Test the strategy. Find where it breaks.
Chat2Invest publishes reproducible research on low-frequency ETF and portfolio strategies — including the assumptions, failure periods, trade-offs and evidence behind each result. Not a stock picker. Not a bot. A research product.
No sign-up required · Reproducible rules · Failure periods included · Failed studies are published too
We do not try to find the highest backtest return.
A beautiful backtest is not sufficient evidence. We test strategies the way you would stress a machine: change one thing, keep everything else fixed, and watch where it breaks.
Every study begins with a simple, well-known starting point — not an exotic strategy tuned to look good.
A controlled comparison changes one thing at a time: a filter, a lookback, a rebalance frequency, a fallback asset.
Both sides of a comparison run on identical dates, symbols and assumptions, so differences are attributable.
Fees, turnover and the share of time in cash are published — the hidden costs that decide whether a backtest survives contact with reality.
The worst year, the longest drawdown and the whipsaw periods are as visible as the CAGR.
Every study states what it does not test, what could change the conclusion, and where the evidence is thin.
A strategy that does not work is a legitimate research result. We do not only publish the winners.
Inspect a real study
The first canonical study is published below with its real evidence — including where the strategy failed. This is the primary proof that Chat2Invest is a research product, not a landing page.
60/40 Portfolio With a Trend Filter: What Actually Changes?
What actually changes when you add a 200-day moving-average trend filter to a classic 60/40 stock/bond portfolio?
Over 2011–2026, adding a 200-day MA trend filter to each leg of a monthly-rebalanced 60/40 (SPY 60% + AGG 40%) cut the maximum drawdown only modestly (−21.8% → −17.4%) while cutting compound growth from 9.5% to 3.1% a year. The filter sat in cash roughly 44% of the time, and repeated whipsaws in 2015–16 and 2022–23 cost far more than the drawdown insurance was worth. As implemented, this filter fails.
| Metric (evaluation window 2011–2026) | Baseline 60/40 | + MA200 trend filter |
|---|---|---|
| CAGR | +9.5% | +3.1% |
| Max drawdown | -21.8% | -17.4% |
| Average cash | +0.2% | +43.8% |
| Turnover (of avg. AUM / yr) | +4.6% | +101.6% |
Where it failed: the filter dodged part of 2022 (−7.2% vs −16.2%) but was whipsawed twice in 2023 (−9.5% vs +17.7%) and never recovered its 2022 drawdown within the window.
Read the full study →What Chat2Invest does not do
No buy/sell recommendations, no “what should I buy today”.
No one-size-fits-all portfolio advice tailored to you.
We do not grid-search parameters until the backtest looks best.
Every rule, assumption and config is published and inspectable.
Start with the research index
One excellent study is more valuable than ten shallow ones. More studies are in review.
Explore Research© 2023–2026 Chat2Invest — evidence-first ETF strategy research.