Methodology & Level-Touch Rate Definition

The exact math, sample size, and honest caveats behind our published 94.2% level-touch rate.

Last updated: September 26, 2026 · Refreshed automatically each month

1. The headline number

Our headline statistic: algorithmically calculated reference levels are touched by intraday price action 94.2% of the time across the top 20 highest-volume symbols on our platform, measured on a rolling 30-day D1 window. This describes how often price interacts with our computed levels — not a trading win rate, not a profit guarantee, and not a prediction of future results.

2. Exact definition of a “touch”

A single plotted level counts as a TOUCH if the following is true:

• The level is plotted at 22:00 UTC on day D (D1 close of the previous session).
• At any point between 00:00 UTC and 23:59 UTC on day D, intraday price crosses within ±0.15% of the level's value.
• Intraday price is measured from the minute-bar high (for a Buy Above zone) or the minute-bar low (for a Sell Below zone), whichever is more favorable to the level.

A level counts as a MISS if the day closes without intraday price coming within ±0.15% of the level's value.

3. Which 20 symbols are in the sample?

The 20 symbols are the highest-volume instruments on our platform by user watchlist frequency over the last 90 days. This list is refreshed monthly. As of September 2026:

XAUUSD BTCUSD EURUSD GBPUSD USDJPY ETHUSD US30 NAS100 US500 USOIL AUDUSD USDCAD USDCHF XAGUSD SOLUSD XRPUSD EURGBP EURJPY GBPJPY DE30

Every one of these symbols contributes equally to the hit-rate calculation. No cherry-picking, no weighting to favor a subset.

4. Sample size & monthly breakdown

Over the last 30 days, we plotted 1,200 levels across the 20 symbols above (approximately 2 zones per symbol per trading day × 30 days). Of those, 1,131 were hit under the definition in section 2, and 69 were missed.

Rolling 6-month average: 93.9%. We publish the current month's number, not the peak.

5. What touch rate is NOT

The hit rate measures one thing only: how often our AI-computed zones align with real intraday order-book behavior. That's useful context; it is not a substitute for skill, risk management, or a trading plan.

6. Data sources & audit trail

All numbers on this page are computed from our own historical records, which log every plotted level with a timestamp and every intraday price tick from our institutional-grade market data feed. The calculation runs automatically each night and refreshes the public numbers on this page without manual editing.

What goes into the calculation:

• Level records: Every level we plot is stored at the moment it is generated, with the symbol, date, level type (buy-above or sell-below), and exact price. Nothing is edited or removed after the fact.
• Price data: Aggregated 1-minute bars from a professional real-time market data feed, resampled per symbol.
• Calculation: A fixed, automated procedure compares each plotted level against the following session's intraday price range under the definition in section 2. The logic does not change between runs.

If you are a serious trader or a journalist and want to independently verify these numbers, we are happy to walk through the calculation with you on request — email support@levelengine.ai.

7. Per-symbol breakdown (last 30 days)

Not every symbol performs identically. Some are stronger (XAUUSD, EURUSD) because their intraday range consistently overlaps with our computed zones; others (SOLUSD, XRPUSD) have wider volatility bands and slightly lower hit rates. Full transparency:

Weighted average across all 20 symbols equals the published 94.2%. Note that we do not hide the underperformers: SOLUSD at 78% and XRPUSD at 75% are lower because those markets have wider intraday ranges that our current model doesn't capture as precisely. We're actively working on a per-symbol volatility adjustment for altcoins.

8. What we don't publish (and why)

We deliberately do NOT publish:

• A profit/loss backtest. Doing so honestly would require modeling entry timing, slippage, spread, and personal risk management — all of which vary per user. Anyone selling you a “backtested strategy” that ignores these is not being honest.
• Individual user P&L. We don't have broker connections yet, so we cannot see whether a user made or lost money after using our levels. When the REST API launches (Q1 2027), users may opt in to share broker performance data for aggregate stats.
• Guarantees. No trading tool can guarantee outcomes. Anyone who does is running a scam.

9. How to reproduce these numbers yourself

If you have a paid subscription and want to verify:

1. Log in to your dashboard.
2. Open any of the 20 symbols listed in section 3.
3. Use the “History” button in the terminal to load the last 30 days of plotted levels.
4. Compare against the intraday high/low of each session (visible in your chart).
5. Count hits and misses under the definition in section 2. You should arrive at the same numbers.

If your count differs by more than 2 percentage points from our published number, please email support@levelengine.ai with your calculation and we'll audit the discrepancy publicly.

10. Change log

• Sep 26, 2026: Page published. Initial disclosure of all methodology details.

Any change to the definition, symbol list, or calculation window will be logged here with a dated entry. Historical numbers will be re-computed and republished if a definition change would materially affect them.