What "it works 85% of the time" really means

It is 09:25 ET (21:25 ICT, winter), five minutes before the New York open. NQ futures trade at 20,402.25. A chat you follow posts a line on the chart at 20,488.00 with the words “naked POC from Tuesday, these get filled 85% of the time”. If price goes back to that line eight or nine times out of ten, buying here and aiming at it looks easy.

This lesson asks one question about that number, and the same question works on every hit rate you will ever be sold: 85% compared with what?

Key terms

Hit rate (tỷ lệ trúng): the share of cases where the promised outcome happened Definition. “85 of 100 naked POCs were revisited” is a hit rate of 85%.

Base rate (tỷ lệ nền): how often the same outcome happens without the signal, over the same window and from the same distance Definition. For a level, the cleanest base rate is a random price placed just as far from the market.

Lift (mức chênh): the hit rate minus the base rate, in percentage points (pp) Definition. A lift near zero means the signal adds nothing you did not already get from distance and time.

Naked POC (POC chưa được chạm lại): the point of control of an earlier session that no later session has traded back to yet Definition.

Term you need: POC. The point of control is the price where the most volume traded in a session or profile. It is a description of where business was done, not a forecast. M2.01 covers it in full; M0.02 has the basic market vocabulary.

Chart first: the line you were told about, and fifty lines nobody named

How it works

Why random prices get revisited so often

In our 58-session NQ sample from EV1.01, the median session travelled 295 points from high to low while the median open-to-close change was −25.9 points Context. Because price wanders back and forth across a wide range before settling, any price close to the market gets traded again often Definition. The closer the price and the longer you wait, the closer that chance gets to 100% Definition. EV1.05 shows the arithmetic.

So a level 85 points away being revisited “most of the time” is what you would expect even if the level meant nothing. The 85% tells you about distance and time. To learn anything about the POC itself, you need the rate for a random price at the same distance over the same window.

Reading a hit rate in three steps

1. Name the outcome and the window. “Revisited” within what time: the next session, a week, ever? A rate with no window can be pushed toward 100% by waiting longer.
2. Find the base rate. The same outcome, the same window, the same distance, without the signal.
3. Take the difference. 85% minus 84% is a lift of about 1 point. That is the part of the number that belongs to the naked POC.

Myth / hypothesis: “Naked POCs are magnets: the market comes back to finish unfinished business, so a naked POC is a target.” Lore The revisit is real and frequent. On the claim that the POC causes it or beats any other price at that distance, a public NQ study was inconclusive: no gap from random prices, with an interval that includes zero (box below). Treat it as a hypothesis you could test, not a rule.

Why it is still worth knowing

A naked POC is useful vocabulary: it marks where the auction did most of its business in a past session, which helps you describe the day. It is a sensible place for an alert or a review. And “naked POCs get revisited more than random prices” is a fair hypothesis that you now know how to test. On current evidence it is not a reason to expect price to go there.

What the data says

Our lab ran its own check on public data, with a different definition and a shorter window, and found the same answer.

Worked example: the long toward a naked POC

An illustrative session, not a real date. The prices are realistic for NQ but made up.

NQ opens at 20,402.25. The naked POC sits above at 20,488.00, 85.75 points away. You consider a long at 20,403.00 with a stop at 20,373.00 and a target at 20,487.75, one tick under the line. One NQ contract moves $5 per 0.25-point tick, so $20 per point.

• Risk: 30 points × $20 = $600.
• Reward: 84.75 points × $20 = $1,695.

If you read “85%” as your chance of winning, the trade looks wonderful: 0.85 × $1,695 − 0.15 × $600 = +$1,351 per trade. That reading has two errors.

First, the 85% is a revisit rate over the study’s window, not “before my stop, today”. Your trade needs price to travel +84.75 points before it travels −30.

Second, the base rate. With no edge at all, a market that wanders with no trend reaches +a before −b with probability b ÷ (a + b) Model. Here that is 30 ÷ (30 + 84.75) = 26.1%. The expected value is 0.261 × $1,695 − 0.739 × $600 ≈ $0 before costs, and below zero after commissions (broker and exchange fees) and slippage (worse fills than planned). That is “no edge” with a stop and a target: the payoff looks good, and the chance of winning shrinks to match it exactly.

The naked POC would have to lift that 26% to earn its place. The data above says it adds about 1 to 2 points to a revisit rate, with an interval that includes zero.

The mirror and the failure

The mirror (short side). Put the naked POC below the market instead: NQ at 20,402.25, a naked POC at 20,318.50, 83.75 points under. “It gets filled 85% of the time” now invites a short. Every step is the same with the signs flipped: random prices 83.75 points below are revisited about as often, and a short with a 30-point stop reaches the line first only about 26% of the time if the market has no trend (30 ÷ 113.75).

The failure. Picture a trend day: NQ opens at 20,402.25 and climbs 250 points by the close. The naked POC 83.75 points below stays untouched that day and for the next week. A trader who shorted toward it, then added because “85% get filled”, is now in the 15%; the number never said when. Whether the misses cluster on trend days is not yet measured.

Do it on the platform

Common mistakes

Common mistake: Treating a base rate as an edge. “85% revisited” is mostly what any price at that distance does. Without the rate for random prices at the same distance (about 84%), the number says nothing about the naked POC.

Common mistake: Reading a revisit rate as a win rate. A level can be revisited 85% of the time while a trade toward it, with a stop, wins about 26% of the time. Different outcome, different window.

Drill: find the lift

For each claim, name the base rate it needs, then compute the lift in percentage points where you can.

1. “Our level gets touched 72% of the time.” No other number is given.
2. “Breakouts after our signal reach 1R (one unit of initial risk) 58% of the time. All breakouts in the same hours reach 1R 55% of the time.”
3. “Prior-day POC touched next session 55%. Random prices at the same distance: 54%.”

Recap

• A hit rate means nothing until you know the base rate: the same outcome, window and distance, without the signal.
• Lift is hit rate minus base rate. For naked POCs it is about 1 to 2 points, with an interval that includes zero.
• Random prices near the market are revisited often because price wanders; distance and time explain most of the 85%.
• A revisit rate is not the win rate of a trade with a stop and a target.

Further reading

• Order Flow, Measured: Headline findings and all claims. A public, pre-registered practitioner study on NQ, 2018–2026; read the volume-profile rows (VP-01 to VP-08) and the baseline tables.
• Freedman, Pisani and Purves, Statistics, the chapters on chance and on comparing two rates.
• Our lab: pre-market level families vs placebo (QQQ and SPY, 721 sessions): naked POCs and prior POCs are touched and respected at the rate of random prices at the same distance.

Prerequisites and next

Read EV1.01 (rates and intervals) first; M0.02 covers the market vocabulary. Next: EV1.03 turns this into the badges you will see on every claim on this site: Measured, Measured: no edge, Plausible and Lore.

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