LevSig launched with 24 leveraged ETFs in its watch universe. As of July 9, 2026, we cover 10. This is the story of how the data told us to cut — and why fewer signals produced materially better strategy performance.
The starting universe
The original 24 covered a mix of index-based leveraged ETFs (TQQQ, TECL, SOXL, UPRO, LABU) and single-stock leveraged ETFs (NVDU, HIMZ, AMZU, TSLT, PLTU, and 14 others). The pitch was breadth: signals across every major leveraged play, from broad-market to sector to single-stock.
Breadth was the wrong metric.
What the backtest said
We ran a 4-year backtest of the LevSig signal logic — 5/20 EMA crossover, MACD histogram confirmation, 1.5× ATR trailing stop with intraday touch, RSI overbought exit — across every underlying in the universe.
Results on the full 24-ticker universe:
FULL 24-TICKER UNIVERSE — 4-YEAR BACKTEST
Profit Factor: 1.36
Max Drawdown: -46.7R
Aggregate Win Rate: 39%
Edge exists but is thin. Drawdown risk is significant.
Not terrible, but not the strategy we wanted to sell.
The pattern in the losers
Per-ticker analysis surfaced something clear: 6-7 underlyings had negative total R over the backtest window. They weren't just underperforming — they were actively dragging the strategy down. Together they accounted for roughly -21R of the -46R aggregate drawdown.
The pattern: low-momentum large caps and choppy names. AMZN, MSFT, PLTR, UBER, MSTR, CRWV, XBI. Each had good reasons to trend on paper. None trended cleanly enough for a momentum strategy to profit.
This wasn't a single-stock vs index problem. XBI is index-based and was on the loser list. AAPL and GOOGL were positive but marginal. The pattern was trend quality, not category.
The curation
We removed the negative-R names and re-ran. Then we added two index ETFs whose underlyings scored strongly in isolation (DIA, XLF) and cut a few marginal winners that weren't earning their spot.
The final 10 underlyings:
| Underlying | LevSig ETF | Category |
|---|---|---|
| QQQ | TQQQ | Nasdaq-100 |
| XLK | TECL | Tech sector |
| SOXX | SOXL | Semiconductors |
| DIA | UDOW | Dow 30 |
| XLF | FAS | Financials |
| NVDA | NVDU | Single-stock |
| HOOD | ROBN | Single-stock |
| HIMS | HIMZ | Single-stock |
| MU | MUU | Single-stock |
| NBIS | NBIG | Single-stock |
What the curation is worth
Same signal logic, same time window, curated universe:
| Metric | Full 24-ticker | Curated 10 | Delta |
|---|---|---|---|
| Profit Factor | 1.36 | 2.01 | +48% |
| Max Drawdown | -46.7R | -15.3R | -67% |
| Win Rate | 39.0% | 45.9% | +7 pts |
| Negative-R Tickers | 6-7 | 0 | — |
Same rules. Same market. Different universe. Materially better strategy.
What we didn't do
We didn't overfit. The 5/20 EMA, 1.5× ATR, and MACD histogram signals are the same rules that ran on the original 24 — we only changed which underlyings we apply them to.
We didn't cherry-pick a favorable window. The 4-year backtest includes the 2022 bear, the 2023 chop, and the 2024-2025 tech run. All three regimes are represented.
We didn't remove single-stock ETFs to make the strategy look safer. Half of the final 10 are single-stock leveraged plays (NVDA, HOOD, HIMS, MU, NBIS). We kept them because the data said they work.
What this means for subscribers
Fewer signals per day. The curated universe produces roughly 1-2 new signals per week rather than 3-4 across the broader watchlist. That's a trade we're comfortable with — quality over noise.
More conviction per signal. Every underlying in the current universe demonstrated positive risk-adjusted returns over 4 years of backtesting. When LevSig flags an entry, it's on an instrument where the strategy has real edge.
A defensible track record from here forward. The prior track record reflected trades on tickers we no longer cover. It wouldn't be honest to keep it — so we reset. Every closed trade on our performance page from July 9, 2026 forward reflects the curated universe.
Why we're publishing this
Because most signal services would quietly change their universe and hope no one noticed. That felt wrong.
The strategy is only as good as the data behind it. When the data says to cut, we cut.
Frequently asked
Why does LevSig only cover 10 ETFs?
A 4-year backtest across 40+ leveraged ETF underlyings identified 10 with the strongest risk-adjusted returns under our signal logic. Underlyings with negative total R or excessive drawdown contribution were removed. The curated universe delivered a profit factor of 2.01 vs 1.36 for the broader watchlist — with 67% less maximum drawdown.
Will you add more ETFs to the universe?
Possibly, if new candidates demonstrate the same edge as the current 10. We track potential additions in a watch list and evaluate them against the same 4-year backtest criteria. Additions require demonstrated edge, not just breadth.
Why were some leveraged ETFs removed?
Backtesting identified 6-7 underlyings that produced negative total R over the 4-year window. These names — primarily choppy or low-momentum large caps like AMZN, MSFT, PLTR, UBER, MSTR — dragged aggregate performance without contributing edge. Removing them lifted overall profit factor while preserving universe diversity.
Does a smaller universe mean less opportunity?
In raw signal count, yes — fewer tickers means fewer weekly signals. In quality-adjusted terms, no — the curated universe produces significantly better risk-adjusted returns. For leveraged ETF subscribers, fewer high-conviction signals typically outperforms more low-conviction ones.
Is the backtest methodology available?
Yes — the LevSig signal logic (5/20 EMA + MACD histogram + 1.5× ATR trailing stop + RSI overbought exit) is documented across our Learn section. The backtest uses adjusted daily bars, assumes intraday-touch stop fills, and applies the same universal parameters across all tickers. No parameter optimization was performed for individual underlyings.
Get the curated signals daily
LevSig runs its signal system on 10 curated leveraged ETFs every trading day at 4:30pm ET. Free tier includes all signals — entries, ATR trailing stops, RSI, and decay risk.
GET FREE SIGNALS