One stock a month. Buy price, sell target, and stop-loss sized to each stock’s own volatility.

The Max strategy scans 96 stocks every market close and names one top-ranked pick with a concrete buy price, a sell target scaled to the stock’s 14-day ATR, and a matching stop-loss (both capped at +25% / −12%). If neither fires within 12 months, close at market. Backtest 2006–2026: +29.3% CAGR vs SPY DCA’s +6.6% — Calmar 0.71, 60% win rate, avg winner +23%, avg loser −12%. For DCA investors who want to know which stock this month and when to sell. Not financial advice.

+29.3% 20-year CAGR, ATR-scaled TP (SPY DCA: +6.6%)
0.71 Calmar (SPY DCA: 0.19) — risk-adj excess
+22.6pp Annual excess over SPY DCA, 2006–2026
−41% Max drawdown (2008 GFC), Sharpe 1.28
This month’s picks

The Max strategy’s picks for this month, ranked by the CAP5+SMA12M opportunity score (trailing 12-month mean of conviction). The headline card is the take-profit play: buy the top-ranked stock at today’s close. The sell target and stop-loss are sized to each stock’s own 14-day ATR — tight around low-vol names like KO/PG, wider on high-vol names like SMCI/AMD — capped at +25% / −12% to preserve tail protection. Close at market after 12 months if neither hits. Backtested 2006–2026: +29.3% CAGR, 60% win rate, avg winner +23%, avg loser −12%. Rows 2–5 show the next picks with their own TP/SL — if you’d rather spread $1,000 across 5 names (CAP5 rank-weighted DCA, hold-forever variant), use the allocation column. These picks come from today’s scan; they are not financial advice.

Computing picks…
Stocks scanned daily
8 Forward horizons from 10 trading days out to 5 years
Last scan completed after market close, New York time
The Take-Profit Strategy, In Seven Rules Concrete buy, sell target, and stop-loss each month
  1. Contribute on the first trading day of each calendar month. No market timing; the strategy runs every month whether markets are up or down.
  2. Rank all 96 stocks by the CAP5+SMA12M opportunity score. Trailing 12-month mean of conviction (quality × 10D recovery probability × pullback gate). Smoothing stabilises the ranker around “persistent quality at fair value” instead of chasing the deepest single-day pullback.
  3. Buy the top-ranked stock at the next trading day’s close. One name, one entry. The one-bar delay matches the real scan-after-close cadence (scan runs tonight; trade fills tomorrow).
  4. Place a GTC limit sell at entry × (1 + 7 × ATR14%), capped at +25%. For a low-vol name (ATR14 ~1%) the target is ~+7%; for a mid-vol name (ATR14 ~2%) it’s ~+14%; for a high-vol name (ATR14 ≥3.5%) it caps at +25%. Each stock gets a target proportional to how much it actually moves.
  5. Place a stop-loss at entry × (1 − 7 × ATR14%), capped at −12%. Same scaling on the downside, with a tighter cap for tail protection. Low-vol names get tight stops (~−7%); everything else is bounded at −12% max loss.
  6. If neither hits within 252 trading days (~12 months), close at market. In the backtest, 0 of 86 trades time out — dynamic TP/SL covers the 12-month window fully. Still kept as a safety net.
  7. Next month, repeat. Every month the scanner names one new stock (top CAP5+SMA12M rank that isn’t a current open position). Median trade lasts just 24 days, so most months start with fresh capital.

Why this beats SPY DCA by 22pp. Fixed +10% / −15% targets misprice volatility: a +10% move on KO is 5σ rare, on SMCI it’s daily noise. Scaling by ATR lets each trade pursue a proportional move. In 20Y backtest, the 60% of trades that hit TP do so at avg +23% (not +10%), while capped stops cap avg loser at −12%. The win/loss magnitude ratio is ~2:1; combined with 60% hit rate, that produces +29.3% CAGR, Calmar 0.71, 41% MDD — winner in all 6 rolling 10Y windows.

Alternative: CAP5 rank-weighted DCA (hold-forever). If you prefer the older “spread $1k across 5 names, never sell” approach, it’s still visible in the weight column of the picks table. Backtest: +18.3% CAGR, gives no sell signal. The take-profit variant with dynamic TP/SL outperforms this by ~11pp CAGR while giving retail-friendly concrete entry/exit prices. Either approach is replicable from the published rules; neither is financial advice.

Portfolio Concentration If you’ve been DCA’ing since 2021, where does your basis sit today?

Snapshot of cumulative cost basis per ticker for a hypothetical investor who started this strategy in April 2021 and contributed $1,000/month. A ticker is “capped” when its share reaches 5% — subsequent months skip it and backfill from the next candidate. PnL% shown is the simple return on cumulative basis vs current position value.

Computing concentration…
20-Year Validation — Dynamic TP Strategy ATR-scaled TP/SL (k=7, cap +25/−12), 12-month time-stop, top-1 by CAP5+SMA12M, 2006–2026

Each month the top-ranked stock is bought at the next trading day’s close. The sell target is set at entry × (1 + 7 × ATR14% of price), capped at +25% max. The stop-loss is symmetric: entry × (1 − 7 × ATR14%), capped at −12%. Low-vol stocks (KO, PG) get tight ~±7% targets; mid-vol names (UNH, JPM) get ~±14%; high-vol names (SMCI, AMD) cap at +25%/−12%. If neither fires within 252 trading days, close at market (happened 0 times in backtest). Benchmark: monthly DCA into SPY.

Headline (2006 → 2026, ~20 years)
  • Dynamic TP CAGR +29.26% vs SPY DCA +6.63% → excess +22.63pp
  • MaxDD −41.0%, Sharpe 1.28, Calmar 0.714 (SPY 0.185 — 3.9× better risk-adj)
  • 86 trades: 60% hit TP (avg +23%), 40% hit SL (avg −15%), 0 time-stopped.
  • Avg 46 days held, median 24 days. Fast turnover = capital rotates through opportunities.
Rolling 10Y walk-forward (all 6 windows beat SPY)
  • 2006–2016: +17.9% vs SPY +4.3% (+13.6pp)
  • 2008–2017: +27.3% vs SPY +6.2% (+21.1pp)
  • 2010–2019: +21.6% vs SPY +6.1% (+15.4pp)
  • 2012–2021: +18.6% vs SPY +7.9% (+10.7pp)
  • 2014–2023: +25.4% vs SPY +5.7% (+19.8pp)
  • 2016–2025: +26.3% vs SPY +7.4% (+18.9pp)
Parameter sensitivity (did we just curve-fit?)
  • ATR multiplier k: swept k ∈ {3, 4, 5, 6, 7, 8, 10}; k=7 is the CAGR + Calmar peak. k=5 and k=10 both within 3pp CAGR.
  • TP cap: swept {20, 25, 30, 40, 100}%; 25% is the peak. Wider (30%+) reduces fill rate too much; tighter (20%) cuts winners short.
  • SL cap: swept {10, 12, 15, 18, 20}%; 12% is optimal. Tighter (10%) cuts too many would-be winners; looser (15%+) widens MDD.
  • Time-stop: 252 / 378 / 504 bars all produce identical results (no trades time out). TP/SL always fires.
Why this works
  • Per-stock sizing: fixed +10%/−15% mis-prices volatility. Scaling by ATR gives each stock a target proportional to how much it actually moves.
  • High-vol winners run: SMCI / AMD / NFLX often rally 15–25% in days. Fixed +10% locks in a fraction; dynamic +25% cap captures the move.
  • Low-vol discipline: KO / PG at +7% TP fires quickly and frees capital for the next pick. Trading frequency rises without cost drag because wins come in days.
  • Tail cap: −12% SL floor keeps avg loser at −15%. The 41% MDD is the accumulation of occasional stops during 2008-09, not a single-name disaster.

Honest caveats: universe-level survivorship bias (128-ticker roster = today’s survivors); the quality multiplier in the scoring feed uses today’s value at historical dates (a known concession — the rest of the score is strictly point-in-time); TP/SL fills assume limit orders fill exactly at target price when intraday High ≥ TP or Low ≤ SL (real fills can be worse in crises); idle-cash drag is real (20–30 bps/yr with a MMF at 4%); the −41% MDD is real — if you can’t hold a position down −12% and maybe 3–4 in a row during a crash, this isn’t for you. Past performance does not predict future results. Not financial advice.

Professional Disclosures If you’re thinking about deploying real capital, read this first
What this is
  • A rules-based, mechanical monthly-pick strategy. One stock per month, concrete buy/sell prices, no discretion, no intraday trading.
  • Validated out-of-sample against 20 years of price history across 128 large-cap US equities, covering the GFC, 2010s bull, COVID, 2022 bear, and the current cycle.
  • Run it yourself: the scanner is open, the backtest code is in max/research/, the monthly picks are published on this page.
What this is not
  • Not investment advice. Nothing on this page is personalized; nobody here knows your tax situation, risk tolerance, or liquidity needs.
  • Not a guarantee. 20 years of favorable backtest does not mean the next 20 years will repeat. A repeat of 2008 would produce a ~40% drawdown; the backtest says so explicitly.
  • Not a hedge fund. No shorting, no leverage, no overlays. Long-only equity, one name at a time.
Known limitations
  • Single-name concentration: only one stock held at a time. A fraud or sudden regulatory shock can breach the stop in one day; −12% isn’t a floor in a gap-down.
  • Survivorship bias: the 128-ticker roster reflects names that exist today. Delistings 2008–2024 aren’t in the universe.
  • Quality multiplier is not strictly point-in-time: the analog matching, probabilities, and edge are PIT, but the quality score uses today’s value. Results remain robust when the quality factor is held out.
  • TP/SL fill assumptions: TP fills when intraday High ≥ target; SL fills at the stop price when Low ≤ stop. Real fills in fast markets can gap through the stop — worse than modeled.
  • Drawdowns are real: the backtest hit −41% in 2008–2009. If that would force you to sell, this strategy is not for you.
  • Idle cash drag: 40% of trades hit stop in under a month; cash sits until next monthly entry. A money-market at 4% adds ~20–30 bps/yr that the backtest doesn’t credit.
Before deploying real capital
  • Review the Python backtester in max/research/ — especially step41 (fixed-TP grid), step44 (stop-loss sweep), step45–47 (dynamic-exit winner validation).
  • Run your own walk-forward on your own universe. The strategy is not magical; it’s a CAP5+SMA12M ranker with ATR-scaled exits. Reproduce the excess before you trust it.
  • Paper-trade for at least 6 months. The hardest part of a mechanical strategy is executing the monthly pick when the previous trade just hit its −12% stop.
  • Accept that 40% of trades lose. The headline is “60% hit TP at +23%”, not “never loses”. Your win-loss emotional tolerance matters more than the backtest CAGR.
  • Have an exit plan. If the strategy deviates from its backtested path by >3 standard deviations of historical tracking error, stop and investigate before adding more capital.
How to Read This The Numbers & Methodology
The numbers
  • Prob @ horizon — The chance this asset was higher after the selected horizon (10D, 30D, 60D, 3M, 6M, 1Y, 3Y, or 5Y) when it was in a similar historical pullback.
  • Typical — Median historical return at the selected horizon across similar past cases.
  • Downside — Bad-but-realistic outcome at the horizon (10th percentile — worst 1-in-10).
  • Pullback — How washed-out the asset is today. 0 = no pullback, 100 = extreme selloff.
  • Quality — Is this a strong asset in a temporary dip, or a weak one still breaking down?
  • Cases — How many similar historical pullbacks the stats are based on. More = more reliable.
  • Best horizon (Top Picks only) — The time frame where this asset's probability most exceeds the universe median at that horizon (biggest "edge"), weighted by how many analog cases back it up.
How it works
  1. Measure today's setup — Score each asset on drawdown, range position, RSI, and 200-day distance.
  2. Find historical matches — For each asset, pull every past day with a similar setup.
  3. Roll forward at eight horizons — Record the actual return 10, 30, 60, 63, 126, 252, 756, and 1260 trading days later (with survivorship correction).
  4. Pick the best horizon per asset — Top Picks ranks by the horizon where the analog-based edge is strongest.
  5. Backtest honestly — Use the point-in-time opportunity score (no look-ahead) to pick positions at each DCA date, then hold for the selected horizon.
Important
  • Not financial advice. This shows what happened in the past, not what will happen next.
  • Not a trading system. No buy/sell signals, position sizing, or timing.
  • No guarantees. Assets can keep falling — including to zero.
  • Long horizons (1Y / 3Y / 5Y) are filled in once enough analog matches exist. Crypto long horizons populate as history builds.
Top Recommendations

Every stock below is ranked by the CAP5+SMA12M opportunity score (trailing 12-month mean of conviction). This is the same scoring used to name this month’s pick. The TP% / SL% / ATR14% columns show the dynamic exit prices a retail investor would place if buying that stock today (GTC limit sell at +TP%, stop-loss at −SL%, close at market after 12 months if neither hits).

Top Stock Picks Ranked by CAP5+SMA12M score — this is the pool the top-1 dynamic TP pick is drawn from each month
Ticker Price TP (+%) SL (−%) ATR14% Pullback Quality Cases
Dynamic TP Backtest — Max Strategy (production) Monthly top-1 pick, ATR-scaled TP/SL (k=7, cap +25/−12), 12-month time-stop — benchmarked to SPY DCA

The production strategy: $1,000 every month into the top-1 stock by CAP5+SMA12M. Each trade gets a dynamic TP/SL sized to the stock’s own ATR14: TP = entry × (1 + max(5%, min(7 × ATR14%, 25%))), SL = entry × (1 − max(5%, min(7 × ATR14%, 12%))). Exit at first close ≥ TP or ≤ SL, else close at market after 252 trading days. Entry at next trading day’s close. Benchmark: SPY DCA on the same monthly cadence. Results below are precomputed server-side on each data refresh over the embedded recent-5Y spine. The full 20-year Python backtest (+29.3% CAGR, +22.6pp over SPY DCA) is summarised in the 20-Year Validation card above.

Running backtest…
Alternative Backtest — CAP5 rank-weighted DCA (hold-forever) For reference: if you split $1k across top-N, held forever, with 5% concentration cap — the prior production strategy

Prior strategy for comparison: $1,000 every month into the top-ranked stocks (by CAP5+SMA12M), split by rank weights (1/1, 1/2, 1/3… normalized so #1 gets ~44%, #5 gets ~9%). Entry at next trading day’s close. Positions held forever — no sell. 5% per-ticker concentration cap drops any name whose cost basis has hit 5% of total invested. Benchmark: SPY DCA. 20Y CAGR +18.3% for the CAP5 hold-forever; the current dynamic TP strategy (above) clocked +29.3% CAGR in the same period — see the 20-Year Validation card.

Click to load backtest data…
Stocks — Multi-Horizon Screener Curated 100-stock universe

Pick a horizon. The listing ranks every stock by its probability at that horizon in similar historical pullbacks.

Ticker Price Prob Typical Downside Pullback Quality Cases
Stocks Backtest — DCA at Selected Horizon Ranked by point-in-time opportunity score, benchmarked to SPY

$1,000 every month into the top-ranked stocks (by point-in-time opportunity score), held for the selected horizon, then sold. Benchmark: SPY DCA on the same schedule.

Click to load backtest data…