A day trading plan as a hypothesis: conditions, trigger, invalidation, target
A day trading plan written as conditions, trigger, invalidation, target, expected base rate and size turns a setup into something you can test. Decide the Path B response before the trade, and change the plan from counts, not from one outcome.
Start in plain English
This lesson turns a setup into a written plan you can test. For a small-cap scalper the plan matters more than usual: decisions happen in seconds, stocks halt, and a loss larger than planned can come from one hesitation. A plan written beforehand makes the decisions while you are calm and gives you something to count afterwards.
The core idea: a day trading plan is a hypothesis. It says, "under these conditions, when this happens, risking this much, I expect outcomes to fall roughly like this." Each trade is one observation, and the plan is judged on many of them, not on the last one.
The template has six parts plus two that most plans leave out:
| Part | Question it answers |
|---|---|
| Conditions | Which stocks and moments qualify? |
| Trigger | What exact event starts the trade? |
| Invalidation (stop) | What price says the idea was wrong? |
| Target and time stop | Where and when does the trade end if it works or stalls? |
| Expected base rate | What split of outcomes do you assume, and where did it come from? |
| Size | How many shares, from dollar risk and liquidity? |
| Path B response | What do you do after a stop, a halt or a partial fill? |
| Costs | Fees, spread and a slippage allowance, in R |
Two paths from the same start
All prices, sizes and costs are Illustrative. Thistledown Semiconductor is fictional, and the plan is a teaching example, not a recommended setup.
The plan, "Pullback v1," written before the session.
- Conditions: gap of 20% or more on a company release with a filing; price $2–$10; spread 1 cent; 09:35–10:30 ET; no halt in the last 15 minutes.
- Trigger: after a pullback of at least three 1-minute bars with lower highs, the first trade above the last pullback bar's high.
- Invalidation: 2 cents under the pullback low.
- Target: 2R, resting limit. Time stop: 10 minutes.
- Expected base rate: unknown; assumed 35% target, 45% stop, 20% time stop (Illustrative) until 20 logged trades replace it.
- Size: $50 risk ÷ per-share risk, capped at a fifth of the displayed size at the best bid.
- Path B response: after a stop, no re-entry on the same pullback; after two stops in this plan, stop trading it for the day; in a halt, exit at the reopening with a marketable limit 5 cents through the bid.
- Costs: $0.005 per share round trip, plus a 2-cent stop-slippage allowance.
The setup at 09:45 ET. Thistledown gapped 34% on a company release with an 8-K. It ran to $3.91 at 09:40, then pulled back for four bars with lower highs to a low of $3.75. Quote at 09:44:59: $3.79 × $3.80.
Applied to this chart. Trigger at a trade above $3.81; marketable limit (ceiling $3.84) fills at $3.83. Stop $3.73 ($3.75 − 2 cents). 1R = $0.10; size 500 shares (displayed bid 3,000 shares, so the liquidity cap does not bind). Target $4.03. Time stop 09:55.
Path ATarget, plan followed
09:45 trades $3.82 and the buy fills at $3.83. Price reaches $4.04 at 09:50 and the $4.03 limit sell fills.
Path BStop, plan followed, Path B response followed
The buy fills at $3.83; at 09:47 the pullback low breaks and the $3.73 stop fills at $3.71. Price bounces to $3.86 at 09:52. The trader does not re-enter, as written.
Path CTime stop, plan followed
The buy fills at $3.83; price chops between $3.78 and $3.90. At 09:55 the time stop sells at the bid, $3.82.
All three reviews say "plan followed," and that is the point: the plan was the decision; the path was the draw. The 09:52 bounce in Path B is the moment many traders break a plan. Because the Path B response was written in advance, there was nothing to decide. The rest of this lesson covers what a deviation looks like in review, the base-rate line, and how to change a plan without reacting to one outcome.
A deviation, reviewed against the plan
Reviewing against the plan separates two questions: did I follow it, and how did it turn out? FennEdge's Trading Journal holds real broker-synced trades only. On Pro and Max, its Review page lets you mark each real trade "Within plan" or "Outside plan", add a Setup tag such as "Pullback v1" and a note, and see statistics by setup (as of 2026-10-05). Practice trades belong in the practice log instead.
Writing each part so it can be counted
- Conditions use things visible at the decision time. "Strong stock" cannot be counted; "gap of 20% or more, spread 1 cent, 09:35–10:30" can.
- Trigger is an event, not a feeling: a price traded, a bar closed. If two people would disagree on whether it happened, rewrite it.
- Invalidation is where the idea is wrong, not where the loss feels too large. Where a stop goes covers the trade-off with noise.
- Target and time stop make Path C explicit. Without a time stop, a stalled trade has no end and no count.
- Size follows from dollar risk and per-share risk, then a liquidity cap; Position sizing and Is it liquid enough to scalp?.
- Costs go into the plan in R, so the expected result is after costs from the start.
Common gaps in written plans
Most plans that fail a review fail on the same few lines. The table lists them with what each costs in the log (Illustrative examples).
| Gap in the plan | What it does to your numbers | A countable fix |
|---|---|---|
| No time stop | Stalled trades run until something else ends them; Path C disappears into A or B and the split is distorted | "Exit at the bid after 10 minutes if neither target nor stop has filled" |
| Trigger described as a feeling ("when it looks strong") | Two reviews of the same chart disagree on whether a trade qualified | A price or bar event: "first trade above the last pullback bar's high" |
| Stop at a dollar amount you can stand rather than at invalidation | Size and stop are chosen together by comfort, so 1R drifts from trade to trade | Stop at a chart level, then size from it |
| No cost line | Results are recorded gross; a plan that breaks even before costs loses after them | Fees and a slippage allowance in R, written once |
| No Path B response | Re-entries and cancelled stops appear after losses, which the plan did not test | "After a stop, no re-entry on the same setup; after two stops, done for the day" |
| Conditions added after the fact | Trades that lost are explained away as "not really the setup" | Conditions dated and versioned before the block starts |
| Base rate copied from a social post | The plan's break-even test compares your log with a number nobody counted | Source written beside the number: study, your log, or "assumption" |
A plan with all six template parts and none of these gaps is still only a hypothesis. It is a hypothesis your log can test.
The base-rate line
Most plans skip the line "how often do I expect each outcome?" Writing it does two things. It forces you to say where the number comes from: a FennEdge study, your own log, or an assumption labelled as such. And it gives the plan a test.
So the hypothesis reads: "Under these conditions, the target share after costs is above about 32%." That is a sentence your log can support or contradict. Expectancy covers the formula.
Observe vs. infer
At 09:45 in Thistledown Semiconductor (fictional; all figures Illustrative):
| What the screen shows (observed) | What a trader might infer | Why it is not established |
|---|---|---|
| Four 1-minute bars with lower highs | "A healthy pullback" | "Healthy" is judged by what happens next |
| Volume fell from 610,000 to 240,000 shares | "Sellers are exhausted" | Falling volume fits a pause and a slow fade |
| The plan's conditions are all met | "This trade works" | Conditions qualify a trade; they do not pick its path |
| A 34% gap on a filed release | "The news supports more upside" | The catalyst is a condition, as in Catalysts |
What shifts the odds
For a new plan, nothing is measured yet: the conditions in the plan are hypotheses to record, not findings. Record, on every trade, the items you suspect matter (pullback depth, bars in the pullback, volume ratio, spread, minutes since the open) so a later version can test them. Add a condition to the plan only when your log shows a difference larger than its noise.
Cost of being wrong
Pullback v1 at 500 shares (Illustrative):
| Outcome | Exit | R after costs | Dollars after costs |
|---|---|---|---|
| Path A: target | $4.03 | +1.95R | +$97.50 |
| Path B: stop with 2-cent slippage | $3.71 | −1.25R | −$62.50 |
| Path C: time stop | $3.82 | −0.15R | −$7.50 |
| Deviation: stop cancelled | $3.58 | −2.55R | −$127.50 |
| Halt, reopening at $3.50 | $3.50 | −3.35R | −$167.50 |
Two stops in a row cost about −2.5R, and the plan then ends the day for this setup. A cap like that is one form of daily limit; Daily loss limits covers the account-level version.
Changing the plan from data, not from one outcome
- Version every change. "Pullback v1" becomes "v2" with a dated note of what changed. Results from v1 and v2 are counted separately.
- Change at block ends. Decide in advance how many trades a block holds (20 is a minimum; more is better), and change the plan only after the block.
- Change for a reason the log shows. "The target share was 4 of 20 (20%), below the 32% break-even" is a reason. "The last trade lost" is not.
- Expect wide intervals. With 7 targets in 20 trades (35%), the 95% interval runs from about 18% to 57% (Wilson interval, computed on these hypothetical counts). That range includes both a losing and a winning plan. How many trades before your numbers mean anything explains the method.
What this lesson does not claim
- Pullback v1 is a teaching example, not a recommended setup. Its split is Illustrative.
- A written plan does not make a trade more likely to work. It makes outcomes countable and decisions consistent.
- Thistledown Semiconductor and its bars are fictional.
Go deeper
- FennEdge first.
- The Trading Journal syncs real fills from supported brokers. Its Review page (Pro and Max, as of 2026-10-05) offers "Within plan" / "Outside plan" marks, Setup tags and notes, statistics by setup, and an optional AI review that cites only your own trades and statistics (it uses your plan's AI allowance). It holds broker-synced trades only, not simulated or practice entries.
- Articles: An active trader's workflow and Trading journal review.
- Implementation intentions. Psychologist Peter Gollwitzer (1999) reviewed evidence that "if X, then I do Y" plans help people act as intended. A Path B response is an if-then plan.
- Checklists. A one-page checklist of the conditions, read before each trade, is a simple way to apply the plan.
- Playbooks. A collection of tested plans is a playbook; the Advanced course builds one in Building a playbook from your own base rates.
- Related lessons: Thinking in probabilities, VWAP reclaims and rejections, Bull flags, Reviewing scalps, Capstone.
Exercises
Six exercises follow this lesson: three concept checks, one scenario that reviews trades against a plan, and two practice tasks. Exercises, answers and progress are available after signing in.
Practice tasks: write two plans, then test each 10 times. First, write two plans with every part of the template, including the base-rate line with its source and the Path B response. Second, run each plan 10 times in a simulator or on replayed charts and log every trade with the FennEdge practice log: download the practice log (CSV), with a column guide. Use the plan name and version in definition_version, mark plan_followed honestly, and log setups you skipped. Keep the log separate from the Trading Journal, which holds only real broker-synced trades.
Primary sources
- Gollwitzer, P. M. (1999). Implementation intentions: Strong effects of simple plans. American Psychologist 54(7).
- Baron, J. and Hershey, J. C. (1988). Outcome bias in decision evaluation. Journal of Personality and Social Psychology 54(4).
- FINRA. Frequent Intraday Trading: Understanding the Basics.
Sources checked 2026-10-05. Thistledown Semiconductor, Pullback v1 and all prices, splits and costs in this lesson are fictional and Illustrative. This lesson is education, not a recommendation to buy or sell any security. Past patterns do not indicate future results.