How you can Lower Volatility without giving up Alpha

Home Forums Logical Invest Forum How you can Lower Volatility without giving up Alpha

Viewing 10 posts - 1 through 10 (of 10 total)
  • Author
  • #80134
    Mark Vincent

    We all want the holy grail which is low volatility and high Alpha. Unfortunately it is very hard to achieve and requires some tradeoffs. My hypothesis is the markets will continue to go up over the next couple of years but volatility will continue to increase. Therefore how do you lower volatility without giving up a lot of the Alpha? By adding VXX to any strategy you will lower the volatility and in some cases increase the return. I have used the UIS 3*(SPXL TMF GLD3*) strategy as a starting point and modified a few of the parameters. I also added it to the US market strategy with surprising results. The problem we all have is how do you choose between all of the different strategies and parameters. What would make me confident to put any of these strategies in production and use real money? The following is some analysis I did that I hope you may find useful any comments are appreciated as always.

    Fixed Parameters
    All optimizations were run over 3 years and appear to me to be stable. My definition of stable is there was stable returns around the lookback and the attenuator parameters that were choosen. Most of the time a large white area.
    VXX has a Minimum Allocation of 10%. (Since you cannot predict market corrections)
    Max Allocation = 60%

    Longer rebalance periods work better for certain strategies but may pose more risk.
    All strategies tested benefit from adding VXX by reducing volatility. Even the US Market strategy that is not leveraged was able to reduce volatility and Drawdown with not giving up much Alpha.

    Results and Rebalance Frequency

    Strategy CAGR Sharp Vol Draw Reb Per
    UIS 3*(SPXL TMF GLD3*) (M 14) 41 1.4 28 41 M
    UIS 3*(SPXL TMF GLD3*) +VXX 30 1.6 18 16 M
    0 SPXL Hedge2x VXX (Q 176 DR) 38 2.2 16 11 Q
    0 SPXL TMF UGL VXX (Q 186 SR) 46 1.8 24 18 Q
    0 TQQQ HED 3X VXX (Q 192 DR) 65 2.5 25 14 Q
    0 US Mkt VXX (Q 168 DR) 22 2.5 9 9 Q

    If my forecast is correct and volatility does increase over the next 3 – 5 years than adding VXX will help most strategies. The argument against this is that over a long period of time you under perform the basic strategies with a hedge. During the period of 1/1/2013 – 12/31/2016 most of these strategies did not do well. That is a long time to wait for alpha. VXX was down over 50% and the Hedge was also negative.
    All strategies have a lower drawdown which I personally like.
    One way to make the strategy better is only use VXX when Volatility is over 30 and leave VXX for 6 months or until volatility is under 20. I have eyeballed the CNN fear and greed index and when it is over 80 use VXX for 6 months shows some promise but that gets into market timing and it may not work. Testing these ideas are not possible in QT but they may help me sleep better at night when I am using VXX as a Hedge.



    I am not a fan of VXX because if you keep it, then you will always lose money and timing is difficult. Better use long term treasuries like TLT which have a negative correlation to equity and will so reduce volatility. To increase income you can sell calls (covered call) on your position after a treasury spike due to a market correction. This works specially well because inverse to equity, TLT volatility increases when TLT spikes up and in such cases you can sell some calls and cash in a good premium. This way you have a very safe investment which produces quite a good income because of the covered calls. It’s a strategy I personally like to do, however I only do it, if TLT volatility (symbol VXTLT) is higher than about 14. Also I only “cover” about half of my position with calls.
    I have to say that instead of doing a covered call which is to buy 100 TLT ETFs and sell 1 ATM (1 month) call, I prefer to write just 1 ATM (1 month) put instead. This is exactly the same but needs one trade less.

    Mark Vincent

    Thanks Frank,

    It’s on my todo list to learn options.

    Mark Vincent

    Frank for those of us who don’t want to use options what do you think of an ETF like TAIL? The backtest looks great. I think to good and I probably curve fitted it with quarterly rebalancing.

    I used Tail with QQQ, HEDGE and achieved following over 3 year backtest:

    CAGR: 25%
    Sharpe: 3.0
    Volatility: 8.5
    Draw Range: -4.1

    Note: I listen to a lot of Resolve Riffs and the focus is on tail risk these days and during market drawdowns liquidity dries up and everything goes down except volatility. LI does not have this in the models I know you trade options to control this just wondering if there is any other way?



    TAIL behaves very similar to a leveraged TLT. I did a quick test of SPY/TAIL 70/30 and SPY/TLT 50/50 you can see here at Portfolio Visualizer. TAIL only started in Jan 2018 so there is not a lot of history, but you can see for similar Stdev, the TAIL portfolio lags in CAGR, MaxDD, Sortino, etc.

    Mark Vincent

    TAIL is not suppose to make money it is only to reduce left TAIL risk.
    Tail is more uncorrelated to stocks than TLT.
    TAIL works better than VXX since it has very little drag.

    Yes it has very little history but it hard to create a strategy over the last 3 years that beats the one I posted. Almost to good to be true.

    Frank Grossmann

    TAIL is buying OTM put options and invests the remaining capital in Treasuries. If you check the chart then you see that it performed very bad 2017 and 2018 when Treasuries did not perform as good as the last 2 years. Buying put options is like buying VXX a very bad strategy to protect your portfolio

    Mark Vincent

    Yes Tail is highly correlated to VXX but with a lot less volatility. It’s only going to go up when the VIX goes up. Just like VXX it lowers volatility but will have a drag on returns. If you don’t want to trade options how can you have a Hedge against left tail risk when all asset classes (Bonds, Gold, Stocks) all go down at the same time. Over the long term your hedged but as we saw from the last few drawdowns during the month all asset classes went down except for volatility. Having a permanent volatility Hedge without options seems impossible for the short term drawdowns? It comes down to are you willing to live with monthly volatility being high and GLD and Bonds will kick in the next month or do you want that permanent protection from Tail or VXX and live with the Drag on returns.

    None of the data providers LI uses provide the full history of TAIL is there a reason why?


    Why not use them all–VXX, TAIL, and TLT. They are all going to behave differently. This month TAIL is up 1% while VXX is down 7%. In March VXX was up 102.76% TAIL was up 11.18%. TAILs worst month this year has been down 4.61%, VXX was down 18.12% so TAIL is something you can be “wrong” on and still be ok, it can be painful to be “wrong” on VXX. But, if we have another downturn like March would be pretty nice to have something around that is up 100%. Would revisit the idea of having a max exposure to VXX of 60% that can get pretty hard to stomach sometimes. You can also backtest using VIXY has a longer history.


    VXX has about a 4x higher volatility than TLT or TAIL. It is very difficult to combine such different assets in one strategy. You will not find a setting where QT is able to switch in time to the best asset. Mainly VXX is the problem as it tends to go constantly down until a market correction happens. No way to use the past performance to allocate it before a crash. VXX is something for day traders as you should keep it only for short time. Normally if VXX had an up spike resulting in a good performance, QT would perhaps allocate to VXX. But then it is already too late because VXX always goes down very fast if volatility drops. TAIL may sometimes be flat or even go up as it buys OTM puts and profits from Treasury performance, but also for TAIL the real profit component is not the put options but the Treasuries.

Viewing 10 posts - 1 through 10 (of 10 total)
  • You must be logged in to reply to this topic.