HEDGEFUNDIE's excellent adventure Part II: The next journey

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Lee_WSP
Posts: 2693
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Lee_WSP » Tue Nov 26, 2019 5:19 pm

Stef wrote:
Tue Nov 26, 2019 5:16 pm
I just did a backtest for July 2006 - October 2019. 10'000$ initial and 1'000$/month, 50% SPY 50% SSO.

No rebalancing: $478,739
Quarterly rebalancing: $498,719
Semi-annually rebalancing: $497,959
Annually rebalancing: $501,898
5% absolut deviation: $499,137
10% absolut deviation: $507,724
20% absolut deviation: $534,869

Too bad we don't have more real data to back this up even further. But the result is very surprising! I mean this is 1.5 years bull market, 1.5 years bear market and then 10 years bull market. So with such a long bull market you would expect that no rebalancing should outperform rebalancing, but it doesn't.

I have to privilege to live in Switzerland. We don't pay taxes or realized profits, just on dividends. So I could even rebalance on a monthy basis. Rebalancing with absolut deviation seems like the best way to do it?
How confident are you that your simulated backtest actually tracks the real UPRO and TMF?
UPRO tracks extremely well. TMF has some issues, but siamond's analysis indicates that since 2013 it has also "fallen in line" with the formula. I have uploaded the data here, if you would like to play with it yourself. Samsdad has provided step-by-step directions on how to run this dataset in PortfolioVisualizer: viewtopic.php?p=4578095#p4578095


https://drive.google.com/open?id=1Byo8z ... 4tBWMebsAB
viewtopic.php?f=10&t=272007#p4364001

RayKeynes
Posts: 230
Joined: Mon Nov 11, 2019 2:14 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by RayKeynes » Tue Nov 26, 2019 5:22 pm

Lee_WSP wrote:
Tue Nov 26, 2019 5:12 pm
RayKeynes wrote:
Tue Nov 26, 2019 5:10 pm
Lee_WSP wrote:
Tue Nov 26, 2019 4:56 pm
The dataset is publicly available here: https://drive.google.com/open?id=16ORud ... WfTP-0FggA

You can see the rebalancing bonus with SSO & VFINX in portfolio visualizer. If you don't understand why the rebalancing bonus exists, I'm not sure I can help you understand this strategy.

https://www.portfoliovisualizer.com/bac ... tion2_1=50


https://www.portfoliovisualizer.com/bac ... tion2_1=50
I've already downloaded the data. However, I am not very familiar with this excel. I could only find the annual returns for these investments.
How did you calculate these figures?

I am using daily returns from 1929.
viewtopic.php?f=10&t=272640

Are you using some sort of proprietary dataset? Since LETF's have only been around for a couple decades at most, any dataset with 2x or 3x returns must be simulated.
I did all the calculation myself in a huge excel with daily simulation data including both TER and borrowing costs.

Look at the warning in his dataset please:
"WARNING: all metrics are derived from annual returns, therefore NOT taking in account intra-year events"

This may explain some of the deviations, I am using daily returns and investing on a MONTHLY basis, not yearly basis.However, I think the deviations between the calculations are still very high and I am open for further investigation, as I would like to rely on my data.

Lee_WSP
Posts: 2693
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Lee_WSP » Tue Nov 26, 2019 5:24 pm

RayKeynes wrote:
Tue Nov 26, 2019 5:22 pm
Lee_WSP wrote:
Tue Nov 26, 2019 5:12 pm
RayKeynes wrote:
Tue Nov 26, 2019 5:10 pm
Lee_WSP wrote:
Tue Nov 26, 2019 4:56 pm
The dataset is publicly available here: https://drive.google.com/open?id=16ORud ... WfTP-0FggA

You can see the rebalancing bonus with SSO & VFINX in portfolio visualizer. If you don't understand why the rebalancing bonus exists, I'm not sure I can help you understand this strategy.

https://www.portfoliovisualizer.com/bac ... tion2_1=50


https://www.portfoliovisualizer.com/bac ... tion2_1=50
I've already downloaded the data. However, I am not very familiar with this excel. I could only find the annual returns for these investments.
How did you calculate these figures?

I am using daily returns from 1929.
viewtopic.php?f=10&t=272640

Are you using some sort of proprietary dataset? Since LETF's have only been around for a couple decades at most, any dataset with 2x or 3x returns must be simulated.
I did all the calculation myself in a huge excel with daily simulation data including both TER and borrowing costs.

Look at the warning in his dataset please:
"WARNING: all metrics are derived from annual returns, therefore NOT taking in account intra-year events"

This may explain some of the deviations, I am using daily returns and investing on a MONTHLY basis, not yearly basis.However, I think the deviations between the calculations are still very high and I am open for further investigation, as I would like to rely on my data.
Upload it to google drive and we can all look at it. That's how peer review works. You need to share the data for us to examine it.

RayKeynes
Posts: 230
Joined: Mon Nov 11, 2019 2:14 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by RayKeynes » Tue Nov 26, 2019 5:28 pm

You may download the latest available version here:

v.3.8alpha
https://gofile.io/?c=hA9ZKw

It does only simulate SSO, UPRO and SPY as well as Nasdaq, tool was created to find the perfect portfolio allocation for leveraged ETFs

Walkure
Posts: 220
Joined: Tue Apr 11, 2017 9:59 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Walkure » Tue Nov 26, 2019 5:40 pm

RayKeynes wrote:
Tue Nov 26, 2019 5:28 pm
You may download the latest available version here:

v.3.8alpha
https://gofile.io/?c=hA9ZKw

It does only simulate SSO, UPRO and SPY as well as Nasdaq, tool was created to find the perfect portfolio allocation for leveraged ETFs
I'm getting a message that says "You are not authorized to download this file."

Lee_WSP
Posts: 2693
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Lee_WSP » Tue Nov 26, 2019 5:44 pm

RayKeynes wrote:
Tue Nov 26, 2019 5:28 pm
You may download the latest available version here:

v.3.8alpha
https://gofile.io/?c=hA9ZKw

It does only simulate SSO, UPRO and SPY as well as Nasdaq, tool was created to find the perfect portfolio allocation for leveraged ETFs
I'm not able to find the simulated 2x & 3x daily data that goes back to 1955.

RayKeynes
Posts: 230
Joined: Mon Nov 11, 2019 2:14 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by RayKeynes » Tue Nov 26, 2019 5:48 pm

Lee_WSP wrote:
Tue Nov 26, 2019 5:44 pm
RayKeynes wrote:
Tue Nov 26, 2019 5:28 pm
You may download the latest available version here:

v.3.8alpha
https://gofile.io/?c=hA9ZKw

It does only simulate SSO, UPRO and SPY as well as Nasdaq, tool was created to find the perfect portfolio allocation for leveraged ETFs
I'm not able to find the simulated 2x & 3x daily data that goes back to 1955.
What do you mean by that? Did you download it?

Lee_WSP
Posts: 2693
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Lee_WSP » Tue Nov 26, 2019 5:50 pm

RayKeynes wrote:
Tue Nov 26, 2019 5:48 pm
Lee_WSP wrote:
Tue Nov 26, 2019 5:44 pm
RayKeynes wrote:
Tue Nov 26, 2019 5:28 pm
You may download the latest available version here:

v.3.8alpha
https://gofile.io/?c=hA9ZKw

It does only simulate SSO, UPRO and SPY as well as Nasdaq, tool was created to find the perfect portfolio allocation for leveraged ETFs
I'm not able to find the simulated 2x & 3x daily data that goes back to 1955.
What do you mean by that? Did you download it?
Yes, and I looked through all the tabs, but I can't find the exact location of the daily data from '55 onwards. Some guidance would be appreciated. Like, tab 5, line 50.

RayKeynes
Posts: 230
Joined: Mon Nov 11, 2019 2:14 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by RayKeynes » Tue Nov 26, 2019 5:52 pm

Lee_WSP wrote:
Tue Nov 26, 2019 5:50 pm
RayKeynes wrote:
Tue Nov 26, 2019 5:48 pm
Lee_WSP wrote:
Tue Nov 26, 2019 5:44 pm
RayKeynes wrote:
Tue Nov 26, 2019 5:28 pm
You may download the latest available version here:

v.3.8alpha
https://gofile.io/?c=hA9ZKw

It does only simulate SSO, UPRO and SPY as well as Nasdaq, tool was created to find the perfect portfolio allocation for leveraged ETFs
I'm not able to find the simulated 2x & 3x daily data that goes back to 1955.
What do you mean by that? Did you download it?
Yes, and I looked through all the tabs, but I can't find the exact location of the daily data from '55 onwards. Some guidance would be appreciated. Like, tab 5, line 50.
You may open all the groupings of the Excel. The Data is all left to the Calculator Input. The tabs are Fürther Analysis and comparison of simulated data to HedgeFundies Data and real upro and sso data.

Data simulated is available since 1929.

BogleBobby
Posts: 30
Joined: Tue Jul 30, 2019 9:56 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by BogleBobby » Wed Nov 27, 2019 1:37 am

RayKeynes wrote:
Tue Nov 26, 2019 5:22 pm
I did all the calculation myself in a huge excel with daily simulation data including both TER and borrowing costs.

Look at the warning in his dataset please:
"WARNING: all metrics are derived from annual returns, therefore NOT taking in account intra-year events"

This may explain some of the deviations, I am using daily returns and investing on a MONTHLY basis, not yearly basis.However, I think the deviations between the calculations are still very high and I am open for further investigation, as I would like to rely on my data.
That footnote about metrics being derived from annual returns is referring to some of the metrics that spreadsheet derives (i.e. max drawdown, etc.). It is NOT referring to how the UPRO/SSO/TMF returns are calculated.

The UPRO/SSO/TMF data in that spreadsheet is based on daily data when available (1955-2018). But the spreadsheet only contains annual returns. The creator of the spreadsheet has daily data in another spreadsheet, but it's not publicly available because some of that daily data is proprietary.

I don't think the Boglehead community has ever found daily data on the SP500 prior to 1955, so would be curious how you got that.

RayKeynes
Posts: 230
Joined: Mon Nov 11, 2019 2:14 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by RayKeynes » Wed Nov 27, 2019 2:34 am

Lee_WSP wrote:
Tue Nov 26, 2019 4:36 pm
RayKeynes wrote:
Tue Nov 26, 2019 12:52 pm
No rebalancing done in my calculations. I can go back to 1929 if you wish with the backtesting.

Period (1955 - 2018, 10'000$ initial, 1'000$ monthly)
Portfolio 1 = 65% SPY (100%) + 35% UPRO (300%) = 170% leverage
Portfolio 2 = 50% SPY (100%) + 30% SSO (200%)+ 20% UPRO (300%) = 170% leverage
Portfolio 3 = 30% SPY (100%) + 70% SSO (200%) = 170% leverage
Portfolio 4 = 100% SPY (100%)

Endvalues:

Portfolio 1 = 12.02m $
Portfolio 2 = 19.74m $
Portfolio 3 = 30.03m $
Portfolio 4 = 12.65m $
I was not able to replicate your results using the Siamond spreadsheet. Please post your dataset.

Total - Rebalanced
Portfolio 1: 13,095,339
Portfolio 2:11,134,542
Portfolio 3: 14,758,360
Portfolio 4: 4,661,186


Total - Not Rebalanced
Portfolio 1: 5,135,675
Portfolio 2: 6,114,577
Portfolio 3: 5,610,165
Portfolio 4: 4,661,186

Although both back tests indicate that SSO may be worth taking another look at.
I am still confuse about the following two things:

1) Why are the results from LEFT backtest (1955 - 2018) so much different than my results shown above (nominal values)?
Hence, could someone provide me with the daily data from this excel so I can compare them to mine?

2) Why does a rebalanced portfolio perform nearly factor 3-times stronger than a non-rebalanced portfolio? To me - this does not make any sense. Rebalanced portfolios can both perform better and worse - its just another risk-return-allocation of the portfolio and thus not lead by RULE to a better return. Is simply not possible to do "market timing". I understand the concept that you guys are explaining in selling in market hights and buying in market lows. HOWEVER, when you are rebalancing in a constantly growing market (like the bull market 2009 - 2019), a rebalanced portfolio would DEFINITELY perform WORSE than a non-rebalanced portfolio. I can say that without even doing the math.

3) I am using the SP500 price index for approximation to simulate UPRO and SSO back to 1929. Below a comparison of the yearly returns I have done between the two data sets. As can bee seen, my data is more conservative as it is based on the price index and not total index return; so why in the name of *** does the other excel deliver such poor performance when investing monthly?

Image

edit: I've done a further plausibility check on the results you're calculator (SIAMOND) and my calculator produced. According to various sources, SP500 has returned a CAGR of 6.9% for the period from 01.01.1955 to 31.10.2019 (64.87 years or 23'679 days). I've then used the simple and straight-forward Compound Interest Calculator (which is mathematically correct) to estimate final portfolio value when investing 1'000$ on a monthly basis, see screenshot below:

Image

SP500 (100%), 1'000$ monthly, 10'000$ initial
Result of plausibility check: 14.37m $
Result of my calculator: 14.31m $
Result of SIAMOND calculator (according to Lee_WSP): 4.66m $

Further, you are claiming that the end portfolio value of a rebalanced portfolio is more than 2.5x higher than the one of a non-rebalanced portfolio. Do you have further evidence? Otherwise, I'd call the myth busted.

MotoTrojan
Posts: 9955
Joined: Wed Feb 01, 2017 8:39 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by MotoTrojan » Wed Nov 27, 2019 7:56 am

RayKeynes wrote:
Wed Nov 27, 2019 2:34 am
Lee_WSP wrote:
Tue Nov 26, 2019 4:36 pm
RayKeynes wrote:
Tue Nov 26, 2019 12:52 pm
No rebalancing done in my calculations. I can go back to 1929 if you wish with the backtesting.

Period (1955 - 2018, 10'000$ initial, 1'000$ monthly)
Portfolio 1 = 65% SPY (100%) + 35% UPRO (300%) = 170% leverage
Portfolio 2 = 50% SPY (100%) + 30% SSO (200%)+ 20% UPRO (300%) = 170% leverage
Portfolio 3 = 30% SPY (100%) + 70% SSO (200%) = 170% leverage
Portfolio 4 = 100% SPY (100%)

Endvalues:

Portfolio 1 = 12.02m $
Portfolio 2 = 19.74m $
Portfolio 3 = 30.03m $
Portfolio 4 = 12.65m $
I was not able to replicate your results using the Siamond spreadsheet. Please post your dataset.

Total - Rebalanced
Portfolio 1: 13,095,339
Portfolio 2:11,134,542
Portfolio 3: 14,758,360
Portfolio 4: 4,661,186


Total - Not Rebalanced
Portfolio 1: 5,135,675
Portfolio 2: 6,114,577
Portfolio 3: 5,610,165
Portfolio 4: 4,661,186

Although both back tests indicate that SSO may be worth taking another look at.
I am still confuse about the following two things:

1) Why are the results from LEFT backtest (1955 - 2018) so much different than my results shown above (nominal values)?
Hence, could someone provide me with the daily data from this excel so I can compare them to mine?

2) Why does a rebalanced portfolio perform nearly factor 3-times stronger than a non-rebalanced portfolio? To me - this does not make any sense. Rebalanced portfolios can both perform better and worse - its just another risk-return-allocation of the portfolio and thus not lead by RULE to a better return. Is simply not possible to do "market timing". I understand the concept that you guys are explaining in selling in market hights and buying in market lows. HOWEVER, when you are rebalancing in a constantly growing market (like the bull market 2009 - 2019), a rebalanced portfolio would DEFINITELY perform WORSE than a non-rebalanced portfolio. I can say that without even doing the math.

3) I am using the SP500 price index for approximation to simulate UPRO and SSO back to 1929. Below a comparison of the yearly returns I have done between the two data sets. As can bee seen, my data is more conservative as it is based on the price index and not total index return; so why in the name of *** does the other excel deliver such poor performance when investing monthly?

Image

edit: I've done a further plausibility check on the results you're calculator (SIAMOND) and my calculator produced. According to various sources, SP500 has returned a CAGR of 6.9% for the period from 01.01.1955 to 31.10.2019 (64.87 years or 23'679 days). I've then used the simple and straight-forward Compound Interest Calculator (which is mathematically correct) to estimate final portfolio value when investing 1'000$ on a monthly basis, see screenshot below:

Image

SP500 (100%), 1'000$ monthly, 10'000$ initial
Result of plausibility check: 14.37m $
Result of my calculator: 14.31m $
Result of SIAMOND calculator (according to Lee_WSP): 4.66m $

Further, you are claiming that the end portfolio value of a rebalanced portfolio is more than 2.5x higher than the one of a non-rebalanced portfolio. Do you have further evidence? Otherwise, I'd call the myth busted.
There is such a significant difference between investing $1000/month through actual markets and investing the same amount through a market growing at the same overall CAGR but equal growth annually (like that calculator) that I don’t even know where to begin.

RayKeynes
Posts: 230
Joined: Mon Nov 11, 2019 2:14 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by RayKeynes » Wed Nov 27, 2019 8:03 am

MotoTrojan wrote:
Wed Nov 27, 2019 7:56 am

There is such a significant difference between investing $1000/month through actual markets and investing the same amount through a market growing at the same overall CAGR but equal growth annually (like that calculator) that I don’t even know where to begin.
Of course there is a deviation around the average CAGR of 6.9% over the total period. Sometimes the amount you invest monthly is above the CAGR, sometimes its below the CAGR line. However, getting a result of 4.6 Million versus an actual result 14.6 Million is - sorry - completely mathematically wrong and nonsense. The monthly investments do nicely smooth out all the overperformance and underperformance periods.

Please recalculate the total return using SIAMOND data, otherwise I'd call this datasheet to be wrong.

User avatar
Stef
Posts: 966
Joined: Thu Oct 10, 2019 10:13 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Stef » Wed Nov 27, 2019 8:58 am

The 6.9% aren't even acounting for dividends. Siamonds data does? So the difference is even bigger.

Lee_WSP
Posts: 2693
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Lee_WSP » Wed Nov 27, 2019 12:02 pm

RayKeynes wrote:
Wed Nov 27, 2019 2:34 am

Further, you are claiming that the end portfolio value of a rebalanced portfolio is more than 2.5x higher than the one of a non-rebalanced portfolio. Do you have further evidence? Otherwise, I'd call the myth busted.
I posted backtesting with the actual daily data of SSO which shows the very clear rebalancing bonus during bear markets. You can easily run the same exact test as I gave you the link.

RayKeynes
Posts: 230
Joined: Mon Nov 11, 2019 2:14 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by RayKeynes » Wed Nov 27, 2019 12:26 pm

Lee_WSP wrote:
Wed Nov 27, 2019 12:02 pm
RayKeynes wrote:
Wed Nov 27, 2019 2:34 am

Further, you are claiming that the end portfolio value of a rebalanced portfolio is more than 2.5x higher than the one of a non-rebalanced portfolio. Do you have further evidence? Otherwise, I'd call the myth busted.
I posted backtesting with the actual daily data of SSO which shows the very clear rebalancing bonus during bear markets. You can easily run the same exact test as I gave you the link.
SSO did not exist since 1955. So please show your results and the rebalancing bonus since inception of SSO. I am interested to see them.

User avatar
Stef
Posts: 966
Joined: Thu Oct 10, 2019 10:13 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Stef » Wed Nov 27, 2019 12:30 pm

SSO is only available for 2006-2019 in Portfolio Visualizer. So it's not really useful, because 13 years aren't statistically relevant. Hell, even 100 years aren't really that great compared to other fields in science.

Hydromod
Posts: 402
Joined: Tue Mar 26, 2019 10:21 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Hydromod » Wed Nov 27, 2019 2:14 pm

Just for grins, I independently used daily S&P 500 price to check the calculations in a spreadsheet. I did full years from 12/31/1954 through 12/31/2018.

I get a nominal CAGR of 6.86% with no dividends.

If I apply each year's annual dividends uniformly at 1/252 per trading day, I get a nominal CAGR of 10.1%.

If I start with $10K and add $1K the first of the month, without dividends I end up with $13.78 M on 12/31/2018. Using just annual values, I get $12.6 M.

This is consistent with what RayKeynes is stating for the S&P 500 calculation.

Another calculation check for RayKeynes: if I add dividends into the daily set, I end up with $61.58 M on 12/31/2018.

As a second calculation check for RayKeynes, I scaled daily changes by 2 and 3 to give an estimate for SSO and UPRO. This is clearly not the correct estimate for SSO and UPRO, because it doesn't account for leverage expenses, it's just for comparison.

Without expenses or dividends, I get a nominal CAGR of 11.4% and 13.1% for 2x and 3x. With 2x and 3x dividends, 18.3% and 23.8%.

I remember about 1% ER for both 2x and 3x leverage. Spreading that evenly over the year, approximate 2x and 3x without dividends is 9.19 and 9.81%; with dividends is 16.0% and 20.2%.

These numbers are consistent with what I remember from some discussions (e.g., 3x daily leverage ends up with 2x overall).

With expenses, the 10K/1K scheme ends at $39.5M (without dividends) and $990M (with dividends) for 2x.
With expenses, the 10K/1K scheme ends at $46.2M (without dividends) and $6915M (with dividends) for 3x.

I don't have the Siamond file available where I am, but I can do similar checks with that tonight. I'm intrigued to see why there seems to be a difference in reported endpoints.

jaj2276
Posts: 493
Joined: Sat Apr 16, 2011 5:13 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by jaj2276 » Wed Nov 27, 2019 2:20 pm

Just took a peek at PV to see what 2 of my 3 slices will get rebalanced to come Friday. The 21-day vol lookback (1/4 of the portfolio) will go to 71% UPRO, 29% TMF. Paired with the 40/60 OG slice (1/2 portfolio and currently drifted to 45/55), this shouldn't be too big of a shift to the portfolio. The 16-vol portfoliio however will go to 99% UPRO, 1% TMF! Let's hope for Xmas rally! :D

Lee_WSP
Posts: 2693
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Lee_WSP » Wed Nov 27, 2019 2:41 pm

RayKeynes wrote:
Wed Nov 27, 2019 12:26 pm
[quote=Lee_WSP post_id=4860348 time=<a href="tel:1574874153">1574874153</a> user_id=147765]
[quote=RayKeynes post_id=4859796 time=<a href="tel:1574840067">1574840067</a> user_id=153916]

Further, you are claiming that the end portfolio value of a rebalanced portfolio is more than 2.5x higher than the one of a non-rebalanced portfolio. Do you have further evidence? Otherwise, I'd call the myth busted.
I posted backtesting with the actual daily data of SSO which shows the very clear rebalancing bonus during bear markets. You can easily run the same exact test as I gave you the link.
[/quote]

SSO did not exist since 1955. So please show your results and the rebalancing bonus since inception of SSO. I am interested to see them.
[/quote]
Stef wrote:
Wed Nov 27, 2019 12:30 pm
SSO is only available for 2006-2019 in Portfolio Visualizer. So it's not really useful, because 13 years aren't statistically relevant. Hell, even 100 years aren't really that great compared to other fields in science.
Well smart redacted, the siamond dataset also shows the rebalance bonus. There isn’t any other actual data to show the bonus other than sso and a few others. So I’m not sure what other proof you’d like.

BogleBobby
Posts: 30
Joined: Tue Jul 30, 2019 9:56 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by BogleBobby » Wed Nov 27, 2019 2:49 pm

I've been able to reproduce Siamond's 3x returns before using daily data from 1955-2018, so I'm fairly confident in those.

One thing to note is that borrowing costs are pretty significant in some years. And I think he assumes something like a 1% expense ratio.

User avatar
Stef
Posts: 966
Joined: Thu Oct 10, 2019 10:13 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Stef » Wed Nov 27, 2019 3:18 pm

Rebalancing doesn't always decrease risk.

Let's asume that there is a crash over 2 years with UPRO hitting almost 0 (-99% drawdown). Not very likely, but possible. Rebalancing with SP500 might destroy >90% of your assets even with a 66/33 SPY/UPRO allocation. You just keep beating a dead horse. Whereas with no rebalancing at least 2/3 of your portfolio would be "safe".

I think it's just pure luck if rebalancing will end up with better or worse results over a 25-30 year period.

RayKeynes
Posts: 230
Joined: Mon Nov 11, 2019 2:14 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by RayKeynes » Wed Nov 27, 2019 3:36 pm

Stef wrote:
Wed Nov 27, 2019 3:18 pm
Rebalancing doesn't always decrease risk.

Let's asume that there is a crash over 2 years with UPRO hitting almost 0 (-99% drawdown). Not very likely, but possible. Rebalancing with SP500 might destroy >90% of your assets even with a 66/33 SPY/UPRO allocation. You just keep beating a dead horse. Whereas with no rebalancing at least 2/3 of your portfolio would be "safe".

I think it's just pure luck if rebalancing will end up with better or worse results over a 25-30 year period.
Absolutely correct, Sir. Rebalancing is a Synonyme for adjusting the risk of a Portfolio and the correlation factor between assets. Hence, rebalancing will only change expected Return, expected volatility and correlation between assets when not already perfecty correlated. As the above mentioned Portfolio Allocation does solely consist of perfectly collerated assets, only expected Return and volatility is being adjusted through Rebalancing.

Rebalancing will thus delivery Vetter returns than a drifting Portfolio when the period consists of a crash in the beginning. Rebalancing will thus deliver worse Returns when the period consists of a bull Market Like 2009 - 2019 or 1990-2001. As a result, whether rebalancing enhances Returns is Poor luck. The idea of rebalancing is too keep ones Leverage (Risk) Constant.

rascott
Posts: 2112
Joined: Wed Apr 15, 2015 10:53 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by rascott » Wed Nov 27, 2019 3:37 pm

Stef wrote:
Wed Nov 27, 2019 3:18 pm
Rebalancing doesn't always decrease risk.

Let's asume that there is a crash over 2 years with UPRO hitting almost 0 (-99% drawdown). Not very likely, but possible. Rebalancing with SP500 might destroy >90% of your assets even with a 66/33 SPY/UPRO allocation. You just keep beating a dead horse. Whereas with no rebalancing at least 2/3 of your portfolio would be "safe".

I think it's just pure luck if rebalancing will end up with better or worse results over a 25-30 year period.

I feel like I'm beating a dead horse.... but you rebalance to maintain your leverage rate constant. If you do not your leverage levels will vary widely when using LETFs. In bull markets your leverage level will keep going up and up..... and vice versa in bear markets.

You can easily backtest constant leverage using PV..... won't give you the exact returns as using LETFs.... but it's in the ballpark.

spae
Posts: 133
Joined: Tue Oct 08, 2019 9:29 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by spae » Thu Nov 28, 2019 8:55 am

RayKeynes wrote:
Wed Nov 27, 2019 3:36 pm
only expected Return and volatility is being adjusted through Rebalancing.
Is this supposed to be a rebuttal to the people who are saying that rebalancing is important? It only impacts volatility and expected return?

RayKeynes
Posts: 230
Joined: Mon Nov 11, 2019 2:14 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by RayKeynes » Thu Nov 28, 2019 9:25 am

spae wrote:
Thu Nov 28, 2019 8:55 am
RayKeynes wrote:
Wed Nov 27, 2019 3:36 pm
only expected Return and volatility is being adjusted through Rebalancing.
Is this supposed to be a rebuttal to the people who are saying that rebalancing is important? It only impacts volatility and expected return?
For perfectly correlated assets - yes. Rebalancing does just amend expected volatility and expected return. For unperfectly correlated assets (e.g. TMF/UPRO), rebalancing does also adjust correlation between assets. Between negatively correlated assets - rebalancing is strongly recommended due to the fact that you've chosen a certain strategic asset allocation to profit from the negative correlation of assets.

However, for positively correlated assets, rebalancing does only influence the above-mentioned and thus does not lead by rule to better risk-adjusted returns - this is pure luck and is depending on the future.

get_g0ing
Posts: 693
Joined: Sat Dec 09, 2017 11:09 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by get_g0ing » Thu Nov 28, 2019 10:09 am

This is perhaps discussed before, so I apologize.

What's a simple way to view the performance of UPRO or TQQQ during the two previous crashes: 2000-2002 and 2008?
Something like a Portfolio Visualizer type output?

Thanks.

MotoTrojan
Posts: 9955
Joined: Wed Feb 01, 2017 8:39 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by MotoTrojan » Thu Nov 28, 2019 10:29 am

get_g0ing wrote:
Thu Nov 28, 2019 10:09 am
This is perhaps discussed before, so I apologize.

What's a simple way to view the performance of UPRO or TQQQ during the two previous crashes: 2000-2002 and 2008?
Something like a Portfolio Visualizer type output?

Thanks.
Original post has 100% UPRO plotted from 1987 to 2019. 2008 was a nearly 98% drawdown. Also should be a link to the raw data which you can upload to PV.

RayKeynes
Posts: 230
Joined: Mon Nov 11, 2019 2:14 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by RayKeynes » Thu Nov 28, 2019 10:38 am

get_g0ing wrote:
Thu Nov 28, 2019 10:09 am
This is perhaps discussed before, so I apologize.

What's a simple way to view the performance of UPRO or TQQQ during the two previous crashes: 2000-2002 and 2008?
Something like a Portfolio Visualizer type output?

Thanks.
Just did a quick simulation for you using my calculator. Growth of 10'000$ from 1995 - 2010 (1st January), non-log and log charts:

Image

Image

langlands
Posts: 261
Joined: Wed Apr 03, 2019 10:05 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by langlands » Thu Nov 28, 2019 11:20 am

RayKeynes wrote:
Thu Nov 28, 2019 9:25 am
spae wrote:
Thu Nov 28, 2019 8:55 am
RayKeynes wrote:
Wed Nov 27, 2019 3:36 pm
only expected Return and volatility is being adjusted through Rebalancing.
Is this supposed to be a rebuttal to the people who are saying that rebalancing is important? It only impacts volatility and expected return?
For perfectly correlated assets - yes. Rebalancing does just amend expected volatility and expected return. For unperfectly correlated assets (e.g. TMF/UPRO), rebalancing does also adjust correlation between assets. Between negatively correlated assets - rebalancing is strongly recommended due to the fact that you've chosen a certain strategic asset allocation to profit from the negative correlation of assets.

However, for positively correlated assets, rebalancing does only influence the above-mentioned and thus does not lead by rule to better risk-adjusted returns - this is pure luck and is depending on the future.
You say that rebalancing "only" affects expected volatility and expected return. I'm a little confused because those are pretty much the only two parameters that factor into the assessment of my portfolio.

There is some leverage ratio (which seems to be around 1.5 if you go by past data) that maximizes expected CAGR (this is equivalent to maximizing expected log wealth or optimal Kelly betting). So assuming this is what you want, it makes sense to rebalance to keep your portfolio at this optimum ratio. Not rebalancing is essentially market timing since you are higher leveraged after a bull market and lower leveraged after a bear market.

Edit: I think I see the heart of the disagreement. Changing the leverage ratio does not change the Sharpe of the portfolio, so indeed you do not get better risk adjusted returns. When you have multiple assets, you want to first optimize the Sharpe by getting the relative weightings between your assets right according to mean variance analysis etc. But once you do that, there is one last parameter to pick which is the overall weight between your portfolio and cash, i.e. the leverage ratio. This doesn't change your Sharpe, but can massively affect your expected CAGR.

MotoTrojan
Posts: 9955
Joined: Wed Feb 01, 2017 8:39 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by MotoTrojan » Thu Nov 28, 2019 11:54 am

langlands wrote:
Thu Nov 28, 2019 11:20 am
RayKeynes wrote:
Thu Nov 28, 2019 9:25 am
spae wrote:
Thu Nov 28, 2019 8:55 am
RayKeynes wrote:
Wed Nov 27, 2019 3:36 pm
only expected Return and volatility is being adjusted through Rebalancing.
Is this supposed to be a rebuttal to the people who are saying that rebalancing is important? It only impacts volatility and expected return?
For perfectly correlated assets - yes. Rebalancing does just amend expected volatility and expected return. For unperfectly correlated assets (e.g. TMF/UPRO), rebalancing does also adjust correlation between assets. Between negatively correlated assets - rebalancing is strongly recommended due to the fact that you've chosen a certain strategic asset allocation to profit from the negative correlation of assets.

However, for positively correlated assets, rebalancing does only influence the above-mentioned and thus does not lead by rule to better risk-adjusted returns - this is pure luck and is depending on the future.
You say that rebalancing "only" affects expected volatility and expected return. I'm a little confused because those are pretty much the only two parameters that factor into the assessment of my portfolio.

There is some leverage ratio (which seems to be around 1.5 if you go by past data) that maximizes expected CAGR (this is equivalent to maximizing expected log wealth or optimal Kelly betting). So assuming this is what you want, it makes sense to rebalance to keep your portfolio at this optimum ratio. Not rebalancing is essentially market timing since you are higher leveraged after a bull market and lower leveraged after a bear market.

Edit: I think I see the heart of the disagreement. Changing the leverage ratio does not change the Sharpe of the portfolio, so indeed you do not get better risk adjusted returns. When you have multiple assets, you want to first optimize the Sharpe by getting the relative weightings between your assets right according to mean variance analysis etc. But once you do that, there is one last parameter to pick which is the overall weight between your portfolio and cash, i.e. the leverage ratio. This doesn't change your Sharpe, but can massively affect your expected CAGR.
Changing the leverage ratio with daily rebalancing funds like these absolutely changes the sharpe ratio. Volatility is proportional to the leverage ratio while returns are not for periods longer than one trading day.

get_g0ing
Posts: 693
Joined: Sat Dec 09, 2017 11:09 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by get_g0ing » Thu Nov 28, 2019 12:00 pm

MotoTrojan wrote:
Thu Nov 28, 2019 10:29 am
Original post has 100% UPRO plotted from 1987 to 2019. 2008 was a nearly 98% drawdown. Also should be a link to the raw data which you can upload to PV.

Thanks Moto. I found the UPROSIM.csv and TMFSIM.csv files and was able to import them to PV.
Do we have a "TQQQSIM.csv" also? I can't seem to locate it.

RayKeynes wrote:
Thu Nov 28, 2019 10:38 am

Just did a quick simulation for you using my calculator. Growth of 10'000$ from 1995 - 2010 (1st January), non-log and log charts:

Image

Image
Hi RayKeynes, are you using the same files as above for your graphs? Can you say what tool you are using to graph please?

When I use PV to graph the above files, I see a mismatch from your graph between October-2003 and June-2008

Image
P1 = UPROSIM
Last edited by get_g0ing on Thu Nov 28, 2019 12:13 pm, edited 5 times in total.

Lee_WSP
Posts: 2693
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Lee_WSP » Thu Nov 28, 2019 12:00 pm

langlands wrote:
Thu Nov 28, 2019 11:20 am
.

Edit: I think I see the heart of the disagreement. Changing the leverage ratio does not change the Sharpe of the portfolio, so indeed you do not get better risk adjusted returns. When you have multiple assets, you want to first optimize the Sharpe by getting the relative weightings between your assets right according to mean variance analysis etc. But once you do that, there is one last parameter to pick which is the overall weight between your portfolio and cash, i.e. the leverage ratio. This doesn't change your Sharpe, but can massively affect your expected CAGR.
I don't think the Sharpe ratio is very useful for leveraged portfolios.

langlands
Posts: 261
Joined: Wed Apr 03, 2019 10:05 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by langlands » Thu Nov 28, 2019 12:05 pm

@MotoTrojan: I meant the one day Sharpe ratio. I'm not sure how optimizing the Sharpe ratio for long periods (with a single asset) fits into any optimal investing framework.

@LeeWSP: Yes, I agree. I got the sense that RayKeynes was talking about the Sharpe ratio though since he mentioned risk-adjusted returns (maybe I'm wrong). I was hoping to connect the two viewpoints to prevent people continuing to talk past one another.
Last edited by langlands on Thu Nov 28, 2019 12:10 pm, edited 1 time in total.

MotoTrojan
Posts: 9955
Joined: Wed Feb 01, 2017 8:39 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by MotoTrojan » Thu Nov 28, 2019 12:10 pm

langlands wrote:
Thu Nov 28, 2019 12:05 pm
@MotoTrojan: I meant the one day Sharpe ratio. I'm not sure how optimizing the Sharpe ratio for long periods fits into any optimal investing framework.

@LeeWSP: Yes, I agree. I got the sense that RayKeynes was talking about the Sharpe ratio though since he mentioned risk-adjusted returns (maybe I'm wrong). I was hoping to connect the two viewpoints to prevent people continuing to talk past one another.
You feel daily sharpe is more important than long term sharpe?

langlands
Posts: 261
Joined: Wed Apr 03, 2019 10:05 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by langlands » Thu Nov 28, 2019 12:16 pm

MotoTrojan wrote:
Thu Nov 28, 2019 12:10 pm
langlands wrote:
Thu Nov 28, 2019 12:05 pm
@MotoTrojan: I meant the one day Sharpe ratio. I'm not sure how optimizing the Sharpe ratio for long periods fits into any optimal investing framework.

@LeeWSP: Yes, I agree. I got the sense that RayKeynes was talking about the Sharpe ratio though since he mentioned risk-adjusted returns (maybe I'm wrong). I was hoping to connect the two viewpoints to prevent people continuing to talk past one another.
You feel daily sharpe is more important than long term sharpe?
My primary utility function is maximizing expected CAGR for my overall portfolio. Assuming no trading (rebalancing) costs, I believe this can be achieved in a two step process that is conducted daily. Every day, maximize the daily Sharpe ratio by getting the relative weighting between your assets correct and then get the leverage multiple correct by figuring out how much cash to go short.

Lee_WSP
Posts: 2693
Joined: Fri Apr 19, 2019 5:15 pm
Location: Arizona

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Lee_WSP » Thu Nov 28, 2019 12:17 pm

langlands wrote:
Thu Nov 28, 2019 12:05 pm
@MotoTrojan: I meant the one day Sharpe ratio. I'm not sure how optimizing the Sharpe ratio for long periods (with a single asset) fits into any optimal investing framework.

@LeeWSP: Yes, I agree. I got the sense that RayKeynes was talking about the Sharpe ratio though since he mentioned risk-adjusted returns (maybe I'm wrong). I was hoping to connect the two viewpoints to prevent people continuing to talk past one another.
I think we're not even looking at the same data. I still can't find the raw simulated data in his spreadsheet.

RayKeynes
Posts: 230
Joined: Mon Nov 11, 2019 2:14 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by RayKeynes » Thu Nov 28, 2019 12:24 pm

get_g0ing wrote:
Thu Nov 28, 2019 12:00 pm
MotoTrojan wrote:
Thu Nov 28, 2019 10:29 am
Original post has 100% UPRO plotted from 1987 to 2019. 2008 was a nearly 98% drawdown. Also should be a link to the raw data which you can upload to PV.

Thanks Moto. I found the UPROSIM.csv and TMFSIM.csv files and was able to import them to PV.
Do we have a "TQQQSIM.csv" also? I can't seem to locate it.

RayKeynes wrote:
Thu Nov 28, 2019 10:38 am

Just did a quick simulation for you using my calculator. Growth of 10'000$ from 1995 - 2010 (1st January), non-log and log charts:

Image

Image
Hi RayKeynes, are you using the same files as above for your graphs? Can you say what tool you are using to graph please?

When I use PV to graph the above files, I see a mismatch from your graph between October-2003 and June-2008

Image
P1 = UPROSIM
Sorry, I should have added: Growth of 10'000$ initial investment AND 2'000$ monthly investment.

I always calculate with monthly investment - as it is more realistic for our investment purpose.

RayKeynes
Posts: 230
Joined: Mon Nov 11, 2019 2:14 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by RayKeynes » Thu Nov 28, 2019 12:33 pm

As you've requested, same period (1995 - 2010) including two crysis without monthly investing, growth of 10'000$

Image
End values in 2010
SP 500 23'565.53$
TQQQ 4'840$
UPRO 10'312$
SSO 25'012$
Last edited by RayKeynes on Thu Nov 28, 2019 1:17 pm, edited 1 time in total.

get_g0ing
Posts: 693
Joined: Sat Dec 09, 2017 11:09 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by get_g0ing » Thu Nov 28, 2019 12:41 pm

RayKeynes wrote:
Thu Nov 28, 2019 12:24 pm

Sorry, I should have added: Growth of 10'000$ initial investment AND 2'000$ monthly investment.

I always calculate with monthly investment - as it is more realistic for our investment purpose.
Sure, that's fair. I just wanted to see if I wasn't messing it up. Thanks for clarifying.

Okay, so now PV with UPROSIM.csv looks to match what you have:

Image

Can you share the TQQQ .csv file so I can load it in PV?

BogleBobby
Posts: 30
Joined: Tue Jul 30, 2019 9:56 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by BogleBobby » Thu Nov 28, 2019 4:43 pm

305pelusa wrote:
Tue Nov 26, 2019 12:50 pm
Maybe this is a dumb question but: Has anyone actually peer reviewed Siamond's calculations, formulas, and how he generated the data? Taking into account fees, daily reconstitution, borrowing costs, etc appropriately?
I'm a bit late to answer this question, but for UPRO and SSO, I was able to get very close to reproducing Siamond's numbers exactly, so we can consider them 'peer reviewed'. Siamond also checked his numbers against historical UPRO and SSO returns when they were available, so that was one check he was able to do when he initially created the simulated returns.

I have not tried to check Siamond's numbers for TMF.
Last edited by BogleBobby on Thu Nov 28, 2019 4:58 pm, edited 1 time in total.

User avatar
Stef
Posts: 966
Joined: Thu Oct 10, 2019 10:13 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Stef » Thu Nov 28, 2019 4:46 pm

Is there any consence on the optimal asset allocation if aiming for a 150% leverage?

50% VT 50% SSO seems very aggressive. I'm not sure if I'm comfortable with more than 35% in LETFs. So 65% VT 20% SSO 15% UPRO would also work.

MotoTrojan
Posts: 9955
Joined: Wed Feb 01, 2017 8:39 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by MotoTrojan » Thu Nov 28, 2019 5:16 pm

Stef wrote:
Thu Nov 28, 2019 4:46 pm
Is there any consence on the optimal asset allocation if aiming for a 150% leverage?

50% VT 50% SSO seems very aggressive. I'm not sure if I'm comfortable with more than 35% in LETFs. So 65% VT 20% SSO 15% UPRO would also work.
Again, you should never hold SSO and UPRO. Just hold enough UPRO and rebalance as needed to maintain the desired leverage.

Overall though I disagree with these overall portfolios.

langlands
Posts: 261
Joined: Wed Apr 03, 2019 10:05 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by langlands » Thu Nov 28, 2019 5:28 pm

One reason to hold SSO is tax efficiency. Say there were an etf that was 1.5x leveraged and that was exactly what you wanted. Then you could just buy and hold that and never have to rebalance. By holding SSO instead of UPRO, you might have to rebalance less which could possibly make up for the expense ratio inefficiency.

User avatar
Stef
Posts: 966
Joined: Thu Oct 10, 2019 10:13 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Stef » Thu Nov 28, 2019 6:19 pm

So I just made a backtest using RayKeynes dataset.

2000$/month from 01.01.1989-01.11.2019
50% SPY 50% SSO

Rebalanced with 10% absolut bands:
01.01.1989-30.10.1996
152'871 SPY 229'194 SSO -> 191'033 SPY 191'033 SSO
30.10.1996-31.12.1998
367'026 SPY 553'108 SSO -> 460'067 SPY 460'067 SSO
31.12.1998-30.06.2002
399'656 SPY 262'616 SSO -> 331'136 SPY 331'136 SSO
30.06.2002-30.11.2008
355'703 SPY 229'135 SSO -> 292'419 SPY 292'419 SSO
30.11.2008-30.04.2013
588'687 SPY 887'857 SSO -> 738'272 SPY 738'272 SSO
30.04.2013-31.07.2017
1'197'223 SPY 1'805'471 SSO -> 1'501'347 SPY 1'501'347 SSO
31.07.2017-01.11.2019
1'884'268 SPY 2'229'114 SSO = 4'113'382
No rebalancing: 4'022'696
100% SPY: 2'228'372

That's almost 31 years and almost no effect. It's pure coincidence that rebalancing turned out to be slightly better. But it might help you sleep better at night.

So even if we have a 2008 crash right now, we would be looking at:

100% SPY: ~1100k
50% SPY 50% SSO not rebalanced: ~1400k
50% SPY 50% SSO rebalanced: ~1590k

150% leverage is ****** awesome!

MotoTrojan
Posts: 9955
Joined: Wed Feb 01, 2017 8:39 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by MotoTrojan » Thu Nov 28, 2019 7:38 pm

Stef wrote:
Thu Nov 28, 2019 6:19 pm
So I just made a backtest using RayKeynes dataset.

2000$/month from 01.01.1989-01.11.2019
50% SPY 50% SSO

Rebalanced with 10% absolut bands:
01.01.1989-30.10.1996
152'871 SPY 229'194 SSO -> 191'033 SPY 191'033 SSO
30.10.1996-31.12.1998
367'026 SPY 553'108 SSO -> 460'067 SPY 460'067 SSO
31.12.1998-30.06.2002
399'656 SPY 262'616 SSO -> 331'136 SPY 331'136 SSO
30.06.2002-30.11.2008
355'703 SPY 229'135 SSO -> 292'419 SPY 292'419 SSO
30.11.2008-30.04.2013
588'687 SPY 887'857 SSO -> 738'272 SPY 738'272 SSO
30.04.2013-31.07.2017
1'197'223 SPY 1'805'471 SSO -> 1'501'347 SPY 1'501'347 SSO
31.07.2017-01.11.2019
1'884'268 SPY 2'229'114 SSO = 4'113'382
No rebalancing: 4'022'696
100% SPY: 2'228'372

That's almost 31 years and almost no effect. It's pure coincidence that rebalancing turned out to be slightly better. But it might help you sleep better at night.

So even if we have a 2008 crash right now, we would be looking at:

100% SPY: ~1100k
50% SPY 50% SSO not rebalanced: ~1400k
50% SPY 50% SSO rebalanced: ~1590k

150% leverage is ****** awesome!
It WAS awesome.

BogleBobby
Posts: 30
Joined: Tue Jul 30, 2019 9:56 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by BogleBobby » Thu Nov 28, 2019 9:33 pm

Stef wrote:
Thu Nov 28, 2019 6:19 pm
So I just made a backtest using RayKeynes dataset.
I wouldn't use that dataset since it's already been shown to not match Siamonds, but Siamonds dataset has been peer reviewed and Ray's has not.

RayKeynes
Posts: 230
Joined: Mon Nov 11, 2019 2:14 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by RayKeynes » Fri Nov 29, 2019 2:47 am

BogleBobby wrote:
Thu Nov 28, 2019 9:33 pm
Stef wrote:
Thu Nov 28, 2019 6:19 pm
So I just made a backtest using RayKeynes dataset.
I wouldn't use that dataset since it's already been shown to not match Siamonds, but Siamonds dataset has been peer reviewed and Ray's has not.
Can someone please show the following results using Siamonds datasheet:

1) 10'000 initial - no monthly investment
periods: 01.01.1955 - 01.11.2019 / 01.01.1980 - 01.01.2010 / 01.01.2000 - 01.11.2019
UPRO
SSO
SPY

2) 10'000 initial - 2'000$ monthly investment
periods: 01.01.1955 - 01.11.2019 / 01.01.1980 - 01.01.2010 / 01.01.2000 - 01.11.2019
UPRO
SSO
SPY

I will then show my results to compare. I hope there arent' any significant deviations.

rascott
Posts: 2112
Joined: Wed Apr 15, 2015 10:53 am

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by rascott » Fri Nov 29, 2019 9:25 am

Stef wrote:
Thu Nov 28, 2019 6:19 pm
So I just made a backtest using RayKeynes dataset.

2000$/month from 01.01.1989-01.11.2019
50% SPY 50% SSO

Rebalanced with 10% absolut bands:
01.01.1989-30.10.1996
152'871 SPY 229'194 SSO -> 191'033 SPY 191'033 SSO
30.10.1996-31.12.1998
367'026 SPY 553'108 SSO -> 460'067 SPY 460'067 SSO
31.12.1998-30.06.2002
399'656 SPY 262'616 SSO -> 331'136 SPY 331'136 SSO
30.06.2002-30.11.2008
355'703 SPY 229'135 SSO -> 292'419 SPY 292'419 SSO
30.11.2008-30.04.2013
588'687 SPY 887'857 SSO -> 738'272 SPY 738'272 SSO
30.04.2013-31.07.2017
1'197'223 SPY 1'805'471 SSO -> 1'501'347 SPY 1'501'347 SSO
31.07.2017-01.11.2019
1'884'268 SPY 2'229'114 SSO = 4'113'382
No rebalancing: 4'022'696
100% SPY: 2'228'372

That's almost 31 years and almost no effect. It's pure coincidence that rebalancing turned out to be slightly better. But it might help you sleep better at night.

So even if we have a 2008 crash right now, we would be looking at:

100% SPY: ~1100k
50% SPY 50% SSO not rebalanced: ~1400k
50% SPY 50% SSO rebalanced: ~1590k

150% leverage is ****** awesome!
You guys are all over the place with this..... Ray first was backtesting to 1955 to the present.... and wondering why rebalancing helped so much. That's over 60 years.... not 31.

It's very simple..... your ongoing contributions are doing much of the natural rebalancing for the a very long time to begin with when starting from zero.... only once you get much further out would the annual rebalancing start making a huge difference. You are now doing $2k/mo over 30 years compared with $1k/mo over 65 years.....

Also since 1989.... ongoing consistent contributions have provided a time weighted ROR of roughly 11% for the SP500. It was a tremendous 30 years to be a dollar cost averager.


The idea that 150% constant leverage is mathematically ideal from history - going back to the 1920s - is something that's been discussed on here quite a bit already, and has been found in other academic white papers that are out there.

The root discussion of rebalancing really was why would you ever use SSO.... when you could get the same leverage at lower ER by using more SP500 and then adding UPRO.

Hydromod
Posts: 402
Joined: Tue Mar 26, 2019 10:21 pm

Re: HEDGEFUNDIE's excellent adventure Part II: The next journey

Post by Hydromod » Fri Nov 29, 2019 2:56 pm

Lee_WSP wrote:
Thu Nov 28, 2019 12:17 pm
I think we're not even looking at the same data. I still can't find the raw simulated data in his spreadsheet.
It's in the command tab in columns A through BA. These are hidden. You have to unhide columns A through BA. Poke around on the top to unhide.

Post Reply