Investment Strategies / Mechanical Investing
No. of Recommendations: 2
GTR1 allows the backtesting of synthetic leveraged indexes -- for instance {!S5TL2} which is 200% leverage on the S&P {!S5T}. Today I noticed that the results using such (highly useful!) tickers do not match those from a couple of months ago, given identical backtest terms and periods. In all cases that I have tried the CAGRs have been appreciably (200-300 basis points) lower than those originally obtained. I can detect no such trouble with synthetic tickers that don't apply leverage.
I was wondering if others have noticed anything similar.
Baltassar
No. of Recommendations: 1
I haven't noticed the issue, however, perhaps the issue is the way dividends are handled in calculating total returns which is probably pretty tricky for something that is a multiplier of a day's return.
Aussi
No. of Recommendations: 0
I only noticed because I ran screens today that were identical to those I ran in November, and got different results. Something must have changed to produce the difference.
There are plenty of tricky calculations involved, no question.
Baltassar
No. of Recommendations: 3
The gtr1 15 year CAGR is close to the ETF proxy (SSO):
gindex 15yCAGR ETFProxy diff
S5T 13.86 13.78 0.08
S5TE 12.70 12.29 0.41
S5TL2 22.83 22.29 0.54
No. of Recommendations: 3
Something has changed with S5TL2. CAGR from 19251231 to 20231229 is now 11.99%. A few months ago the CAGR was 17.03%. S5T and S5TE CAGRs look the same.
No. of Recommendations: 2
The CAGR might be closer to the ETF:
CAGR from 20151231 to 20201231:
21.7% etf SSO
25.17% S5TL2 today
28.05% S5TL2 2 months ago
All of the long leveraged gindexes have lower CAGR compared to the results from 2 months ago:
Screen CAGR oldCAGR difference From To
S5T 10.24 10.24 0.00 19251231 20241115
S5 6.27 6.27 0.00 19251231 20241115
S5TE 11.40 11.40 0.00 19251231 20241115
S5TL2 12.30 17.38 -5.08 19251231 20241115
S5TL3 10.40 20.62 -10.22 19251231 20241115
S5TS1 -10.46 -12.35 1.89 19251231 20241115
S5TS3 -39.47 -39.37 -0.10 19251231 20241115
N1T 13.36 13.36 0.00 19721214 20241115
N1 11.91 11.91 0.00 19721214 20241115
N1TE 10.90 10.90 0.00 19721214 20241115
N1TL2 15.08 21.59 -6.51 19721214 20241115
N1TL3 10.40 23.27 -12.87 19721214 20241115
N1TS1 -13.81 -16.54 2.73 19721214 20241115
N1TS3 -50.51 -50.96 0.45 19721214 20241115
RUTPT 10.26 10.26 0.00 19251231 20241115
RUTP 6.91 6.91 0.00 19251231 20241115
RUTPTE 9.95 9.95 0.00 19251231 20241115
RUTPTL2 12.06 17.13 -5.07 19251231 20241115
RUTPTL3 9.60 19.74 -10.15 19251231 20241115
RUTPTS1 -10.72 -12.60 1.88 19251231 20241115
RUTPTS3 -40.50 -40.41 -0.10 19251231 20241115
ZY100T 8.49 8.49 0.00 19261230 20241115
ZY100 7.93 7.93 0.00 19261230 20241115
ZY100TE 7.87 7.87 0.00 19261230 20241115
ZY100TL2 4.30 9.02 -4.72 19261230 20241115
ZY100TL3 -7.40 1.20 -8.59 19261230 20241115
ZY100TS1 -12.73 -14.57 1.84 19261230 20241115
ZY100TS3 -50.93 -50.85 -0.08 19261230 20241115
No. of Recommendations: 0
Something has changed with S5TL2. CAGR from 19251231 to 20231229 is now 11.99%. A few months ago the CAGR was 17.03%. S5T and S5TE CAGRs look the same.
Yes. I see something similar with !S4TL2 (Midcap 400 x 2). Synthetic tickers that don't involve leverage seem to work fine.
Baltassar
No. of Recommendations: 1
Forgive my ignorance, but what is ZY100?
Is the a comprehensive list anywhere of the synthetic indexes GTR1 will recognize?
Baltassar
No. of Recommendations: 2
Zero Yield stocks. Maybe somewhat of a proxy for QQQ before 1970. No list that I know of.
No. of Recommendations: 4
I only noticed because I ran screens today that were identical to those I ran in November, and got different results. Something must have changed to produce the difference.
I believe - but could well be wrong - that for non-standard tickers the GTR1 downloads the data (from Yahoo!?) at the time the backtest is requested and uses the downloaded data to perform the calculation. In this case non-standard means non-SI-Pro and other datasets that the GTR1 normally uses. If this is true, it's quite possible that the underlying data changed - for Yahoo! for example this happens quite frequently (and especially often for 'adjusted close' as the Dividends are recorded. I've also personally found Yahoo's calculation to be suspect on this). I believe that it was dissatisfaction with those types of datasets & the frequent errors that led to the creation of the GTR1 in the first place.
(Unless a lot of people are running the exact same request. In those instances the GTR1 can detect this and use the previously cached calculation - as it'll be the same.)
And so personally I always look at those 'downloaded' backtests with a bit of skepticism. They'll be approximately correct, but not necessarily exactly IMO.
No. of Recommendations: 2
I found some GTR1 data problems in several of my screens in early January.
Huge GSD's (like 3000-5000) suddenly appeared for holding 5 or more positions, starting on 20241224.
I emailed Robbie and a week or so later the problem disappeared.
Presumably, some wild data got into an update.
It is possible there is some data issue, as 2 or 3% seems pretty large.
But I'm not familiar with those indices and how far off Yahoo might be.
I'd think that platykurtic is correct.
Mark
No. of Recommendations: 1
I agree about "approximately correct". That is the only way to look at any backtest if you ask me. But the discrepancies in this case only arise among leveraged tickers. The scale of the error (several hundred basis points) is also cause for concern.
Baltassar
