Posted on

More on “Outsourcing”

In the “real” world, Hedge Funds pay their staff. In the fake world of “Outsourcing” they don’t. Rather like hiring an indian from Bangladesh living on the floodplains to do your telephone sales – but at least they get paid SOMETHING. I agree that technically Q have come up with some good research and a good back testing engine (if you have the infinite time and patience to look through the more than voluminous source code).

I very much fear that no statistical test and no amount of live trading verifies anything very much. How many fund managers let alone hedge fund managers have we seen come and go after stratospheric periods during which they appeared to be omniscient. JW Henry and Bill Dunn are classic examples. Maybe their time will come again, maybe not. There are periods where a strategy fits the markets and rather more periods where it does not. But this is not a diatribe about the foolhardiness of prediction…so lest I get carried away I’m just saying that people on the end of the outsourcing chain, the mugs hoping for some tiny allocation, are about as likely to achieve their dream as the average lottery ticket punter.

The financial world in particular is overly populated by FWOT: systems, participants, the works. A giant “Bonfire of the Vanities”

My distaste and extreme scepticism result from years of disappointment following contacts on the internet. Magazines who pay nothing for articles they have published, conference and seminar promoters who expect you to perform for nothing when they are charging punters a couple of thousand quid for entry, an endless stream of people who want to talk about ideas or proposals which inevitably lead nowhere. Life is too precious to waste on such nonsense.

Posted on

Hedge Fund “Outsourcing” and the something for Nothing mentality.

Is all well with Quantopian? They seem to have gone remarkably quiet since announcing the tie up with Big Steve Cohen earlier in the year.

I dropped in recently and the activity on the forum seems to have declined. The blog by the (CEO?) or whoever seems to have been dropped and they have disappeared from the news.

One problem in my view is that without exception EVERYBODY uses the internet to try and get something for nothing.  And “outsourcing” by hedge funds seems to be the latest example of that. To be fair Quantopian provides an excellent on-line backtesting engine for US stocks which is scott free and includes data. Their technical staff is excellent and their is much to be learnt from analysts such as Thomas Wiecki.

The problem is that it is hard to make any money – either for the punters who use the platform or the punters who provide it. For one thing given the open source environment for software and data most of us can program our own back testing engines and source our own free data. For another we don’t really relish working on systems and publishing them on websites just for the hell of it. It would be nice to actually make some money for a change from a lot of hard work. Probably difficult unless you have thighs to flash like the Kardashians or the bare faced cheek and rhino thick skin of some of the truly awful systems sellers out there on the internet. The systems sellers seem to profit from selling dreams to fools and by getting paid for books, seminars and general buffoonery.

If such people were the magicians they claim to be they would not need to engage in shameless self promotion – they would trade for themselves.

Back to Quantopian and their ilk.  They are an excellent bunch but have lost their way. Finance is not a hobby for people like myself – it is a way of life and has been for almost 35 years.  I have had good years and bad but at my age I would welcome taking less risk – hence I have cut way back on my own trading.

But that does not mean I enjoy giving my expertise away for free – and that is what the internet seems to require. Unless you are a seller of impossible trading dreams with a huge ego and the cynicism and morals of Bernie Madoff.

Guys like Quantopian (and there are quite a few others out there by now) need to find a way to recompense good quality research. At present the “carrot” is that by providing algos and entering into trading competitions you may (eventually….or may not) receive an allocation to trade and profit from that.

Well thanks but that does not pay the rent, nor will it ever. It’s a something for nothing world out there and most of these “outsourcing” hedge funds have teams of in-house quants anyway.  And they have to pay those guys.  Not so the fools who plug away day in day out publishing their research on their own blogs or, worse, on competition websites.

If the outsourcers found a fair way to pay for the outsourced research, matters might take on a more interesting mantle.

Until such time, goodnight sooty – why should we bother?

Posted on




I was flattered to be asked by Christopher Parker of Harriman House to contribute a chapter to their Book of Investing Rules.  Other(rather better known!) contributors include Jack Bogle,William Bernstein and Burton Malkiel.

The book will be released at the end of October.

Here is my chapter.

Since I was not paid for writing it, I hope Harriman House will have no objection to my providing it here.

Posted on

Vix Calls to Hedge Short Vol Strat?

As a brief addendum if you treat the VIX as a mean reverting series (which it is) you can choose to enter monthly calls (weekly, whatever) only if the VIX is above a certain level (IE above its normal range).

The sample size is likely to be small and the scheme very prone to curve fitting.  But it is worth consideration.

Posted on

Call Options for Hedging – FWOT

Like the great herd of sheep like morons you read about in the financial press and blogosphere I am currently short volatility having only discovered the trade last year. As a Johnny Come Lately (as usual) I have spent many weeks looking into protecting my long position in the inverse volatility ETF XIV with various hedging methods.

I have already written a post on hedging it with TMF which has been highly successful so far but will undoubtedly let me down at some crucial point in the future.

I then turned to options and bought VIX data from CBOE.  I have spent weeks cleaning up the messy and badly presented data and then turned to writing a back testing program in Python/Pandas to back test various option strategies.

I had never bothered much with options before partly because I was put off by the ridiculous names people give to their various strategies and partly because the largely useless “greeks” were all greek to me, and in any event have absolutely no predictive value whatsoever.

Here is the punchline: protecting your position with long call options is a FWOT unless you are an expert in market timing and know exactly when to put the hedge on. If you are permanently long puts you will destroy your returns with the huge cost of the option premiums. And if you think you can time markets save yourself the bother and expense of options and simply exit you short volatility position.  

Having disappeared down the options rabbit hole for a couple of months I am now absolutely staggered by the sheer bloody nonsense out there  and the plethora of stupid or dishonest sharks peddling schemes to the hapless fools seeking Eldorado.

It is particularly interesting to see that all the CBOE indices involve selling options not buying them. Option selling is not without its huge perils but compared to option buying it is the holy grail – which is actually damning the practice with faint praise.

As anyone who has bothered to do any back testing with be aware, option selling can be every bit of much as a disaster as option buying.

So why, in the name of god, is there an entire vast industry out there built around option trading? The only answer has to be “because there are a lot of bloody fools who think they have discovered the philosopher’s Stone”.

I have almost come to the end of my option Odyssey but my last vague interest is in back testing The CBOE S&P 500 Range Bound Premium Income Index Series . It looks interesting prima facie but as is usual with any trading scheme scheme it relies on historic data to predict the future (in this case the historical volatility of the S&P 500).

I will report further but suspect that a promising idea will turn out to be an over optimistic chimera.

Posted on

CBOE Option Data – Caveat Emptor

I recently bought EOD data stretching back to 2006 on an option series from CBOE. Frankly, the data is a disgrace and I got no reply from my complaint to CBOE.

Problems include strike prices out by up to 3 decimal points in some cases and phantom expiration months which suddenly appear and equally suddenly disappear.

I am having to go through the data line by line to correct the errors. Any attempt to perform systematised analysis on the data I bought was a farce before the tedious hand conducted clean up.

I also have a problem with the expiration dates. There is no trading on the day of expiration – instead the 0’s are shown for each strike. It seems absurd that the exercise-settlement value is not stated and there is no way to calculate or approximate it since the data does not even include the day’s spot opening price for the VIX.

Unless I have badly misinterpreted something (always possible) I would say shame on the CBOE for selling such misleading data.  Its value for my purpose is rendered almost useless.

Posted on

Gold Currency Strategy

I recently looked at Logical Invest’s Gold Currency Strategy

Here is their summary:

  • The Gold-Currency strategy trades Gold ETF vs 3 major currencies.
  • It is based on the negative correlation between the Gold ETF and the U.S. dollar Index.
  • It is an excellent addition to existing equity or bond portfolios as it holds very little correlation to either.
  • It can be traded using ETFs, Futures or even low-margin/low-cost FX pairs.

It is always partly invested in gold and partly in the best performer (presumably on a momentum basis) between Long USD/ Short Yen, Long USD / Short Euro, Long USD / Short Australian Dollar.

The weighting scheme is the same as for their Universal Investment Strategy but it is unclear what parameters they used.

I used the same parameters as they recommended for the UIP:
Increment (for gridsearch) = 0.01
Lookback (days) = 73
Volatility Factor = 2.5
Risk free rate = 0.0

I rebalanced the weightings monthly.

I am deeply averse to fiddling with parameters to obtain better results and thus left these parameters well alone.

Logical Invest tested back to 2009 which did not suit my purposes at all so I took futures contracts to represent the currency pairs and spot gold right back to 1987. I did not add any interest on deposits, however neither did I include management fees.

The results were underwhelming over this period:

CAGR 4.69%  Max DD 41.84%

Very similar to the metrics for a simple buy and hold of spot gold.

I took exactly the same idea but used a more traditional Markowitz mean variance optimisation on the same portfolio using risk targeting (IE the highest return for the desired level of risk). I targeted risk as medium (10 on a scale of 1 to 20). Again I rebalanced monthly.

I got a much better performance:

CAGR 7.35%  MAX DD 23.58%

I am sure if I had fiddled with the UIP I could have improved the results.  But that is a curve fitting death trap.

Annualised standard deviation for both systems was around 15%.  I don’t have huge interest in this system although I can see it may have its uses as a portfolio “diversifier” so I won’t post the usual plethora of charts. Just the equity curve for my version of the system. The benchmark is spot gold.



Posted on

CAGR 40% – Buy and Hold Inverse Vix + 3x Long Bond

I’m going to make this brief. And I’m going to start with a couple of caveats. The major caveat is that this is only a back test and as ever it is highly dangerous to forecast the future based on the past. The other caveat is that the specific features which make this strategy work (contango in the VIX futures and negative correlation between the ETFs XIV and TMF) may not persist in the future.

This strategy has worked during the period of the test because of two factors:

  1. Shorting the VIX has been profitable since VIX futures have been in contango 80% of the time. This means you can short the new front month at a higher level than the expiring month you are exiting. Technically speaking the market is in contango most of the time.
  2. The long bond during the period of this back test has had a negative correlation of -0.4 to the inverse VIX. The best way to short the VIX for the man in the street is to use an inverse VIX ETF such as XIV. For the long bond you can use an ETF such as TLT or a geared version such as the 3x geared TMF. Negative correlation means that over time (in back testing at least) XIV and TMF have tended to move in opposite direction.  A loss on one instrument has tended to be mitigated by a gain in the other.

Without further ado I will proceed to the back test.  As long suffering followers will know, I have ceased to “believe” in trading where there is a “probabilistic” edge and thus concentrate these days on letting the market give what it will give.

For the 3x geared long bond I have used TLT returns multiplied by 3 since it has a long price history back to 2002. For XIV (short the VIX) I have used actual returns of XIV backfilled by futures market data to 2004.

You buy XIV and TMF and rebalance each month. It does not seem to matter which day you re-balance on – a big bonus and a sharp contrast to many systems I have looked at.

I have used a simple risk parity weighting strategy and a 1 month look back period to assess volatility. The look back period is robust to varying this parameter but unsurprisingly CAGR declines somewhat as the length of the look back increases to 12 months.

I’m not going to comment further. This is a very high risk strategy and I am trading it in small size. BUY and HOLD of short VIX position over the period would have produced a similar CAGR but with a far higher volatility and drawdown.

This “system” or combination of holdings has in back testing produced an absolute and risk adjusted return far higher than the benchmark.

I will leave it to readers to ponder this monstrous scheme and draw from it whatever conclusions they wish.

Stat                 Inverse Volatility    S&P 500 TR
-------------------  --------------------  -----------
Start                2004-04-26            2004-04-26
End                  2016-12-20            2016-12-20
Risk-free rate       0.00%                 0.00%

Total Return         7671.36%              160.47%
Daily Sharpe         1.34                  0.49
CAGR                 41.07%                7.86%
Max Drawdown         -41.10%               -55.25%

MTD                  3.93%                 3.38%
3m                   -5.27%                6.69%
6m                   2.87%                 10.17%
YTD                  38.47%                13.50%
1Y                   35.59%                15.74%
3Y (ann.)            27.83%                9.99%
5Y (ann.)            30.69%                15.28%
10Y (ann.)           32.64%                7.06%
Since Incep. (ann.)  41.07%                7.86%

Daily Sharpe         1.34                  0.49
Daily Mean (ann.)    38.65%                9.41%
Daily Vol (ann.)     28.94%                19.18%
Daily Skew           -0.38                 -0.09
Daily Kurt           2.70                  11.76
Best Day             11.09%                11.58%
Worst Day            -11.46%               -9.03%

Monthly Sharpe       1.31                  0.63
Monthly Mean (ann.)  39.34%                8.76%
Monthly Vol (ann.)   29.97%                13.95%
Monthly Skew         -0.34                 -0.78
Monthly Kurt         2.54                  2.26
Best Month           37.70%                10.93%
Worst Month          -28.46%               -16.79%






Allocation over time between XIV and TMF:



Posted on

Bond Investment and Rising Interest Rates


Some time ago I took the annual coupon of the US 10 year Treasury Bond going back to 1927 investigated the effect on a bond index of rising interest rates.  I created a total return index based on constant maturity which is about as good as you can get given the lack of public figures (if they even exist) on bond funds going back that far.

Much alarm has been raised in financial circles and the blogosphere on how bond investors will fare in a rising rate environment, with particular and (largely unjustified?) concern surrounding risk parity funds such as Bridgewater and AQR.

I have no idea how or the extent to which they gear up their bond investments so that the volatility on bonds equals that on stocks. If they use the long bond, I assume they use little to no gearing. I do not imagine they would be able to gain much by gearing up fed funds or the Eurodollar – you can only gear a coupon by borrowing cheap to invest in a lesser credit or at a longer maturity if the interest rate curve slopes upwards.

Nonetheless an index of the 10 year is instructive in showing that widespread panic and gloom is probably unnecessary, unless we were to face a rapid and very steep increase in rates. A swift 1% hike in rates for the ten year would make for a painful 8% drawdown, which would take a while to recover from.

Interest rates rose throughout the period 1949 to 1981. You can see that the cumulative drawdown in price amounted to around 60% in the period. Very painful you might think. But this is to ignore the fact that fund managers would be re-investing as they went along into higher coupon bonds which helps to make up for the price action.

You can see from the drawdown chart of the total return series that drawdown during the period of rising rates was not significant.

Yes, return was lower during the earlier period – CAGR of around 5% as opposed for the later period of close to 9%.  But the period 1949 to 1981 was clearly no disaster.

If there is one thing I have learnt over the years, it is that forecasting is close to useless. The future may not resemble the past. Rates may rise rapidly and steeply. But perhaps looking at history gives some comfort that bond investing over the next decade may not be too painful after all?

Here is my spreadsheet for downloading.




Posted on

Cliff Asness is correct: Leverage Aversion and Risk Parity

Many have complained that gearing into bonds in a rising interest rate environment will lead to disaster. Not so says Cliff Asness and I wholeheartedly agree with him.

Can Risk Parity Outperform If Yields Rise?

I have spent many an hour looking at market data going back for centuries. To 1650 in the case of the UK (courtesy of the Bank of England).  Without going into boring details my endless research leads me to similar views. The vast proportion of bond return is from coupon not price. Provided the rise in interest rates is not too violent, funds can re-invest at higher coupons as rates rise which in time will overcome short term price loss. Calculate a long term Constant Maturity Bond Index yourself (with data from the Fed or the Bank of England) and you will come to the same conclusion.

For what its worth you will find links below to spreadsheets taken from Robert Schiller and the Bank of England containing data for bonds and stocks going back centuries. I have added constant maturity bond indices (which you can chart) which show the relative UNIMPORTANCE over time of price. Coupon is all.  You will, I hope excuse any idiocies or mistakes I have made. I have not looked at these spreadsheets for a while.

Schiller Spreadsheet

Bank of England Spreadsheet


I set out below a simple back test of a very simple risk parity 2 instrument portfolio: S&P 500 TR and 2.5x leverage on the 10 Yr US Treasury bond calculated from futures prices (understated since no interest on capital is included). Yes, AQR and Bridgewater are far more sophisticated and of course include commodities and well as different stock markets and bonds (maturities, currencies and so forth).  The lookback to calculate volatility used was 3 months.

Perhaps the information below might be of help to those looking at the concept for the first time.  Similar volatility, far higher R Sq of equity curve and lower drawdown for the bond/equity/ risk parity scheme.  Will it continue this way? Who knows but it seems a reasonable way to invest.

Stat                 Inverse Volatility    S&P 500 TR
-------------------  --------------------  -----------
Start                1988-04-04            1988-04-04
End                  2016-12-20            2016-12-20
Risk-free rate       0.00%                 0.00%

Total Return         1677.08%              1578.16%
Daily Sharpe         1.02                  0.65
CAGR                 10.54%                10.32%
Max Drawdown         -20.34%               -55.25%
Monthly Mean (ann.)  10.58%                10.83%



Portfolio Weightings


Posted on

Avoid Market Drawdown!

The TMV ETF should stay in place for quite some long time, and it´s a great investment to harvest time decay and avoid drawdown. The big tapering drawdowns of 2013 are past history. You don’t need to look daily at the TMV short hedge. Just keep it. The ETF TMV is a loser and if you stay short it will be a long term winner. It should return about 10-15% per year.

I came across the following yesterday on Logical Invests website:

Avoid Market Drawdown

I’m afraid there has been an error here.  The real figure for the example quoted in a CAGR of around 8.4% not 10 to 15%.

This is because Frank has not taken into account the necessity of maintaining  a constant ratio of short x 1 the instrument concerned. If all you do is to short on day 1 in 2009 and simply maintain that same short until today the profit will simply be the decline in the adjusted short price of $285 to the current price of $18. If you want to obtain the figures quoted you will need to look at your position each day and make adjustments each day to maintain a constant 1x short position.  Then indeed you can achieve the sort of figures quoted.

But for my money the effort is not worthwhile – you can achieve much the same simply remaining long. And then you really can just buy and hold. No daily adjustments necessary, no borrowing costs, no effort.

I like the Logical Invest website very much and they are doing a good job in keeping folk out of the hands of discretionary fund managers so I hope they will not mind me pointing out their mistake!

I don’t like using short histories so instead of using ETFs I used the US 20 Year bond futures. I did not include interest on money not needed for margin, nevertheless the figures are accurate enough to illustrate my point. Have a look at the figures and charts below.


dd us20Yr


Here are the figures for the LONG side:

1 x US 20 Year Bond
1985-01-02 -26.51444365
2017-06-20 156.40625
CAGR 5.62%
Std Dev 10.27%
Max DD -21.56%
2 x US 20 Year Bond
1985-01-02 -26.51444365
2017-06-20 655.1533969
CAGR 10.38%
Std Dev 20.53%
Max DD -40.30%
3 x US 20 Year Bond
1985-01-11 -26.83452635
2017-06-20 1947.883671
CAGR 14.11%
Std Dev 30.81%
Max DD -56.15%


Here are the figures if you short the 3 x Long on day one and simply hold:

1985-01-02 -26.5144
2017-06-20 53.0288
CAGR 2.16%


If instead you had shorted and maintained a short ratio of 1 x the 3 x Long you would have achieved around the same as the long position: 14%  ….give or take…..


Posted on

Why I Loathe Market Timing

I have come to believe that all systems which attempt to choose assets based on past price performance are market timing systems.

As against that I still believe a small modicum of “momentum” chasing is acceptable especially when combined with risk parity weighting and a large portfolio equally balanced between high and low risk assets.   See: Trend Following Without Terror

On top of that I run scared from any re-balancing system whose results vary too greatly depending on which day of the month one re-allocates.  And any sort of on/off binary choice of asset is a dangerous proposition where the re-allocation dates can affect performance dramatically.

Some time ago I did extensive research and back testing on a monthly re-balancing system based on investing in the top “x” US stocks, being careful to include randomly chosen de-listed stocks in the mix.

I read with interest AQR’s momentum prospectus and noted the small pick up in returns over a conventional approach.  Cliff Asness has it right: he rotates into the top “x” stocks out of 1000 and “x” is a healthy 300.  Fine, drawdown and volatility will likely be no better than the Russell 1,000 or whatever but profit may well be a little increased.  And it will be resilient to changes in the rebalancing dates.

All of this lead me to realise that momentum is a fine thing but that concentrated on too small a figure for “x” it is an unstable disaster. Investing in the top few hundred is fine and reasonably resistant to changing the dates of the month on which a monthly reallocation occurs. A small number for “x” such as 10 or 50 is an unstable disaster and leads to wide and wild swings in performance metrics depending on which day of the month one picks.

So any re-allocation method which chooses on or off for a small number of stocks does NOT get my vote.  While in back testing it may look good to “go all on or all off stocks / bonds” using TLT and SPY as representative classes, in the long run it will slap you round the back of the head with a lead cosh.

Great danger lies in trying to time the market with a limited number of instruments in an attempt to increase profits and reduce volatility and drawdown. It looks tempting in back testing but will almost certainly leave a sour taste in the mouth in the long term.

These views have been reinforced over the past couple of days by my re-creation and back testing of the Logical Invest Universal Investment Strategy

Their approach is a good attempt to avoid binary “risk on / risk off” systems:

The idea for this Universal Investment Strategy was to develop a strategy which has an adaptive allocation between 0% and 100% for each ETF (TLT / SPY) depending on the market situation.

Unfortunately it still leaves many periods where risk is entirely “on or off”. If you can count investment in a 20 year bond as “risk off”!

Worse, testing different days of the month re-balancing reveals some stress which can add to the instability of a system which is at least in part binary.

A system such as risk parity avoids this problem and is highly resilient and impervious to “day of the month” problems since it is never wholly “in” or wholly “out”. Prediction of “volatility” is a great deal more stable than prediction of “return”.

Here is what I tested. Different “end of weeks” on which to reallocate (13 rolls in a year):

self.week_list1 = 1, 5, 9, 13, 17, 21, 25, 29, 33, 37, 41, 45, 49
self.week_list2 = 2, 6, 10, 14, 18, 22, 26, 30, 34, 38, 42, 46, 50
self.week_list3 = 3, 7, 11, 15, 19, 23, 27, 31, 35, 39, 43, 47, 51
self.week_list4 = 4, 8, 12, 16, 20, 24, 28, 32, 36, 40, 44, 48, 52

And here are the performance metrics for the Logical Invest UIS and a simple inverse volatility weighting strategy using Spy/TLT since 2002 to end 2016 (using a three month lookback period to calculate volatility):

Settings CAGR Daily Sharpe Max DD
Logical Invest UIS
1 8.53 0.87 21.43
2 9.51 0.93 20.63
3 10.34 0.98 32.17
4 11 1.08 16.55
Std Dev 1.067785871 0.088881944 6.668360618
Risk Parity
1 8.8 1.08 17.89
2 8.69 1.07 18.08
3 8.87 1.09 17.89
4 8.77 1.08 16.74
Std Dev 0.074554231 0.008164966 0.613242747


Good effort Logical Invest and certainly an improvement on “Dual Momentum” but for my choice I will stick to risk parity.