I want a list of all fairly unique ETFs. When I say fairly unique, I mean that the past 120 day correlation between any ETF and all other ETFs in the list is lower than some threshold (they are fairly close because holdings substantially overlap). When there's a tie, I'd keep the more liquid ETF and drop the other. Basically, I don't want both VOO and SPY in the list, keep SPY, drop VOO. Same with XLK and VTI, keep XLK, drop VTI. But something like IBB is different enough from XBI that I would keep both. Does anyone happen to have such a list? This is probably close, but still a lot of overlap within the list: https://etfdb.com/compare/market-cap/
I think you mean same with XLK and QQQ. VTI is much more closely correlated with SPY (1.00 Pearson Correlation) than with QQQ (0.88 correlation). From https://www.portfoliovisualizer.com/asset-correlations
Correct. I was thinking VTI was the Vanguard Tech ETF. It's actually total market index. I was thinking of VGT. Been a while since I had a Vanguard account. The risk exposure and returns vary greatly among different ETFs. While I could just run my screener across all known ETFs, I don't really want to have to update a database of EOD data for all known ETFs.
I periodically create something similar to that with scripts that Get the top 550 passively-managed, non-leveraged, non-inverse ETFs by asset value from etfdb.com. Calculate Pearson's correlation for the past 3-years of closing prices adjusted for dividends and splits for every pair of ETFs from that list. Calculate liquidity for each ETF by multiplying the ETF's median 3-month adjusted close price by the median 3-month adjusted volume. For each ETF from highest to lowest liquidity, look at the correlation of the more liquid ETF with all less liquid ETFs, and discard the less liquid ETFs whose correlation with the more liquid ETF is at or above a 0.90 threshold. The remaining ETFs have no correlations at or above 0.90 with any other remaining ETF from the original 550 ETFs. The attached correlations.csv.zip has for the period ending November 20, 2020 semicolon-separated files with the correlations for all 550 ETFs and the ETFs having no correlations at or above 0.90 between any two of them. Code: zipinfo correlations.csv.zip Archive: correlations.csv.zip Zip file size: 3770094 bytes, number of entries: 2 -rw-r--r-- 3.0 unx 23708403 tx defN 20-Nov-22 14:10 correlations.csv -rw-r--r-- 3.0 unx 50248 tx defN 20-Nov-22 14:10 low_correlations.csv 2 files, 23758651 bytes uncompressed, 3769744 bytes compressed: 84.1%
%% True; but 80% or more if one mainly does spy/qqq/LOL TQQQ QLD,SPXL,UPRO,sds, dxd, gets a bigger pic................................................[NOT a prediction + .22/ .223+ 7mm Remington Magnum + 30-06\ .270 WINchester help with the bigger pic.]
ETF's are good for correlating but I only use the very large cap ones. Turtle may wish to consider this type large cap...
TT That %% That + metals sometimes /LOL; sometimes the mid cap, even though it can have huge drawdowns......................................................................................