Frank Grossmann

Long/short equity, momentum, bonds, newsletter provider
Frank Grossmann
Long/short equity, momentum, bonds, newsletter provider
Contributor since: 2013
You are right. normally I am using a volatility factor f>1 which favours low volatility in the formula: Sharpe = rd/(sd^f). You can calculate this with ETFReplay as you described.
Using ETFReplay you optimize for normal Sharpe (=Return/Volatility). During very volatile times you may optimize for minimum volatility, so only look at volatility. After a big drop you may not go 100% to the better ETF because of a possible mean reversion.
Dont rebalance too often and try not to rebalance in a market correction. You risk to switch 100 to "risk off" and this way you lose another time when the market bounces back.
100% TLT is not a good allocation. Rather use a longer look back period and use a minimum volatility allocation.
I would use the ETFReplay Backtest Portfolio option
Even without subscribing you can manually check for which combination of SPY-TLT you had the best Sharpe or lowest volatility for the last 3 months. Then just use this allocation for the next month.
We do monthly rebalancing.
I tested the strategy with trading 1,2 or 3 days delayed at close. I also tested the strategy with only liquid (>500M NAV) ETFs. It still works. Also I always run such a strategy at least for 6 month myself before I publish it.
For the Trading costs remove about 2% per year, but this depends a lot on your discount broker and on the size of your investment.
Backtests are always done like this. Very popular web sites like also do it like this. The backtests just verifies that a strategy worked in the past.
A lot of the selected ETFs have been really beaten down before and then have recovered and have been selected in the strategy. A good example is Ireland EIRL which was really beaten down together with Greece, Spain, Italy ....
Ireland really recovered well because they did a good turnaround job.
The 66 day lookback period was chosen, because the strategy delivers good returns around this slightly more than 3 month period. However the strategy works in quite a wide range. If the lookback period becomes too short, then you risk to react to every small drawdown by selling low and later buying high. We do not want to react on market noise, but reallocate to economic up and down trends in these countries.
One thing what you can be sure is that unless we launch such a strategy as a tradable product (ETF ETN or Fund) the strategy we will not be able to influence the underlying country indexes. This would need much more subscribers than we have now. These subscription strategies which you have to execute by yourself, only address a very small percentage of investors.
I will check ISRA again. EIS has the much better volume and lower spreads. This is also quite important for such monthly re-balanced strategies. Anyway, I always check the ETFs and add or remove if necessary.
There is the Treasury as a hedge, but not Gold like in the original PP
There are several ways to reduce the 36% drawdown. One is to add cash, but probably the best is to impose a volatility limit. Limiting volatility to 13% halves the max drawdown in 2008. I backtested these possibilities, but long term they always reduce the performance. Going to cash always means that you realize a loss. Switching to a different asset than cash always gives you the possibility of a recovery after a short market correction. This is important, because every year we have several short about -5% corrections.
However because this drawdown was more a mean reversion after a 35% Treasury spike, I don't think it was so painful.
I also run a Top5 Nasdaq 100 strategy built exactly the same way as the country strategy. It works also very well, but the advantage of countries is the global diversification.
Nasqaq 100 is one of the ways to beat the S&P 500. An other way to beat it is the same type of strategy with US industry ETFs. I run a Top5 of 27 US industry ETF strategy since half a year.
I don't like PHDG. It is hedged using VIX futures and I think that such a Hedge is way too expensive. The only way to make money with volatility is to sell volatility premium.
The strategy does also work reasonably well with cash instead of the Treasury. Also Gold as a safe haven asset works reasonably well. Treasuries however are by far the best way to reduce volatility and risk, even if the would just go sideways these next years.
Btw. US (Nasdaq 100) is also included in the top world country strategy.
You could buy ACWI - iShares MSCI All-World ACWI Index, but this way you only get mediocre results because such an all world ETF includes also all the looser countries of the world.
There are some months where the strategy only invests in country ETFs and no Treasury. We try to find the max Sharpe ratio always of Stock-Treasury pairs, but if the top 4 country ETFs all have the max Sharpe at 100% stock-0% Treasury, then at the end there is no Treasury in the Top4 ranking. But normally the best Sharpe nearly always has some 10% or more of Treasuries in it.
I may include it later on, but at the moment it is too small. It has an AUM of only 55M$
No, but we do an input parameter stability plot to check that the CAGR is stable over a large variation of input parameters. You can see one of these plots here:
No, there are no commissions and no spread or slippage taken into account.
The UIS strategy had a lower drawdown of 11% during winter 2008/2009 because it was more hedged and SPY is a very low volatility ETF compared to average world country ETFs. However this drawdown during the sub-prime crisis was followed by a +35% panick spike of the safe haven TLT Treasury and it was more of a mean reversion.
At the end the strategies where about flat after these ups and downs during this 6 month period.
The strategy does go partially to cash if no positive sharpe ratio is found for all possible SPY-TLT allocations for the lookback period.
The tax problem is a big one and I personally as a foreign investor don't invest in normal US bonds because of this. There are some ways around it. If you for example invest in leveraged Treasuries (TMF instead of TLT), you do not pay taxes. Another way is to invest in Treasury futures. This way you get the dividend as roll yield.
It is possible that a strategy like UIS does nortwork well for some time because of rising rates, but with strategies it is like with ETFs. If a strategy underperforms, then you have to switch to better performing strategies.
These strategies only work if you do daily rebalancing. Then you earn about a calculated 9% per year. Now substract 4-5% borrowing fee, spreads, slippage and commissions and you will be down to 3%-4% annually.
The risk however is high with ETFs like UVXY. If there would be some terrorist attack in New York or something similar, then you can still have huge losses, because your XIV position is only balanced with UVXY within small VIX variations. You can see this already if you switch from daily to monthly rebalancing.
I tried all this, but finally the 3%-4% yearly was not worth the risk.
It is just very difficult to construct well working strategies with inverse ETFs. Try for example a strategy which switches between SPY and SH which is the inverse SPY ETF. It is much simpler to do a well working strategies with SPY TLT or SPY GLD.
Or try to do a strategy which always shorts the worst ranked ETFs out of a selection which would be similar to go long in inverse ETFs. This just does not work unless you can really time very well the trades.
One thing I have learned is never to invest against a long term trend with ETFs. So, even if Treasuries will go down for some time, long term they always go up. So I would not try to invest in a -2x leveraged Treasury. Same is for volatility. I would never go long VXZ, because this ETF is nearly always going down. It is this general trend which makes that you have more than 50% winning months.
If both SPY-TLT would go down for a longer period, then our strategy would reduce the total allocation up to 0% = cash. There may be quite periods where TLT shows positive correlation to SPY. However if there is a bigger market correction, then TLT normally quickly gets back to its safe haven role with negative correlation.
The correlation in normal market periods is less important.
Yes, here I meant to hedge the stock market ETF SPY with an inverse Treasury or in this case a short TLT put option.
There are two important facts to add. If your investment is flat, then you are right, leveraged ETFs are slightly worse than the normal ETFs, however if you rebalance or your investment goes only a little bit in the normal long term up direction, then most of the time a 3x leveraged ETF is the better way to go.
Here an example:
you buy 100'000$ TLT and 33'333$ of the 3x leveraged TMF (=3xTLT) on January 1. 2014. Both positions are substantially the same.
April 13. 2015 your two positions have the following value:
TLT = 131433$ = 31433$ profit = 30166$ profit after tax*
TMF = 70390$ = 37057$ profit
*In the TLT profit, there are 4224$ dividends of which you have to pay 30% witholding tax (=1267$). Leveraged ETFs normally do not pay dividends.
Another important consideration is, that you only have to invest 1/3rd in the leveraged ETF. This means also that as a worst case scenario you can only lose 1/3rd of your money.
So the 3x leveraged TMF is the better investment than TLT.
No, I sold SPY and TLT puts as a replacement of Spy and TLT. Selling puts is similar to a long position of the ETFs when they have a long expiry.
As this is only an US strategy, it is well possible that we have a correction of both, US stocks and treasuries.
Then you probably need to change to a strategy with more international exposure.
You are right. The strategy profits a lot from that negative correlation. I don't know how long this will go on like this. Maybee some more years, but one day you will probably need to change to another strategy, however until then you should profit.
Strategies are not made forever. Its the same like for the underlying ETFs. From time to time you have to change to another strategy which works better.