No ads. No sponsors. Self-funded. Built for you, not advertisers.

Ghost Man (current rules): performance report

The brand strategy. Over 15 years it trailed SPY; over the last five years it did better. Its results lean on today's quality labels (see the look-ahead caveat).

January 3, 2012 to the November 17, 2026 close · backtests computed Nov 18, 2026 · static report.

The rules. Buy when the weekly Ghost Wave turns up from 20 or lower while the stock trades near or below its 4-year (200-week) average; sell half at +20%, sell the rest on the take-profit signal or after 2 years. 5% of the account per stock, up to 20 stocks; idle cash earns T-bill interest.

Over these 15 years, Ghost Man finished behind SPY (median of 200 runs): $53,170 vs $73,974. Ghost Man alone trailed the S&P 500 over 15 years.

Growth of $10,000: Ghost Man (current rules) vs SPYGhost Man (current rules)S&P 500 (SPY)
$10k$20k$50k20122014201620182020202220242026 ■ S&P 500 (SPY)$73,974■ Ghost Man$53,236

Log scale: each step up the left axis is the same percentage gain. Weekly closes shown; hover or tap the chart to read a date.

$10,000 became (median of 200 runs)
$53,170SPY: $73,974
Per year (CAGR)
11.9%SPY: 14.4%
Worst drop
−40.8%SPY: −33.7%
Sharpe ratio (curve shown)
0.71SPY: 0.90
5-year windows beating SPY
16 of 41
Honest win rate
70.7%
Trades
15 a year
Average hold
92 weeks

Year by year vs SPY

Calendar-year returns
Calendar year201220132014201520162017201820192020202120222023202420252026*
Ghost Man (current rules)+6.5%+4.1%+4.0%−13.8%+33.9%+15.7%−1.8%+18.8%+4.8%+39.0%+6.9%+26.8%+2.0%+24.4%+18.5%
S&P 500 (SPY)+13.9%+32.3%+13.5%+1.2%+12.0%+21.7%−4.6%+31.2%+18.3%+28.7%−18.2%+26.2%+24.9%+17.7%+8.2%
Difference vs SPY−7.4 pts−28.2 pts−9.5 pts−15.1 pts+21.9 pts−6.0 pts+2.7 pts−12.4 pts−13.5 pts+10.2 pts+25.1 pts+0.6 pts−22.9 pts+6.6 pts+10.3 pts

*2026 through the November 17 close. 2012 starts from the January 3 open. Calendar-year returns come from the curve shown in the chart.

Caveats that matter most here

  • Trailed SPY over the full 15 years in the median run, and in 200 runs only a minority beat SPY.
  • Uses today's Blue Chip and Growth labels (look-ahead); on a point-in-time list it made less.
Read this before any number on this page.
  • Ghost Man alone trailed the S&P 500 over 15 years. $10,000 in the current Ghost Man rules became $53,170 (11.9% a year, median of 200 runs); the same money in SPY became $73,974 (14.4% a year).
  • Survivorship bias. The stock lists are companies listed today. Stocks that went bankrupt or were delisted between 2012 and 2026 are missing, so every stock strategy here looks better than it would have in real time. Our rough estimate is that this adds 1 to 3 percentage points a year; that is an estimate, not something these tests measured. Momentum Leaders is the most exposed, because it buys hot stocks, and some hot stocks later collapsed.
  • Look-ahead in the quality list. Ghost Man's results use today's Blue Chip and Growth labels, which already know which companies turned out fine. Run on a point-in-time list of the 500 most-traded stocks instead, the same rules made 9.3% a year (vs 11.9%) and beat SPY in 9 of 41 five-year windows (vs 16).
  • No taxes, fees or slippage. Every trade fills at the next open at the exact price. Momentum Leaders makes about 104 trades a year and nearly all its gains would be short-term (taxed as ordinary income); Ghost Man holds for about 92 weeks on average.
  • Many variants were tested on the same data. Six strategy families in several stock lists, plus 24 rule changes to Ghost Man. The best of many tries usually looks better than its future.
  • Past results don't predict future results. These are historical simulations, not real-money accounts. Education only, not financial advice.

Where every number comes from

Window: Jan 3, 2012 to Nov 17, 2026 (14.87 years). Backtests computed Nov 18, 2026 by Ghost Man Trades. The daily curves are published in this site's data folder (data/performance/equity_daily.csv) with the headline numbers (summary.json), exported from the verified backtest files below; ending values were rebuilt trade by trade from the ledgers and matched the simulator to the cent.

  • Ghost Man (current rules): bt2/lab/results.csv, lab/rolling.csv, lab/out/GM_Ghost_Man_V3+2yr__BlueChip+Growth__main.pkl (Strategy Lab, 2026-11-18); curve shown: seed 131 (ranked 100th of 200 by ending value), ending at $53,236.
  • Ghost Man, idle cash in SPY: bt2/edge/results.csv, edge/rolling5.csv, edge/runs_phase1.csv.gz, edge/verify_E01.txt (Edge study, 2026-11-18); curve shown: seed 33 (ranked 100th of 200; ledger rebuilt in edge/verify_E01.txt), ending at $66,392.
  • Momentum Leaders (HIGH RISK): bt2/lab/results.csv, lab/rolling.csv, lab/verify_S4_P.txt (Strategy Lab, 2026-11-18); curve shown: one path (no tie order), ending at $352,324.
  • 50% SPY + 50% Ghost Man: bt2/edge/mixes.csv, edge/rolling5.csv (Edge study, 2026-11-18); curve shown: E01 seed 33 curve mixed 50/50 with SPY, ending at $71,333.
  • S&P 500 (SPY): bt2/edge/prices/SPY.csv (Yahoo, dividend-adjusted), bought at the 2012-01-03 open.

How the numbers were made: $10,000 starts on January 3, 2012 and is valued every trading day through the November 17, 2026 close. Yahoo Finance daily prices, adjusted for dividends and splits. Decisions on a candle's close, trades at the next open, each stock position 5% of the account (Momentum Leaders: 20 equal slices), one position per stock, fractional shares. When more signals print on one day than there is cash for, which ones get bought changes the result, so Ghost Man was run 200 times with the same-day signals in random orders. Tables show the median run (and the 10th–90th percentile range); the chart shows the run whose ending value ranks 100th of 200. Momentum Leaders has no ties to break, so it has one run. Honest win rate counts positions still open at the end below their cost as losses. 5-year windows: 41 overlapping five-year periods starting every quarter from January 2012 to 2021 plus November 17, 2021, each compared with SPY over the same dates. Sharpe ratio: average daily return ÷ its standard deviation × √252, with no risk-free rate subtracted (QuantStats' default definition).

Credits

Report statistics follow the definitions in QuantStats by Ran Aroussi and were cross-checked against QuantStats 0.0.86. QuantStats is licensed under the Apache License, Version 2.0; this site renders its own static pages and does not ship QuantStats code. Price data: Yahoo Finance. No paid data services.

All strategy reports · The Ghost Report · What is survivorship bias?