Extended ETF Results Now Available for Download

Synthetic ETF Results - Now Available for Download
The new DMS website beautifully displays and compares strategies online, and the new engine does all the calculating in the web app - which finally got it out of my cumbersome Excel models.
A side project that came along with the engine was building new synthetic return data used by the strategies. Extending ETF results is not easy, and it is more difficult for leveraged ETFs - but it is necessary if you want to see meaningful back-tests.
My approach follows a waterfall, using the best available source at each step and falling back only when needed:
- Actual ETF data - used from inception forward.
- Sister mutual fund - often extends the record further back.
- Index data - where a clean index exists.
- Proxy source - not exact, but the best available fit.
- Linear regression - a last resort to estimate earlier results; this is not ideal, but preferable to no data at all.
My old data set worked, but it wasn't well documented, I had a feeling that some of the linear regressed data probably had better index or mutual fund sources that could be used. I felt an obligation to produce synthetic fund results that were new, documented, and properly extended, and open to public scrutiny. I'm posting all of it here for anyone to inspect and use.
On that note: I wish that strategy creators would band together in order to share a common set of fund return data. It would remove a big hidden variable and let strategies be compared on their logic rather than on whose back-test data ran deeper, or maybe is more ‘optimistic’ than somebody else’s data. That is in part why I documented this thoroughly and am making it available for download.
Using the data
You can use the results as-is - no Google Sheets setup is required. If you would like to extend the dataset yourself, I have included the Extension Apps Script: paste it in, add your personal Tiingo token and St. Louis FRED API key, save, and run it.
You'll also find DMS Synthetic Model Documentation.xlsx, which documents the full project in detail. It contains four sheets:
- Extended Monthly Returns
- Extended Daily Returns
- Synthetic Monthly
- Synthetic Daily
The Extended sheets use actual ETF data as far back as it exists, then fill earlier history with synthetic returns - these are the ones you'll want. The Synthetic sheets are internal (they're what the waterfall draws from to build the Extended series), but I've included them for full transparency.
Most series cover 1x/2x/3x variants; a few are 1x only. Fund data include are:
US Large Cap / S&P 500 - VOO 1x, SSO 2x, UPRO 3x
Russell 1000 - IWB 1x
Cash / T-Bills - BIL 1x
International Ex-US (Developed + EM) - VXUS 1x
20+ Year Treasury Bonds - TLT 1x, UBT 2x, TMF 3x
Gold - SGOL 1x, UGL 2x, SHNY 3x
Emerging Markets - EEM 1x, EET 2x, EDC 3x
US Real Estate (REITs) - VNQ 1x, URE 2x, DRN 3x
Small Cap (Russell 2000) - IWM 1x, UWM 2x, URTY 3x
Nasdaq-100 - QQQ 1x, QLD 2x, TQQQ 3x