The thread on contrarian investing got me thinking especially on Steve’s TLT position. I was wondering what people thought of using UST (2x intermediate treasuries) in place of TLT. Seems like there is less duration risk but you get the same or better hedge/performance as TLT. Although the duration risk may seem the same as TLT, it should recover faster in an increasing rate environment than an equivalent TLT position.
Charles - I would stay away from leveraging on this.
Steve
Are you saying you find UST more risky than TLT?
Goodness yes. I addressed leveraged ETFs in that pdf I uploaded a couple of posts ago.
Interesting. I disagree that 2x leveraged intermediate term treasuries are more risky than unleveraged long term treasuries.
UST is less volatile than TLT, so any volatility drag for this fund is actually less than that of TLT. I don’t see why it should be riskier though I understand returns will not be 2x the underlying index over holding times greater than one day.
There’s a good paper here:http://ddnum.com/articles/leveragedETFs.php
that addresses the mathematics of the risks of holding leveraged ETF’s long term. The author finds that historically for most equity markets 2x leverage was near optimal (increased returns with limited tracking error). An understanding of the math or a quick glance at the charts of leveraged ETFs during 2008-2009 shows that volatility decay, the primary cause of concern for leveraged etfs, quite naturally varies based on the volatility of the underlying index. So leveraging a volatile index is much riskier than leveraging a not so volatile one.
Interestingly the paper finds that the extra fee’s associated with leveraged ETF’s are in many cases a bigger drag than the leverage of the ETF itself.
Don - I only know that there were class action lawsuits over 2X stock market index funds that lost money during a period of time where the index increased significantly. So this isn’t just media hype. And from my experience, leveraged ETFs are “hot” to touch, much more so than non-leveraged ETFs. It is hard enough using stops with non-leveraged ETFs (actually stop/limit orders which seem to be the norm after the flash crash).
Perhaps it is my biased opinion but I would stay away from the leveraged products.
Steve
I heard the warning about inverse/leveraged etfs just last week at a Stock Twits investment meeting. The presenter had something like a 10yr chart of SDS. With a painful look on his face he make a proclamation to never hold these ETFs overnight!
However, I see no problem holding these ETFs long-term if they’re a hedge and if one rebalances often.
Since book sims now support ETFs, it’s easy to sim 2*SPY + SDS. With weekly rebalancing, you would have a slight loss after 10yrs (0.20% slippage).
Walter
PS Today is SDS’s birthday! Inception date was 7/11/2006.
Don,
Interesting paper. Thanks for the link. I looked at the curves in the paper and I could not help but think it looks exactly like what happens with Kelly betting, optimal F, optimal Kelly or Kelly criteria: whichever term you like. In other words there is an optimal bet size that maximizes returns. Sometimes–especially with ETFs that limit your potential losses–the optimal Kelly bet can involve leverage.
The curve always increases up to optimal Kelly, decreasing afterwards and usually going to zero at about 2X optimal Kelly. There are papers that confirm that they are equivalent concepts. But they usually talk about geometric versus arithmetic averages: and are hard to read. I did not find one worth linking to.
But clearly it is like placing a bet each day and determining the proper bet size for each of those bets.
Bottom line. There will always be an optimal bet size. Sometimes that bet size will involve leverage. And it is always easy to determine optimal Kelly using past data.
Knowing the optimal bet size in the future is where the risk comes in. The other source of risk is being sure ahead of time that you will not capitulate or wash-out at the bottom of a drawdown.
-Jim
Can someone summarize why holding SSO or QLD is bad over the last ten years? The returns are better especially for QLD. If you had held QLD over the last 3 years you would be up 27% and never made a trade. That’s better than 95% of smart alpha. Why is that bad? Yes I understand the drawdown can kill you but the strategy I would employ is:
- I only use SSO and QLD. The leveraged bonds don’t seem to work as well.
- Use market timing even a simple moving average
- Always use with other uncorrelated strategies in a book
- The last one which I am considering is reduce exposure as we are in a 8 year bull market.
These are all the same strategies you would use if you were not leveraged but being leveraged increases your returns. What are the risks:
- The markets drop by 50% over night not likely but the markets will drop by 20% or more over 2 weeks and the market timing will kick in. That’s what the simulations are for.
- The assets your hedging with like TLT implode at the same time. Possible but if that happens everyone is in big trouble. I am Hedging with Gold also.
- The other uncorrelated assets in your book all crash and burn. Again possible but what is the alternative?
What am I missing?
MV
I think hedging with short SSO or QLD makes a lot of sense. If you have a good model then hedging short 35% SSO works well. QLD is not available in the P123 hedge module.
This is a better hedge than Treasuries, especially now when yields can not go much lower. Assume market tanks 30%. The 35% short SSO protects you with 2x0.35 x 30% = 20%, so your net loss is only 10% assuming your model tracks the general market. Hopefully your model does better than that.
Now if the market gains 10% before the hedge is off, then you lose 7% with the hedge, but you still have a small positive return of 3% if your port follows the market.
TLT or other Treasuries provides no certainty of protection, whereas the 2x leveraged SSO does.
Geov,
Why short SSO vs going long SDS?
Walter
Geov - you can find yourself losing in good times and bad if the long stocks do not similar in composition to the S&P index, in which case there may not be any advantage. What we do need are adjustable book percentage holdings, adjustable over various market conditions.
Steve
Walter,
for long SDS you have to sell a portion of your holdings. If you have a good model which does better during down-markets than SPY then it is advantageous to be short SSO.
Hi Geov,
Right, you need margin to short. Too bad that can’t be done within IRAs. But also, the short seller has to pay out the dividends that SSO provides. How expensive is it to carry a SSO short.
Walter
Walter,
SSO dividends paid and carry costs are included in the P123 simulations and ports. You need to sell 26% of your holdings and buy SDS for a short:long ratio of 0.7:1.0.
You can buy SDS on margin. Depending on your broker, you may find that there are some outrageous borrowing fees for shorting 2x leveraged ETFs such as SSO.
Steve