🎯 AQ Trend Screener
A screener of trend leaders based on the Trend Template of Mark Minervini's methodology (SEPA): eight objective criteria that, according to their author, eliminate ~95% of the market and leave only the genuine leaders in Stage 2 advance.
You get in through Stocks → 🎯 Trend Screener. It scans your whole catalog in seconds; the filters refine without re-scanning. Right-click a row = that stock's chart.
The quick view: reviewing 50 stocks in a minute
As you move through the grid (by clicking or with the arrow keys), the side panel instantly draws the stock's last year: close, MA50 and MA200, the 52-week high line, the VCP pivot in green and the volume. It is the review flow of a professional watchlist: arrow, arrow, arrow… "I like this structure" → double-click (or click the mini chart itself) and AniQuant's full chart opens.
The 8 criteria and the AQ RS Rating
Every stock shows its 8 traffic lights: price above MA50/150/200, the averages correctly stacked (50>150>200), MA200 rising for at least 21 sessions, price less than 25% below its 52-week high, and an AQ RS Rating ≥ 70. The RS is the percentile (1-99) of the weighted 3/6/9/12-month return — computed against our entire universe, delisted stocks included: a relative-strength ranking with no survivorship bias, something few commercial screeners can claim.
Template (40) + RS (30) + regime (10), normalized. Technical criteria only: the score does not mix in fundamentals, which are applied separately as a filter.
Accumulation / Advance / Distribution / Decline, detected from the price's position relative to the MA200 and its slope. SEPA only trades Stage 2 — the filter comes ticked by default.
The breadth of the universe itself (% of stocks above their MA200/MA50) as a proxy for the environment: Minervini does not trade against the market — on 🔴, go to cash.
The screener also detects the Volatility Contraction Pattern: successive pullbacks each one smaller than the last (18% → 11% → 5%…) with the volume drying up — the fingerprint of supply running out while institutions accumulate. Each stock shows its state: 🌀 forming, 🎯 READY (the last contraction tight and one step from the pivot), 🚀 confirmed breakout (it cleared the pivot on volume ≥40% above average) or ⚠️ weak breakout (it broke out on mediocre volume: statistically, far more likely to fail). Breakouts already running more than 5% above the pivot are flagged +N% ⛔ (extended): the entry was at the breakout — chasing it moves the stop away from the structure, and the screener reminds you by pushing them down the ranking. The pivot is drawn in green in the quick view, and the pattern's quality contributes up to 20 points to the Score (full scale: Template 40 + RS 30 + VCP 20 + Regime 10).
What it is NOT (yet): the E of Earnings (EPS, sales, margins) and the catalyst require fundamental data that will arrive in a future phase. Reference: Trade Like a Stock Market Wizard (M. Minervini, 2013) — the methodology is cited, not affiliated.
AQ Legends: the duel of the legendary strategies
In Stocks → ⚔️ AQ Legends, the famous methodologies stop being promises from a book and become comparable numbers: you tick the legends you want to pit against each other and they all run on exactly the same arena — the complete US universe with no survivorship bias (delisted included), the same capital, the same costs and the same competition for a slot in the portfolio. The result is a fair duel: each legend's equity overlaid, a ranking with medals (and the risk columns that say at what cost it was won), and full detail per strategy — cards, IS/OOS validation and trade log — choosing with one click which legend to inspect.
The price-and-volume-only legends: SEPA · VCP Breakout (Minervini style) — trend template + VCP + entry at the close of the pivot breakout on confirmed volume, a 7-8% stop (gaps exit at the real open, with no window dressing) and a MA50 trailing stop; and Momentum 12-1 (Jegadeesh & Titman, 1993) — the most documented factor in academic history: buy what rose most over 12 months excluding the last one (which tends to revert), rotating the portfolio at each month end; no stops — its risk management is the rotation itself. And 52-Week High (George & Hwang, 2004) — buy what closes less than 5% below its yearly high (proximity filters; momentum ranks — we measured that ranking by proximity bought the flattest stocks in the market), monthly rotation: the anchoring anomaly, the academic cousin of Minervini's pivot. And the market wizards: Weinstein-style Stage 2 — the Stage 1→2 transition across the 30-week average with volume expansion, selling on the first close below it (the most patient trailing stop in the duel) — Darvas Box — boxes at 52-week highs (the ceiling and floor fixed after 3 days without being cleared or pierced), buy on the breakout with volume and the stop rising with each new box: the original 1960 trailing stop — and O'Neil-style Growth Breakout (technical) — a base of 7+ weeks with an 8-35% correction, a pivot breakout on volume, and its textbook exits: an 8% stop, profit taken at +20% except for the 8-week rule (if it gets there in under 3, you hold the full 8) and a defensive sale below the 50-day line. Its letters C and A (quarterly earnings and annual growth) can now be demanded since the fundamentals arrived; the I (institutional sponsorship) still has no data, which is why we do not simply call it CAN SLIM. Piotroski is complete signal by signal —what is missing is his original high book-to-market universe, because we do not have book value—, and the Magic Formula is still pending.
🥊 Beat the Masters — the ANIQUANT legend. The third contender is you: write your strategy's ENTRY in the AQ language (the Lab's own — one condition per line, // for comments), choose its exits (stop %, MA50 trailing, exit after N bars — whichever comes first) and tick its checkbox: your system competes against Minervini and academia under the same rules of the game — the same universe, capital, costs and selection. Your code and exits are remembered between sessions. On top of that, the AQ filters in the editable SEPA code are demanded at the entry of every legend in the duel (yours included).
«When the market is in correction, I go to cash»: it vetoes entries with the market broken. The Breadth model (% of the universe above its MA200, previous week — no peeking at the future) or O'Neil's Follow-Through Day model on SPY: a rally attempt whose day 4+ rises ≥1.5% on higher volume = market CONFIRMED; 5 distribution days in 25 sessions = back to correction.
Each legend's first pass walks ~25 years of the universe (minutes). After that, changing capital, positions, costs, regime or IS/OOS dates re-simulates the whole duel instantly. Changing the SEPA rules only re-collects the SEPA legend (Momentum does not even notice); changing the AQ filter or the universe re-collects them all.
Equity on a log scale against two yardsticks (detail below), a random selection band (merit or luck?) and IS/OOS validation with fresh capital per stretch and a verdict in plain language.
It demands a company as well as a chart: quarterly EPS growing ≥X% year on year (O'Neil's C), sales, debt/equity, margin. Every signal is judged with the last quarter published before its entry —zero lookahead— using Sharadar's point-in-time fundamentals. The star recipe: SEPA + EPS≥25% = full SEPA. It applies to every legend and, like the regime, re-simulates instantly.
⚪ Equal-weight universe (the gray line) — the THEORETICAL CEILING. Each month the month-end close of every live stock in the universe is taken (delisted included, contributing their collapses until they disappear — point-in-time, no survivorship bias) and the index advances with the average of their returns: an equal-weight of thousands of stocks rebalanced monthly at no cost whatsoever. That rebalancing is a theoretical buy-low-sell-high machine (the "rebalancing premium") amplified by microcaps — which is why its CAGR comes out sky-high. Nobody can buy it: it would take thousands of positions and a massive monthly rotation with zero friction. Use it as the raw material's ceiling, never as an investable alternative.
🟡 SPY (the golden line) — the FAIR BAR. The S&P 500 ETF bought on day one and held: value = capital × adjusted close ÷ initial close. The adjusted close includes reinvested dividends (total return), so it is the complete and honest comparison: what anyone can genuinely do with one click. It has existed since 1993.
🔄 Where the SPY data comes from (and how it keeps itself current). If SPY is in your catalog, its BIN is used (and the Manager's «Update quotes» keeps it current, like all the rest). If it is not — "Common Stock only" imports exclude ETFs —, AniQuant downloads it automatically with your EODHD key and saves it in datos\SPY.bin. That file has a 30-day expiry: when its last bar ages beyond a month, it is re-downloaded in full on the next run (you will see «updating SPY» in the progress — always the whole series, never a patch: adjusted prices shift backwards with every dividend). If the download were to fail (no key or no network), the existing file is used: better a stale benchmark than none, and it retries next time.
🔵 The legends' portfolios use realized equity (the P&L is booked at each trade's close): which is why a winner that ran for months shows up as a vertical step on its exit date — that is not a data error, it is the engine's accounting convention. And mind the comparison of philosophies: Momentum 12-1 pays rotation costs every month (even when it «holds» a stock, which is sold and rebought — the conservative version), whereas SEPA lets it run; slippage is the duel's great silent referee.
📜 The method's code — visible and EDITABLE. The panel shows the complete strategy as a formula (professional screener style), generated live from the parameters: no small print. And with ✏️ you can edit it as code: change the values in the text, delete the trailing line to disable it, or add your own AQ conditions in the additional filters section (one per line, // for comments — the same language as the Lab's AQ editor); on Apply, the code is poured into the controls and the extra conditions are demanded (AND) on every breakout bar. And the «E» is no longer a promise: the fundamental filter (above) adds the earnings/sales/debt rules with Sharadar's point-in-time data, honest by publication date.
🔍 Inspecting everything. The chart has a crosshair (move the cursor and you will see the date and return of every curve in the duel at that instant) and its legend is a control panel: each line has its checkbox to hide it — remove the Universe and the scale zooms automatically onto the duel (your selection is remembered). The colored chips above the cards choose the active legend: its tiles, its IS/OOS validation and its trade log. The Trades card has three tabs: the active legend (the full trade log — a virtual grid filterable by ticker, sortable by any column, keyboard-navigable and with a double-click that opens the stock's chart, reusing the window), SPY by month and Universe by month (the benchmarks have no "trades": their honest detail is the month-by-month return, with the universe's live stocks at each date: point-in-time in action, from ~2,000 stocks in 1991 to ~6,000 today).
🏁 The race by years. A table with each calendar year's return per strategy: the cell shaded in its color is that year's winner among the legends and SPY (the Universe runs outside the contest), with the 🏆 years won and total CAGR rows at the foot. This is where each philosophy's character shows: who wins in the trending years, who survives the crashes and who shines precisely when the others bleed.
🎯 The companies that would get in. Tick the box, pick a date (TODAY by default) and each legend lists its last known batch of entries as of that date — point-in-time: only what was known that day, with the duel's own filters, delisted excluded and sorted by its own ranking (the first N 🎯 would fill the portfolio's slots). And it is a time machine: enter 15 March 2009 and you will see what each method would have bought coming out of the crash. Instant — it uses the cached signals.
🤖 The duel's AI Report. With your Claude API key, the ✨ button generates an extensive, educational report on the whole duel: what has been measured and why the arena is honest, the podium, each strategy analyzed (with the reading of its band percentile: does its selection add value, or is it worse than chance?), the comparisons between philosophies, the role of costs and regime, the methodological caveats and specific experiments to try in the window itself. Skeptical by design, and it never recommends trading for real.
💾 Saving a legend as a portfolio strategy. In the row where you choose which legend to inspect, at the far right, 💾 Save appears after running the duel — before that there is nothing to save. And saving matters: without it, a legend would not be a thing, but something living in the browser's memory, and clearing that would erase the configuration that has just beaten eleven legendary methods. Saved, it is an object with an identity of its own (AQC-XXXXXXXX), a name, notes and a date, holding everything needed to reproduce it: the rule, the method's parameters and also the arena —capital, slots, per-position cap, minimum liquidity, commission and slippage— plus the regime and fundamental filters. That last part matters: two portfolios with the same rule and a different number of slots are not the same strategy, they are two. The metrics are stored as a snapshot with its date, not as a promise. Below, in the left column, the 💼 Saved portfolio strategies card lists them, loads them (pouring the parameters and arena back into the window: all that is left is to press Start) and deletes them.
A legend is not a single-symbol strategy: it is a portfolio policy — a ranking rule over a whole universe, a number of slots and a rebalancing cadence. Pine, EasyLanguage and the rest think in terms of «one symbol, one chart, one position» and have no such concept, so there is nowhere to export it to. Its place of execution is AniTrader, which is what talks to the broker and can hold ten positions at once; this format is the first stone of that bridge. Single-symbol strategies —Lab, Genesis, Library— still export to the five languages, as always.
Momentum 12-1 usually shows up as the winner of any factor duel. Here it is not, and the reason is not in the rules —they are the original paper's— but in what it is allowed to buy.
The universe tells the type of security apart by the category the provider declares, not by guessing it from the ticker's suffix. If it were guessed, warrants, units and duplicate series would find their way into the portfolio, and they are not ordinary shares — and Momentum loves that, because it buys whatever rose most over twelve months and a warrant multiplies by ten easily. Measured with the badly filtered universe, its second largest contributor was BLNKW: Blink Charging's warrant, not the stock.
How much that weighs: 12% of everything that dirty universe made was put there by securities you cannot buy. On the bare factor —the same 336 months, the same rules, no costs— that is the difference between a CAGR of 16.5% and one of 6.5%. The high number was not a better strategy: it was a strategy buying smoke. The one here is the honest one, and it matches what the literature says when momentum is measured with costs and without exotic instruments.
AniQuant can be tried free for 30 days, with every module and no card.