Composer · investment strategy
by Composer Technologies · via Composer
Daily, it hunts for overheated markets using RSI across major and leveraged ETFs. When things look too hot, it buys UVXY (long volatility). Otherwise it usually shorts volatility with SVXY—or, when tech leads, holds Apple. Very high risk.
Description from Composer
of the return and risk figures:as the source publishes them
What $10,000 would have become, from the strategy's own published history, against the S&P 500 over the same dates. The basis is labelled.
This strategy doesn’t publish a performance history we can chart.
Each rule is quoted word for word from the strategy's own documents.
No rules extracted from this strategy’s text yet.
1 positions as of 2026-10-11
: the platform's backtest holdings, not a client account.
Compares the stated rules with the holdings (or, where none are published, a mix estimated from returns), using tests fixed before any results were seen.
0.0% of weight in (IVV: 3.0%)
No funds
This strategy’s rules haven’t been extracted yet, so there’s no claim to compare with. The measurement is still shown. : 100% · 2026-10-11 ()
0.0% of weight in (SPY: 0%)
This strategy’s rules haven’t been extracted yet, so there’s no claim to compare with. The measurement is still shown. : 100% · 2026-10-11 ()
5 of 12 funds in its are : FAS, SVXY, TECL, TQQQ, UVXY
This strategy’s rules haven’t been extracted yet, so there’s no claim to compare with. The measurement is still shown. : 100%
Strategies that share its Aggressive risk label, from other sources first, strongest risk-adjusted return first.
Composer Technologies · Composer
Composer Technologies · Composer
How much of what we look for this strategy has, and why anything is missing. Fields that don't apply to it aren't counted.
4 missing
The text the rules are quoted from, with where it came from.
Each day it checks if tech (QQQ) is stronger or weaker than the market (SPY). It then scans many big/leveraged ETFs for very “hot” readings using RSI—a 0–100 gauge of recent price speed. If any read ~75–80+, it buys UVXY to bet on a near‑term volatility jump. If not hot: when tech is weaker it holds SVXY (short volatility); when tech is stronger it either buys UVXY after a big TQQQ up‑day (>4.5%) or else parks in Apple.
Daily, it hunts for overheated markets using RSI across major and leveraged ETFs. When things look too hot, it buys UVXY (long volatility). Otherwise it usually shorts volatility with SVXY—or, when tech leads, holds Apple. Very high risk.
Out-of-sample, this volatility-timing strategy hunts for spikes to boost returns, aiming for ~31% annualized vs SPY’s ~18%, with Calmar ~0.67. Higher drawdowns can occur, but the upside may outpace the S&P over time.
Using relationships between the tech-heavy QQQ, and standard of evaluation SPY, to determine the optimal strategy for catching pops in volatility, backtesting to late 2011.
Every value on file, each labelled with its calculation basis and as-of date.
| Backtest final value (from $10,000) | $32.8B | 2026-10-11 | |
| Backtest regulatory fees | $112.8M | 2026-10-11 | |
| Backtest slippage cost | $1.1B | 2026-10-11 |
| AI categories (Composer) | daily rebalanced, leveraged etfs, rsi momentum, tactical allocation, vix etps, volatility timing | 2026-10-11 | |
| Asset classes / instruments | equities | 2026-10-11 |
| Logic complexity (node count) | 87 | 2026-10-11 | |
| Rebalance frequency | daily | 2026-10-11 |
| Concentration (Herfindahl index, out-of-sample, as Composer reports it) | 1 | 2026-10-11 | |
| Ticker universe (assets the logic can hold) | show[ "AAPL", "FAS", "QQQ", "SPY", "SVXY", "TECL", "TQQQ", "UVXY", "VOX", "VTV", "XLP", "XLY" ] | 2026-10-11 |
| Alpha vs SPY, out-of-sample (out-of-sample) | 0.34 | 2026-10-11 | |
| Beta vs SPY, out-of-sample (out-of-sample) | 0.38 | 2026-10-11 | |
| Correlation (R) vs SPY, out-of-sample (out-of-sample) | 0.12 | 2026-10-11 |
| Remixes (copies) | 1 | 2026-10-11 | |
| Watchers | 0 | 2026-10-11 |
| Community review status | ACTIVE | 2026-10-11 |
| Backtest length | 3,775 | 2026-10-11 | |
| Out-of-sample length | 536 | 2026-10-11 | |
| Return, backtest period (annualized) (backtest) | 172.20% | 2026-10-11 | |
| Return, backtest period (cumulative) (backtest) | 328497k% | 2026-10-11 | |
| Return, out-of-sample period (annualized) (out-of-sample) | 31.40% | 2026-10-11 | |
| Return, out-of-sample period (cumulative) (out-of-sample) | 78.95% | 2026-10-11 | |
| Return, trailing 1 day (out-of-sample) | -1.11% | 2026-10-11 | |
| Return, trailing 1 month (out-of-sample) | 2.28% | 2026-10-11 | |
| Return, trailing 1 week (out-of-sample) | 0.88% | 2026-10-11 | |
| Return, trailing 1 year (backtest) | 18.41% | 2026-10-11 | |
| Return, trailing 1 year (out-of-sample) | 18.39% | 2026-10-11 | |
| Return, trailing 2 weeks (out-of-sample) | -1.30% | 2026-10-11 | |
| Return, trailing 3 months (out-of-sample) | 10.97% | 2026-10-11 | |
| SPY annualized return over the backtest window (backtest) | 15.79% | 2026-10-11 |
| Calmar ratio, backtest period (backtest) | 3.68 | 2026-10-11 | |
| Calmar ratio, out-of-sample period (out-of-sample) | 0.67 | 2026-10-11 | |
| Maximum drawdown, backtest period (backtest) | -46.78% | 2026-10-11 | |
| Maximum drawdown, out-of-sample period (out-of-sample) | -46.78% | 2026-10-11 | |
| Sharpe ratio, backtest period (backtest) | 1.96 | 2026-10-11 | |
| Sharpe ratio, out-of-sample period (out-of-sample) | 0.77 | 2026-10-11 | |
| Sortino ratio, backtest period (backtest) | 3.69 | 2026-10-11 | |
| Sortino ratio, out-of-sample period (out-of-sample) | 1.33 | 2026-10-11 | |
| Standard deviation, backtest period (backtest) | 59.58% | 2026-10-11 | |
| Standard deviation, out-of-sample period (out-of-sample) | 53.05% | 2026-10-11 |
| Annualized turnover, backtest period (backtest) | 58.48 | 2026-10-11 | |
| Annualized turnover, out-of-sample period (out-of-sample) | 59.24 | 2026-10-11 | |
| Daily win rate, backtest period (backtest) | 57.25% | 2026-10-11 | |
| Daily win rate, out-of-sample period (out-of-sample) | 55.41% | 2026-10-11 |
Composer Technologies · Composer
Composer Technologies · Composer
