Placebos and baselines

It is 10:05 ET (22:05 ICT, winter). NQ drops into yesterday’s low, trades a few ticks through it, and bounces 90 points in half an hour. Someone in your chat writes “prior-day low held again”. It feels like proof. But price has to turn somewhere, and on most days it turns many times. The question this lesson answers: if you had drawn a line at a random price the same distance from the open, how often would that line have “held” too? Until you know that number, the bounce tells you nothing about the level.

Key terms

Baseline (mức nền): the rate of an outcome without the thing you are testing Definition. If prior-day lows hold 20% of the time, the baseline is how often any price at that distance holds.

Placebo level (mức giả lập): a random price placed at the same distance from the open as a real level, scored with exactly the same rules Definition. It is the baseline made concrete: a line with no story behind it.

Touched, held, broke (chạm, giữ, thủng): three outcomes you score after price reaches a level Definition. Touched: price trades at the level. Held: after the touch, price moves your bounce distance away before it moves your break distance through. Broke: it goes your break distance through first.

Terms you need: ATR (average true range) is the average daily range over the last 14 days, in points; 0.5 ATR is half a normal day’s range. Overnight high/low are the highest and lowest prices from the 18:00–09:30 ET Globex session; prior value area is the price zone where 70% of yesterday’s volume traded. A stop is an order that exits a losing trade at a set price, MNQ is the Micro E-mini Nasdaq-100 future ($2 a point), and pp means percentage points.

Chart first: distance decides most of the count

How it works

Why distance is the hidden variable

A level 0.1 ATR from the open gets touched on most days, because most sessions travel at least that far from the open Definition. A level 1.5 ATR away gets touched rarely Definition. So “touched 80% of the time” mostly tells you how far the level sits from price, not what the level is. The same goes for “held”: every touch is followed by some bounce, and a small bounce counts as “held” often, wherever the line is Definition. A placebo strips the story out and keeps the distance, so whatever difference remains belongs to the level.

Building a placebo, step by step

1. Measure the distance. Take the real level’s distance from the 09:30 ET open and divide by ATR. Prior-day low 210 points below the open with ATR at 420 points is 0.50 ATR.
2. Place the placebo at the same distance. The simplest rule is the mirror: the same distance on the other side of the open. Another rule keeps it on the same side as the real level, a small seeded step away. Pick one and keep it.
3. Keep it clear of real levels. If the placebo lands within 0.1 ATR of another real level you have drawn, step it outward in small fixed steps until it is clear House rule.
4. Seed the draw. Use a fixed rule (such as the mirror, or a coin flip written down before the session) so you cannot nudge the placebo to a “worse” spot after the fact.

Write your rules before you look

Fix the outcome rules before the session House rule. Write three numbers in your journal: a bounce distance (how far away counts as “held”), a break distance (how far through counts as “broke”), and a window (how long you wait), scaled in ATR. Our worked example below uses 0.20 ATR bounce, 0.10 ATR break and a 60-minute window as teaching numbers; your own numbers can differ as long as they are written first and applied to both lines. Choosing rules after seeing the bounce inflates any hit rate.

What the data says

Worked example: a long at the prior-day low

Illustrative numbers, not a real date. NQ opens at 25,180.00. ATR is 420 points, so your written teaching rules are a bounce of 84 points (0.20 ATR), a break of 42 points (0.10 ATR) and a 60-minute window.

• Real level: prior-day low 24,970.00, 210 points below the open (0.50 ATR).
• Placebo: the mirror is 25,390.00. The overnight high sits at 25,362.00, only 28 points away (under 0.1 ATR). One outward step of 0.05 ATR (21 points) moves it to 25,411.00, 49 points clear (231 points above the open, 0.55 ATR).

At 10:05 ET price trades down to 24,970.00 (touched). The low prints 24,961.25, 8.75 points through, well short of the 42-point break. By 10:38 ET price reaches 25,060.50, 90.5 points off the level: held. Had you bought 1 MNQ at 24,972.00 with a stop at 24,927.75, your risk was 44.25 points × $2 = $88.50 plus fees.

Now score the placebo with the same rules. At 14:20 ET price rallies to 25,411.00 and falls back to 25,322.00, 89 points off: held as well. One session cannot separate them; only a scored log across many sessions can.

The mirror and the failure

The mirror. A short at the prior-day high flips the signs: held means price falls the bounce distance before trading the break distance above, compared against a mirror price below the open.

The failure. In the 5-minute check on NQ and ES futures, prior-day high, low and close held on 22.9% of touches, while random prices at the same distance held on 35.1% Measured: no edgeNQ. Across only 96 touches in one regime, the levels held less often than random prices did (so non-holds—breaks or timeouts—were more common at the real levels). A trader who counted only the bounces at prior-day levels would never have seen this.

Do it on the platform

If you trade an evaluation account, the firm’s loss limits decide what a broken level costs you; the Prop Firm course (PROP1.01) covers them. A placebo test changes what you believe about a level, not your stop.

Common mistake: treating a base rate as an edge. “Prior-day low held 19% of the time” sounds like a fact about the prior-day low. It is mostly a fact about price moving: random prices at the same distance held 19.4%.

Drill: score your own map

For five sessions, mark three levels before 09:30 ET and one placebo for each, using the steps above.

1. Before the first session, write your bounce distance, break distance (both in ATR) and window. Do not change them during the drill.
2. After each session, log every line as untouched, touched-held or touched-broke.
3. After five sessions, compute the held rate for real levels and for placebos.

Recap

• A hit rate is only informative next to its baseline: the same outcome at a random price the same distance away.
• Build the placebo at the same distance in ATR, clear of real levels, with a seeded rule.
• Write your bounce, break and window before you look, and score both lines with them.
• Our lab’s test across 721 sessions per instrument (QQQ and SPY pooled, 10,288 touches) found real pre-market levels held at the placebo rate Measured: no edgeNasdaq index.

Further reading

• J. Welles Wilder Jr., New Concepts in Technical Trading Systems (1978): the original definition of true range and ATR.
• Efron and Tibshirani, An Introduction to the Bootstrap (1993): the resampling method behind the intervals in this lesson.
• Our lab: pre-market level families vs a distance-matched placebo on 721 sessions per instrument (QQQ and SPY pooled, 1,437 instrument-sessions), and the 46-session 5-minute check on NQ and ES futures. Neither found a level family that held more often than its placebo.

Next: EV1.05 applies the same comparison to far price targets, where volatility alone explains most of the hits.

Educational content only, not financial advice. Futures trading carries substantial risk of loss and is not suitable for every investor. Past results, measured or simulated, do not guarantee future results.