The 6-year backtest, with the table you should see before the marketing copy.
Three equal-weighted, correlation-optimized strategies (momentum, mean-reversion, regime-aware trend) versus a buy-and-hold SPY benchmark. Same window, same fees, same fill model. The methodology below shows every assumption.
Growth portfolio vs. SPY benchmark
| Metric | TradeLocus Growth | SPY buy-and-hold | Delta |
|---|---|---|---|
| CAGR | ▲ 21.4% | ▲ 18.1% | +3.3 pts |
| Max drawdown | ▼ 19.7% | ▼ 24.5% | ▲ 4.8 pts shallower |
| Sharpe (rf=4%) | 1.18 | 0.81 | +0.37 |
| Sortino | 1.74 | 1.12 | +0.62 |
| Worst month | ▼ 8.3% | ▼ 12.5% | ▲ 4.2 pts shallower |
Portfolio composition
Equity curve · same $100k starting capital
Every closed trade gets a verdict: what worked, what cost you, what to change.
The Autopsy decomposes the gap between your backtest's expected P&L and the trade's actual P&L into four causes — market drift, execution drag, override drag, and strategy decay. The biggest cause is named in plain English.
MeanRev-RSI(2) · TSLA · 2026-05-14
expected +$184 · actual −$71
market −$22 · exec −$118 · override −$94 · decay −$21
Cause: mid-quote fill assumption vs 38 bps real spread.
Backtest at the spread you actually pay — not at the mid.
Most platforms quote a backtest at mid-quote. Real fills happen at the spread plus broker latency. TradeLocus measures your broker's actual fill drag from connected accounts and re-runs your backtest at the price you'd really get.
Connected broker · 2,143 fills observed
mid → real fill: −38 bps avg
backtest CAGR @ mid: +24.1%
backtest CAGR @ real fill drag: +21.4%
Gap explained: 2.7 pts of "unexplained underperformance."
Every retired strategy is autopsied and archived. Yours stays yours.
Strategies that fail validation or get retired don't disappear — they go to the Graveyard with a full post-mortem (what broke, when, in what regime). You learn what doesn't work in your regime, not what failed for somebody else.
47 strategies retired · 12 reborn after fixes
most common cause: look-ahead bias (31%)
second cause: fill assumption (24%)
Survivor stats: the 18 in the live book are the 18 that survived this filter.
How the autopsy assigns blame
Every closed live or paper trade is compared against what the backtest expected for the same signal, and the gap is split into four additive causes:
- Market drift — the underlying moved between signal and fill; the part of the gap explained by price movement you could not have traded.
- Execution drag — spread, slippage, and latency: the difference between the modeled fill and the fill your broker actually reported.
- Override drag — the cost of manual intervention: exits taken early, entries skipped, or size changed versus what the strategy specified.
- Strategy decay — the residual: systematic underperformance versus the backtest that persists after the first three causes are removed, the signature of an edge going stale.
The largest cause is named on the trade card in plain English, and cause totals are aggregated per strategy so a pattern (say, persistent execution drag) is visible before it compounds.
How fill calibration works
A backtest priced at mid-quote assumes fills you will not get. With a connected brokerage account, TradeLocus observes your actual fills and measures the average gap between the quoted mid at signal time and the price your broker reported — your personal fill drag, in basis points, per symbol and time of day.
That measured drag replaces the generic slippage assumption: the backtest re-runs with entries and exits adjusted by the drag you actually pay, so the equity curve you evaluate is the one your account could plausibly have produced. No connected account means no calibration — the backtest states its slippage assumption explicitly instead of implying precision it does not have.
What happens to a retired strategy
A strategy that fails validation or gets retired is archived with a post-mortem: which check it failed (walk-forward robustness, Monte Carlo downside, live decay), when it broke, and what market regime it broke in. The record keeps the full parameter set and test history, so a fix is testable against the same bar of evidence that retired it — some come back, most stay buried.
The lessons are yours alone: your graveyard is built from your strategies in the regimes you traded, not pooled from other accounts. Its patterns feed the promotion checklist as advisories, so the mistake that killed one strategy is surfaced before the next one goes live.
Every number on this page is reproducible.
Strategy specs, parameter sweeps, slippage assumptions, fill model, and survivorship-bias-free data sources are documented and re-runnable on your own account. Commission $0 · 0.05% per-trade slippage · price returns · no look-ahead, no survivorship bias. The author is Aaron, founder & quant lead.
Hypothetical Performance Disclosure
Hypothetical performance results have many inherent limitations, some of which are described below. No representation is being made that any account will or is likely to achieve profits or losses similar to those shown. In fact, there are frequently sharp differences between hypothetical performance results and the actual results subsequently achieved by any particular trading program.
One of the limitations of hypothetical performance results is that they are generally prepared with the benefit of hindsight. In addition, hypothetical trading does not involve financial risk, and no hypothetical trading record can completely account for the impact of financial risk in actual trading.
This disclosure follows NFA Interpretive Notice 9025 guidelines for hypothetical performance presentations.
Engine: TradeLocus production backtest engine (master branch). Sharpe calculation: (mean_daily_return / std_daily_return) × √252. Monte Carlo: 1,000-path bootstrap resampling with $100,000 initial capital.