Volatility is not direction

It is a Sunday evening, before ES reopens at 18:00 ET (06:00 ICT Monday, winter). A widely followed paid desk reports that its published targets were reached on 97% of sessions. Does the desk know where price is going? This lesson shows why a number like that can carry no information beyond volatility, and how to check it with a calculator and a random level. Our lab ran that check on 66 sessions of one widely followed desk.

Key terms

Volatility (biến động): how far price usually travels in a period, whichever way it goes Definition. It says nothing about the direction.

Touch (chạm): price trades at the level at least once in the window you chose, for example 18:00 ET to 16:00 ET the next day Definition. A touch is not a win and not a reversal.

Distance in σ (khoảng cách tính bằng độ lệch chuẩn): how far a level sits from the current price, divided by one day’s σ Definition. A level 80 points away on a day whose σ is 315 points sits 0.25σ away.

Terms you need: σ (sigma, độ lệch chuẩn) is the spread of daily moves (EV1.01). For an index future, one day’s σ ≈ price × (IV ÷ 100) ÷ √252 Definition, where IV is options-implied volatility (VXN for the Nasdaq-100, VIX for the S&P 500, Cboe 30-day indices) and √252 ≈ 15.87. Settle is the exchange’s official daily closing price, RTH is regular trading hours (09:30–16:00 ET), stop and target are exit orders for a loss and a profit, drift is a steady average push in one direction, and pp means percentage points (M0.05 covers contracts; OPF4.03 tests this σ ruler).

Chart first: slide a level toward price

How it works

Volatility sets the reach

Every session, price wanders: a little on a quiet day, a lot on a busy one. Implied volatility is the options market’s price for that wandering, so σ gives you a ruler before the open Definition. Any level inside the day’s normal wander gets touched often, whether it came from a chart pattern, an options model, a round number or a dart.

The touch rule: 2·(1−Φ(k))

Picture price as a random walk with no drift. Put a level k σ above the start. The chance that the day closes beyond it is 1−Φ(k), where Φ is the bell-curve share you met in EV1.01. Now the trick, called the reflection rule: every path that touched the level and ended back below has a mirror twin that touched it and ended above. So the chance of a touch is twice the chance of closing beyond:

touch ≈ 2·(1−Φ(k)) Model

At k = 2 that is about 5%; at k = 1, 32%; at k = 0.5, 62%; at k = 0.25, 80%. Real days differ from the model MeasuredNasdaq index (see “The mirror and the failure”), but the shape holds: the closer the level, the higher the hit rate.

Publishing targets on both sides

If a service publishes targets at modest σ distances and counts a hit whenever a level is reached during the session, price almost never stays inside a narrow band for a whole day, so a high session touch rate is expected from volatility alone Model.

Touching is not winning

A hit rate counts touches at any time in the window, including days when your stop was taken first. With a stop and a target at equal distances and no drift, each is reached first about half the time Model. The 80% touch rate and the 50% win rate are both true on the same day.

What the data says

The ruler itself is not perfect either. On the Nasdaq-100 index (2001 to 2026, 6,482 days) a level 1σ above the prior close was touched on 18.2% of days and a level 1σ below on 19.8%, against 32% from the touch rule MeasuredNasdaq index. Implied volatility usually runs above what the day delivers, so real touch rates sit below the model. That does not rescue the 97% claim: the placebo was counted on the same real bars.

Worked example: the long side

An illustrative NQ day, not a real date. NQ settled at 25,000.00 and VXN is 20, so one day’s σ ≈ 315 points.

1. You plan a long from 25,000.00 with a target at 25,080.00: 80 points, 320 ticks of 0.25. On one MNQ ($2 a point) that is $160; on one NQ ($20 a point), $1,600.
2. Distance in σ: 80 ÷ 315 ≈ 0.25σ.
3. Touch chance from the rule: 2·(1−Φ(0.25)) ≈ 80% Model.
4. Your stop sits at 24,920.00, also 80 points away, so it is also touched on about 80% of days, and about half the time it comes first.

Publish this target daily and you can honestly say “reached about 80% of the time” while the trade wins about half the time and loses money after costs.

The mirror and the failure

The mirror. Sell at 25,000.00 instead, target 24,920.00. The same rule gives the same 80%. The hit rate is identical whichever direction you pick, which is the whole point: volatility is not direction. A forecast of direction has to beat the coin flip between the long and the short target, not the touch rate.

Where the distance-only view fails.

• Calm or high-volatility regimes. The real touch rate shifts with the volatility regime: on the Nasdaq-100 a 1σ level above was touched on 12.0% of days when VXN was under 15 and 21.0% when it was 30 or higher MeasuredNasdaq index.
• “No edge detected” is not “no edge”. A 66-session sample is small; subtle effects take a larger sample to settle (EV3.01).
• Placebos must be honest. Same distance, same window, same sessions (EV1.04); a placebo placed farther away flatters the real level.

Do it on the platform

Prop-firm note. A target’s touch rate says nothing about passing an evaluation: stops, costs and the firm’s loss limits decide that. The Prop Firm course covers it with each firm’s dated rules.

Common mistake: treating a base rate as an edge. “Reached 97% of the time” is a base rate until a random level at the same distance has been counted beside it.

Common mistake: reading a touch as a win. Ask what was hit first: the target or the stop.

Drill: put three levels on the ruler

Take NQ at 24,000.00 with VXN at 24. Before you look at the solution:

1. Compute one day’s σ in points.
2. Place three levels: 90 points above, 360 points below, and 720 points above. Write each one’s distance in σ.
3. Estimate each level’s touch chance with 2·(1−Φ(k)). Use Φ(0.25) ≈ 0.60, Φ(1) ≈ 0.84, Φ(2) ≈ 0.977.
4. In replay, log ten sessions: beside each level, a random level at the same distance on the other side; record touch or no touch for both.

What we don’t know yet

• Touch rates on NQ futures, overnight session included: not yet measured (the numbers above are index closes and RTH ranges).
• Event-day touch rates on scheduled release days such as FOMC and CPI: not yet measured.

Recap

• A level’s hit rate mostly reflects its distance from price in σ.
• Touch chance under the model is 2·(1−Φ(k)): about 80% at 0.25σ, 32% at 1σ, 5% at 2σ.
• A long target and a short target at the same distance get the same touch rate, so a touch rate cannot show direction.
• Our lab found no information beyond volatility in one paid desk’s 97% target rate (ES, 66 sessions).
• Before trusting any hit rate, count a random level at the same distance beside it.

Further reading

• Cboe, Volatility Index (VIX) Methodology (public document): what implied volatility measures.
• Steven Shreve, Stochastic Calculus for Finance II, chapter 3: the reflection principle behind 2·(1−Φ(k)).
• Our lab: one widely followed paid desk’s ES levels against distance-matched placebo levels (66 sessions, 2026): the published target was reached as often as random levels at the same distance.
• Our lab: Nasdaq-100 and S&P 500 index moves against the VXN and VIX daily ruler (2001–2026 and 1990–2026).

Before this: EV1.01 (the bell curve) and EV1.04 (building a placebo). Next: EV2.01 looks at what retail day trading really returns: once you can tell an edge from a base rate, you can judge the business honestly.

Educational content only — not financial advice. Futures trading involves substantial risk of loss.