Parameter & Regeln
So wird die Empfehlung berechnet
Jun 13Parameter
die einstellbaren Parameter — direkt bearbeiten oder via config/overrides.jsonPower Trend
when the index is in a confirmed, sustained uptrend
Trend-Regime & Entscheidungsgate
MAs die das Regime definieren + Quality-Schwellen
Exposure nach Marktgesundheit
wie viel Kapital der Engine je Gesundheitsgrad einsetzt
Risikoplan
Positions-Disziplin je Position (Einstieg / Stop / Ziel)
Markt-Playbook-Regeln
abgestufte Negativsignale die Exposure reduzieren — aus Analysten-Quellen, config-gesteuert
An extended index closing below its 21-day EMA is the first sign the run is stalling — sharper/severe when it's well below on heavy volume.
Loss of the 50-day / 10-week line is a bigger character change — the line institutions defend.
A cluster of distribution days (index down >=0.2% on higher volume than the prior day) = institutions selling. 5+ in a few weeks pressures the uptrend.
2s10s inversion (10Y < 2Y): late-cycle context (does not reduce exposure by default).
HY OAS at/above the stress threshold widens → risk-off (counts as a negative).
The 3 recommendation paths
each row shows which path triggered (① ② ③) — they're mutually exclusive by market context
| Path | When (context) | Trigger | Idea |
|---|---|---|---|
| ① Breakout | Healthy uptrend (close > 200-MA, 50 > 200) | Breaks out of a tight base on volume ≥ 1.4× average | Momentum — ride a fresh leader |
| ② Pullback | Established uptrend (close > 50 > 200, 50 rising) | Dips to the rising 21-EMA and reclaims it | Add to a winner on the dip |
| ③ Reversal | Overdone selloff (≥ 25–30% off high / below 200-MA) | RSI recovering + higher low + breaks the down-trendline | Mean-reversion — beaten-down quality turning |
The Status carries the path number, e.g. Watch ③ = a confirmed reversal candidate — a reversal is never an auto-buy; you judge the turn. Each path is backtested separately (see Backtests) and judged on expectancy, not win-rate.
Two strategies — read them separately
every name carries two independent reads; never blended
Classic CAN SLIM: buy leaders that are already strong. Uses the blended Quality score (Fundamentals + Relative Strength). Paths ① Breakout & ② Pullback. A breakout with weak RS (e.g. CRM, RS 11) is not a buy — no leadership.
“Quality on sale”: a fundamentally good company that got beaten down (so its RS is low by definition). We grade it on the Fund score (EPS rating, SMR, margins, growth) — RS deliberately ignored. Path ③ Reversal, and it caps at Watch — never an auto-buy; you judge the turn. It also must not be a falling knife: if the live quality has collapsed (e.g. NOW — Fund 72 but live-Q 23) it reads Avoid, not Watch. This is intentionally not strict IBD.
In the Stocks table the Q · Fund column shows both at once (e.g. 87 · 95). When they disagree, that’s the point — one strategy may pass while the other vetoes. The expand row spells out which.
Pillar 2 · QualityScore (0–100)
Fundamentals 65% + Relative Strength 35%
Deliberately excludes the chart setup and the market — those are separate pillars. A high QualityScore says “this is a leader”, not “buy it today”.
Pillar 3 · The checks
all positive — green dot = the good condition is met, red = not met
How the Status is decided
checked top-to-bottom — the first matching rule wins
“Setup triggered” means the chart engine confirmed a 1-2-3, cup-with-handle or base breakout. “Not extended” gates out names that have already run too far past the pivot — that is why a triggered base that has run +15% reads Hold, not New buy. This same Status drives every list in the app, so a ticker never reads Avoid on one page and Hold on another.
The rules above are the Strategy-1 ladder (① Breakout / ② Pullback). A ③ Reversal is the exception: it can never earn New buy or Add. The best it reads is Watch(“you judge the turn”), and only when the fundamentals are sound andthe live quality hasn’t washed out — otherwise it’s a falling knife and reads Avoid.
Confidence (on Signals) is separate: it comes from the backtest of that setup family — High / Medium / Low / Experimental, always read with the sample size.
Glossary
every term used in the appMethod
Market → Theme → Quality → Setup → Risk. We only take fresh risk when the market allows it, in a leading theme, on a high-quality name, at an actionable setup, with a defined risk plan.
Pillar 1 · Market
IBD market state: index low above the 21-EMA ≥10 days, 21-EMA above the 50-SMA ≥5 days, 50-SMA rising, positive close on the trigger. Permits maximum exposure.
Pillar 2 · Quality (ratings)
0–100 leader score from Fundamentals + Relative Strength only. Excludes setup and market by design.
Shorthand for QualityScore (0–100): Fundamentals + Relative Strength. The leader score.
Overall leader score (our Quality). Higher = stronger combination of fundamentals + relative strength.
Earnings strength, 1–99. Recent + long-term earnings-per-share growth vs all stocks.
Relative Strength vs the benchmark — persistent price strength, not a timing trigger.
Relative Strength rating, 1–99. Price performance vs all stocks over the last year.
Strength of the stock's industry group. Leaders tend to come from top-ranked groups.
Sales + profit Margins + Return on equity, A–E. A = strongest profitability profile.
Accumulation/Distribution, A–E. Are institutions net buying (A/B) or selling (D/E) lately?
The engine's composite leader score (0–100): a blend of relative strength, fundamentals, setup quality and risk checks. The single headline read for a name.
RS-free business-quality score (0–100): earnings power, margins and growth — the company itself, ignoring price action. High here + low RS = a quality name 'on sale'.
Whether institutional demand is confirming the move — volume expanding on up-moves vs the average. Strong demand backs a breakout; thin volume warns.
How many of the chart/trend checks pass (above key MAs, RS line at highs, in buy range, not extended). The price-structure half of the read.
Strength of the stock's industry group / theme. Leaders tend to come from top-ranked groups — a tailwind when the group is strong.
Pillar 3 · Setup
The chart pattern right now: 1-2-3, base breakout, near pivot, watch, or none.
Sperandeo reversal: A (swing low), B (high breaking the downtrend line), C (higher low above A), confirmed on a close above B.
How far price is from the proper buy point. Negative = below (can buy); above ~5% = extended.
Too far above the proper buy point — poor risk/reward for a fresh entry.
Bullish checks (above 50-MA, RS new high, volume, in buy range) vs bearish ones. Green circle = the good condition is met.
Pillar 4 · Risk / Trade plan
The price range where a fresh buy is valid (at/just above the pivot).
Where you'd add to a winning position once it's working.
Protective exit. Cut the loss if price closes below this (Classic IBD: ~7–8% under entry).
Price levels to take profits or trim into strength.
Suggested position size as a % of the portfolio.
Expected holding period. 'Swing · weeks' = a swing trade held days to weeks, not minutes or years.
Signals & status
The final verdict (synthesis) — what to do, combining all pillars: New buy / Add / Hold / Watch / Avoid.
When this signal fired. Older signals fade ('aging') — a Monday call may be stale by Friday.
Signal strength rank — best opportunities first.
Backtest-derived bucket (High/Medium/Low/Experimental); always read with the sample size.
The industry theme/group the stock belongs to (context, not a score).
Market internals
How far price sits above its moving average. Large = extended (overbought), poor for fresh buys.
Average True Range as % of price — daily volatility. Higher = wider swings.
Distance below the 52-week high. Near 0% = at/near highs (leadership).
Stocks whose price actually printed a fresh 52-week high today — the honest breadth read. Different from the common 'new highs' figure that counts names merely within ~1% of their high (which inflates the count).
McClellan Summation Index — a running total of market breadth momentum (Francesco's gauge). Rising = breadth broadening (healthy); falling = breadth narrowing even if the index holds up (a warning under the surface).
Arms Index — (Advancing Issues / Declining Issues) ÷ (Advancing Volume / Declining Volume). Below 0.75: volume concentrating in advancing stocks (bullish). Above 1.25: volume concentrating in decliners (bearish / oversold). Murphy's gauge of buying vs selling pressure intensity.
Price/breadth divergence: price makes a new high but NASI (breadth) is falling = warning, sellers hiding under the surface. Reverse: price makes a new low but breadth improves = bullish early reversal signal.
The four-market model (Murphy): bonds vs equities (rising yields = headwind), dollar vs commodities (strong dollar = suppresses gold/commodities), credit spreads (HY OAS widening = equity stress). These are directional headwind/tailwind signals, not entry triggers.
Statistical regime (companion read)
A second opinion to the rule-based health: a model groups past days into bear / range / bull from return behaviour and gives today's most-likely state with a probability. If it disagrees with the rules (e.g. rules say uptrend but the model says bear + stressed), exposure is capped.
How today's volatility ranks vs its own history: calm → normal → elevated → stressed (by percentile). Rising into 'stressed' is an early risk-off warning — volatility clusters.
The model's estimate that the market will be in its bull state N days out — a probabilistic 'where is this likely headed' read, not a guarantee.
Honest scorecard (validation)
Average return per trade — the number that decides profit. A positive expectancy with a low win rate still makes money if winners outweigh losers.
Return per unit of total volatility (annualised). Higher is better, but it treats up- and down-swings the same.
The probability the edge is REAL once you account for how few trades there are and how many settings were tried. Below 0.9 = could just be luck. On small samples it's correctly near 0 — that's the point: it stops us trusting a pretty number.
Like Sharpe but penalises only DOWNSIDE swings — a fairer read for a strategy that cuts losers and lets winners run.
Return ÷ worst drawdown — how much you earn per unit of maximum pain. The key ratio for a strategy whose job is cutting drawdowns.
Conditional Value-at-Risk: the average loss on the worst ~5% of days — the tail risk that actually hurts, not the typical day.
How well-calibrated a probability forecast is (0 = perfect, 0.25 = always guessing 50%). Lower means 'when it says 70%, it happens ~70% of the time'.
A slice of history never used while building/tuning. Performance there is the honest test — in-sample numbers always look better than they really are.
How much of buy-and-hold's worst peak-to-trough fall the timed strategy avoided (1.0 = removed it entirely).
What fraction of buy-and-hold's return the timed strategy kept. Below 100% = it sat partly in cash; the trade-off is shallower drawdowns.
Chart bases (Pillar 3 setups)
Volatility Contraction Pattern (Minervini): a run of progressively SHALLOWER pullbacks on shrinking volume — sellers drying up — then a breakout on a volume surge.
A rounded base then a shallow handle; buy point at the right-rim high.
A W-shaped base; buy point at the middle peak between the two lows.
A powerful ~80%+ run in a few weeks, then a tight shallow flag. Rare and strong — but high-risk.
Three pullbacks, each bottoming higher than the last (a staircase up).
A long, shallow rounded base — gentler and longer than a cup.
How many bases deep the advance is. 1st/2nd-stage bases are the reliable ones; 3rd-stage-or-later are late and fail more often.
The stock-vs-index strength line at a new high — true leaders' RS line often leads price out of a base.