Skip to content

EivindAamodt/Stock-Market-Leverage-Backtests

Repository files navigation

Big backtest on daily resetting leverage on the S&P 500 index

"Leveraged ETFs Are Not a Long-Term Bet" myth

Daily resetting ETFs are often called a poor long term investment. This is mainly because of volatility decay, also called beta decay. The most common example I see is that whenever the underlying index drops 10% then gains 10% the next day, a leveraged portfolio would lose a lot more value compared to the underlying.

Underlying: 100 -> 90 -> 99 - 1% loss

3x Leverage: 100 -> 70 -> 91 - 9% loss

A 9% loss is not a 3x of 1% loss!

A plot showing what it means in practice:

Volatility decay

What is often forgotten, is that the daily resetting also helps and serves as protection in some cases. Let's take an example where the underlying drops 10% four days in a row:

Underlying: 100 -> 90 -> 81 -> 73 -> 65 - 35% loss

3x Leverage: 100 -> 70 -> 49 -> 35 -> 24 - 76% loss

A 76% loss is a lot less than 3x of 35% loss. If it did not reset daily, the leveraged portfolio would be wiped out as 35*3 = 105% loss!

The same is also true when the underlying increases multiple days in a row:

Underlying: 100 -> 110 -> 121 -> 133 -> 146 - 46% gain

3x Leverage: 100 -> 130 -> 169 -> 220 -> 286 - 186% gain

A 186% gain is a lot better than the expected 46*3 = 138% gain.

Backtests from 6months up to 40 years

5k lump sum + 500/month DCA:

Lots of data - mean, median, percentiles, probabilities etc. Plots below:

250 trading days = 1 year

DCA

Plots

End value compared to SPY Raw end values
DCA125 ValueDCA125
DCA250 ValueDCA250
DCA500 ValueDCA500
DCA750 ValueDCA750
DCA1000 ValueDCA1000
DCA1500 ValueDCA1500
DCA2500 ValueDCA2500
DCA5000 ValueDCA5000
DCA7500 ValueDCA7500
DCA1000 ValueDCA1000

10k lump sum no DCA:

LOTS OF NUMBERS - mean, median, percentiles, probabilities etc:

Lumpsum

Plots

End value compared to SPY Raw end values
LumpSum125 ValueLumpsum125
LumpSum250 ValueLumpsum250
LumpSum500 ValueLumpsum500
LumpSum750 ValueLumpsum750
LumpSum1000 ValueLumpsum1000
LumpSum1500 ValueLumpsum1500
LumpSum2500 ValueLumpsum2500
LumpSum5000 ValueLumpsum5000
LumpSum7500 ValueLumpsum7500
LumpSum1000 ValueLumpsum1000

Some of the later graphs zoomed in for more clarity:

5000 days (20 years) DCA:

DCA5000

7500 days (30 years) DCA:

DCA5000

10000 days (40 years) DCA:

DCA5000

Conclusion

There is not a single 30 or 40 year timeframe since 1927 where DCAing into either 2x SPY or 3x SPY lost money compared to just buying SPY, even when holding through the depression in the 1930s, 1970s stagflation, the lost decade from 1999 to 2009, or ending the period at the bottom of the Covid-19 crash.

Past performance does not guarantee future results and all that stuff, but it does seem like having at least a portion of your portfolio in leveraged index funds is a great way to increase wealth, with the rewards heavily outweighing the risks. The hard part is having to stomach watching the extreme portfolio drawdowns during market corrections.

tl:dr

About

Backtesting leverage in the stock market.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages