How the PnLock engine works.
A plain-English explanation of the stop models, quality score, portfolio analytics, and re-entry scoring behind PnLock.
Principles
PnLock is decision support, not automation. Every number below is an estimate built from public market data, designed to make a manual decision more disciplined, never to predict the future or place trades for you.
- Stops are anchored to the current price (the latest close), not your entry, so protection reflects where the stock is now.
- A model returns nothing rather than a bad answer: if a computed level would sit at or above the current price, it is shown as unavailable.
- Nothing here is financial advice, and past volatility never guarantees future behaviour.
The six stop models
Profit Lock computes up to six exit levels for the same position, each with a different philosophy, then lets you compare them side by side.
| Model | How it places the exit |
|---|---|
| Conservative Shield | A fixed percentage below the current price. Simple and predictable; ignores volatility. |
| Balanced Defender | 1.5× the 14-day ATR below price. A volatility-aware default for most stocks. |
| Volatility Guard | 2× the 21-day ATR below price. Roomier and more patient for longer holds. |
| Momentum Protector | Just under the 20-day swing low, the most recent level buyers defended. |
| Crash Hunter | At the 50-day moving average, a trend-break line for clearly uptrending names. |
| Adaptive Protector | A Chandelier Exit: 3× ATR below the 22-day high. Trails up with the trend, never down. |
The quality score (0–100)
Every candidate stop is graded across four sub-scores that sum to 100, then mapped to an A–F grade. The score answers "is this a sensible level?", not "will this trade win?".
- Grades: A/B = sound, C = workable, D/F = likely to misfire, reconsider the model or re-run.
| Component | Weight | What it checks |
|---|---|---|
| Noise tolerance | 40 pts | Is the stop far enough from price that ordinary daily wiggles won’t trip it? |
| Distance from price | 30 pts | Is the distance sensible as a multiple of ATR, neither hair-trigger nor far too loose? |
| Model agreement | 20 pts | How tightly do the other models cluster around this level? |
| Freshness | 10 pts | How recent is the data and analysis behind the level? |
Portfolio analytics
Portfolio Analytics decomposes holdings, including the companies inside your ETFs, where data is available, and maps exposure across several dimensions.
- Concentration is graded with a Herfindahl-Hirschman Index (HHI) over holding weights, so "many tickers" doesn’t get mistaken for "diversified".
- Exposure is broken out by holding, sector, industry, country, currency, market-cap band, and portfolio beta.
- ETF passthrough uses issuer-fed holdings where available and falls back to a top-10 estimate otherwise.
Re-entry scoring
Reentry Intelligence runs five checks on a stopped-out name and rolls them into a 0–100 readiness score with a Re-enter / Watching / Wait status.
- Stabilisation, has disorderly price action calmed down?
- Volatility reset, are daily ranges contracting from the panic?
- Higher-low structure, is price carving constructive higher lows?
- Volume context, is participation confirming the recovery?
- Market regime, is the broad market (SPY) supportive rather than hostile?
Data & limitations
Models run on daily OHLC candles from market-data providers; enrichment (sector, market cap, ETF composition) comes from public sources and is cached.
- Disclosure feeds (SEC Form 4, US lawmaker trades) are public, delayed, and can be amended, verify before acting.
- Backtests are historical simulations and can mislead if overfit; they are not proof of future results.
- All levels are estimates from past data. News gaps and regime shifts can move straight through any stop.