Strategy vs QQQ Review

All strategy curves are clipped to the same available window and normalized to start at 100. This makes the comparison about relative performance over the same dates.

Main window: 2024-05-08 09:00 -> 2026-05-08 08:00 Benchmark: QQQ buy-and-hold BTC V1-2 data: Alpaca Crypto 1-minute bars

Main Candidate Curves

Main candidate strategies versus QQQ normalized equity curves

All Local Curves

All local strategies versus QQQ normalized equity curves

Main-Window Results

Highest Return
+73.66%
Wei ETH NR 1h p=22 long_only
QQQ Benchmark
+63.40%
Same date window, daily close series
Best Sharpe
9.94
BTC V1-2 window360 on Alpaca 1m data
Lowest Drawdown
-0.17%
BTC V1-2 window360 preliminary PnL model

Ranking

Strategy Return MaxDD Sharpe Read
Wei_ETH_NR_1h_p22_long_only +73.66% -17.24% 1.42 Highest absolute return among main candidates; beats QQQ but drawdown is material.
QQQ buy-and-hold +63.40% -22.77% 1.05 Strong baseline over this period.
Wei_ETH_NR_1m_current_best +29.62% -12.23% 1.09 Lower drawdown, but far below QQQ return.
BTC_ETH_ZScore_1h +28.77% -34.51% 0.51 Too much drawdown for the return profile.
BTC_V1_2_window360_alpaca_1m +23.63% -0.17% 9.94 Uses Alpaca Crypto 1-minute BTC/USD and ETH/USD; looks extremely smooth, so execution/PnL assumptions need a second audit before launch.
BTC_V1_2_window720_alpaca_1m +12.42% -0.34% 4.61 Same family, slower window; lower return but still high risk-adjusted profile in this model.
BTC_CTA_Donchian_stored_1m -37.57% -49.12% -0.59 Current stored run is not acceptable.
BTC_ETH_ZScore_stored_1m -99.98% -99.98% -8.71 Stored minute run is broken or badly specified; keep out of shortlist.

What NR Means

NR is Normalized Return: EWMA(log return) / EWMA(abs(log return)). It measures directional persistence after scaling by recent volatility. High positive NR means recent returns are persistently positive; values near zero mean direction is weak.

Current Takeaway

BTC V1-2 is promising on 1-minute Alpaca crypto bars by Sharpe and drawdown, but its return is below QQQ in this conservative spread-PnL model. The next check is to reconcile this model against the original live-monitor accounting and KuCoin/Binance execution assumptions.