favourable move pays more than the first. A fast forty cents captures both before anything decays; the same forty cents over eight minutes hands part of it back to theta and to implied volatility contracting from the spike the break itself caused.

Then the cost, where most 0DTE breakout plans quietly die. Our measurement of option round-trip cost puts it at 0.06 to 0.27 R — six to twenty-seven per cent of a unit of risk gone to spread and fees before the position does anything. T1 has to clear that by a wide margin or the trade works for the broker.

That cost also settles the strike question. Far-OTM tickets look attractive because the percentage moves are larger, and our study of real 0DTE prices found them a poor vehicle. The near-the-money contract pays at T1, the exit that fires on the largest share of attempts; the far-OTM contract optimises for the rarest outcome and turns illiquid at the moment you need out. Stay near the money for this use.

The rule that keeps the two layers separate: premium P&L is an overlay on market structure, not a substitute for it. "I am up thirty per cent" is not a level. "I am down twenty-five per cent" is not an invalidation. The stop and the targets live on the underlying. Premium behaviour is diagnostic — a contract moving far less than delta predicted means volatility contracted or your mark is stale — and that is an execution input, never a replacement for the structure.

What you should take from this lesson

1. The target exists before the entry, and the first one is set to fill. T1 is the only target that fires on the modal outcome — failed breakouts in our 489-attempt sample still reached a median 0.16W at three minutes, which is what makes it collectable.
2. Risk budget first, contract count second. The count is the output of a minimum across budget, hard cap, liquidity and concentration — and on tight structural stops the hard cap binds.
3. Harvest, protect, participate — in that order of frequency. The phase needing the least evidence fires most often, funding the speculative part of the position with the reliable part.
4. One contract cannot run this method. With a single unit you can harvest or participate, never both, and the edge comes from doing both.
5. Scaling out lowers expectancy when the move continues. That is its honest price. What it buys is being paid on the 75% of attempts that fail — the majority of what you will experience.

Next: scoring the setup, the state machine that runs the whole sequence, and the no-trade rules that decide most of your results before a single order is placed.