You can do some other things. First, give away all your income producing assets (joking).
SOME ‘solutions’
Or, try to build sell rules to maximize ‘losses’ and the ‘acrruing of losses.’ For example, on the ETF front, somebody like the roboadvisor Wealthfront will basically sell all ‘losing ETF’s’ from the buy and hold ETF portfolio and replace them with similar ETF’s. In this way you are accruing tax losses to offset future gains. This is a smart thing to do and fairly easy. You can build these in P123 fairly easily.
Re: FEATURE REQUESTS that have been around forever. It would help if P123 showed after tax returns and allowed users to input ‘tax assumptions’ for short and long term gains. So, the system didn’t have to calculate tax rates, but did calculate after tax returns based on user specified tax rates. This has been a feature request forever.
Re: STOCKS. The question is, at what point to cap gains drags on compounding outweight excess returns of shorter holding periods. I have built many long-hold fundamental based models for myself and use one (a yearly rebalance 30 holding system). I settled on 1 year for this. I actually offered one ( a 50 holding large cap system) for 6 months as an R2G and it did better than many of my higher-turn systems out of sample (both pre and post tax), but no one signed up so I took it down a few months ago. However, given the short data history in P123, you are basically betting on a philosophy (i.e. quality-value forever) - not on a backtest. Holding periods longer than 1 year is not really something you can get any meaningful backtests on.
Re: ETF’s. For the dynamic hedging to work, would be very helpful if P123 significantly upgraded the book capability. If you could add a) variable leverage at the book level based on rule based conditions (i.e. book level DD or subsystem performance evaluation) and b) variable hedging or use of additional ‘systems’ based on rule based hedging. So… bear market one system is added to a book if and only if it’s ‘in the market’, that kind of thing. If you want ‘simple options’, just create a basket of hedged systems and/or a book of only timing systems that enter SH or short IWM or whatever, when various conditions are met. They will gradually hedge the total portfolio.
PRACTICAL SUGGESTIONS:
Divide port into buckets. I have about 33% of what I do in buy and hold ETF’s that I annually rebalance if/when they are more than 5% out of whack. It used to be a lot more.
Know why you are holding different buckets. My higher turn stuff is intended to be much better on a risk-adjusted basis and ‘smoother’ in terms of returns (lower DD’s). If it doesn’t achieve this, I have to replace it.
Re: Expectations / Place in the market cycle. Discretionary market timing is not even close to an exact science and can be off by years. However, if you believe buy and hold will return, say 4%/year for the next 5 years and vol will be high, then a 20% reduction in annual returns paid to taxes is costing you under 1%/yr and higher turn systems may give much better risk-adjusted results. If you think buy and hold will get you 20% AR / yr over next 5 years, then the 20% reduction for annual rebalance is 4% / year roughly and may be better in buy and hold. So, I would suggest a ‘market timing’ / ‘market cycle’ component to the bucket sizes. At the current time, I would be more inclined to hedge more and be long less.
If anyone knows, I am interested in how strict the “Wash Sale” rules are. If one of us rebalanced once per year–possibly holding the winners 366 days and the losers 364 days–and followed the rebalance recommendations would we be likely to get into problems with the wash sale rules? Would some of the new stocks in the portfolio be similar enough to the sold stocks that the IRS would give us a problem?
The wash sale rule doesn’t apply to the sell of one stock and then buying a different company even if the new company is in most respects similar as the sold company. In fact, it doesn’t apply to selling the common shares on one company and buying the preferred shares of the same company unless the preferred stock is convertible into common stock without any restriction, has the same voting rights as the common stock, and trades at a price close to the conversion ratio.
That includes selling one index ETF and buying a different ETF based on the same index.
I’ve tried to create versions of my ports that hold stocks for at least a year, but the potential tax savings didn’t seem to be large enough to offset the lower returns from being constrained. Perhaps others have had more success, like Steve’s R2G.
My solution thus far has been to maximize contributions to tax-advantaged accounts.
Have you heard of the backdoor Roth IRA that could add $5,500 per year?
As always, thanks for the great advice. The tax feature, and the improved dynamic hedging Book capabilities would be great.
A simple approach to the dynamic hedging book might be to combine multiple hedges in a book simulation. For example, include your favorite system, and include 2 TLT systems. Each TLT system buys and sells under different conditions. Adjust the percent contribution of each TLT system in your book. The problem with this is that the book holds back cash for each TLT system that is out of the market.
Before Book was available, I had created my own version of Book using excel. Exact same concept as the book, except when systems go out of the market it puts the resulting cash back into systems that are still in the market. I could dust it off if interested.
The performance isn’t great but slightly outperforms the S&P 500 equal weight benchmark. I would have to put a lot of hours in on this to get better performance and I don’t have the time.
Losses are cut short so you can write them off against winners in other ports. The winning trades are held greater than 366 days. You would have to do your own hedge to cover the 2008 bear market. I did a 10 year run because there aren’t that many ETFs prior to 2005.
If anyone comes up with an improvement to this port then perhaps they can share/make it public.
I made a few mods to your ETF Test Sim. Here is my mod Sim .
I moves your Taxonomy buy rule to a custom universe so I could play with Rank rules.
I added RankPos < 4 to your first sell rule and removed the GainPct>60 requirement.
I added a simple bench buy rule requiring the 6 month bench return to be > 0% or the 1 month bench to be > 10%.
The annual return = 16%, the max DD = 20%, and the % winners increased to over 50% from 30%.
1/2 of the holdings were held for over 1 year and all of the rest were sold with loses between -4.4% & -13.6%.
Thanks Denny - the issue I have with you mod is that the improvements have to do with market timing, whereas I think Stu wants to keep his hedging separate?
Nice work. A consideration is that this etf systems performance will change based on which provider is chosen for a starting universe (try putting in the other etf providers) as well as the type of etfs that are added to that providers universe.
Tom,
It would be great to be able to create a post-tax backtest. This might encourage development of a tax efficient category within r2go. I would also like to see leverage at the book level as you suggest.
Alan,
The backdoor traditional to roth ira conversion is a great way to diversify your tax risk and is underutilized. The mega backdoor ira conversion is difficult to implement unless you meet certain well defined criteria.
All,
The etn, vqt (switches between spy and vix), is a reasonable tax efficient replacement for a portion of a bond portfolio.
Scott - I haven’t tested Denny’s version but mine gives similar results for iSHARES (in addition to the original SPDR). Both of these providers have 80+ long equity ETFs and a lot of the ETFs have a narrow focus i.e. specific industry/sector. The ranking system demands the strongest performance over the last 1-2 years in order to be picked and then sells a year or more later with large % gain. This is why one needs the narrow focus, very difficult to do this with a broad ETF. The rules are not suitable for other asset classes such as Fixed Income or mixed assets, I’m not sure about commodities.
Steve and Denny, thank you! I need to spend more time reviewing both systems.
Here is something I put together quickly as an example of what I’m thinking. This needs work as it is not diversified enough, and the hedge needs work. Additional hedge systems could/should be included in the mix. Unfortunately, the overall return assumes only 80% cash invested in the market when the hedge is out of the market. This needs to be fixed as a book option. I wish I had more time to experiment.
[quote]
I added RankPos < 4 to your first sell rule and removed the GainPct>60 requirement.
[/quote]With the rule “RankPos<4” you are selling the etf if it is in the top 3 ranked etfs. Is there any reason for this or did you mean “RankPos>4”?
Wow! KJ you caught a random typing error. Missing the > key and Hitting the < key instead sure makes a BIG difference. That is what happens when I only spend a few minutes trying various rules. The mistake significantly improved the results so I just went on and didn’t recheck the rule. Since the Sim only traded 29 ETFs over the full time period, a random goof can make a big difference where if the Sim traded hundreds of stocks it would have made the results worse.
This is a good example of why we must understand why a change made an improvement and not simply accept every change that improves a Sim. Although selling the stocks ranked 1 through 3 and keeping stocks ranked 4 & 5 improved this Sim, I see no reason to believe that will work going forward. Data mining at its finest!
I see a reason that your random typing error might work going forward. In a word (okay 4 words): reversion to the mean. There is good reason to think that the ETFs that have performed the best over a long time period (i.e., best one and two year sortino ratios in this case) will revert to the mean (in a negative way).
I run no ETF sims but excluding stocks that have performed exceptionally well over a long period does lead to a mild improvement in my stock sims.
For stocks at least, I have found this to be too dependent on market conditions to be useful.
My thoughts exactly. The current hedge shorts SSO which is a 2x leverage S&P500 ETF.
My general assumptions…
Speaking in rough numbers, assuming 100% invested cash when hedge is out of the market (instead of 80% invested cash using the Book…please fix), I believe the return would increase about 2% to around 13% in this systems example. I believe in my tax bracket, the equivalent annual return using a system with short term cap gains would need to be roughly 40% higher or 13%/.6=21%. A system using 1 year hold time qualifying for long term capital gains would need a return roughly 20% higher (15% fed, 5% state) or 13%/.8=16%
Stu / Scott - I’m not really following your trail of thought. For tax purposes, you pay capital gains the year you sell a security. (I assume that is how it works in the USA). So if you short sell, then this presents a problem because you will have to buy back the ETF at or before the end of the year, thus incurring short term capital gains, what you were trying to avoid.
[quote]
I see a reason that your random typing error might work going forward. In a word (okay 4 words): reversion to the mean. There is good reason to think that the ETFs that have performed the best over a long time period (i.e., best one and two year sortino ratios in this case) will revert to the mean (in a negative way).
[/quote] Jim,
While that may be true, you cannot have two competing philosophies in buy and sell rules. Your buy rules are buying the highest ranked etfs (assuming they will go up) while your sell rules are selling the top ranked etfs (assuming they are done going up).
KJ
Yep. Better ways to do the sim I’m sure. Finding why a typo works would just be a start. And as I said in my post, the improvement–even when the buy and sell rules match up–is not that impressive for a stock sim.
