Trading in a low vol world

I wanted to take a look at what works in low vol environments, such as we are currently experiencing. I am open to the idea we have entered a period of structurally low volatility due to increased regulatory burden and flow on effects from the decline of institutional FICC trading. Or it may just be a function of QE, and post-tapering we will see a return to higher levels.

The plan

The main idea is to compare mean reversion (MR) vs. follow through (FT). For simplicity I define mean reversion as an up day being followed by a down day, and a down day being followed by an up day. Conversely, follow through sees an up day followed by another up day, and a down day followed by a down day.

I took a look at the major US equity indices, SPX (GSPC), NDX and RUT.

For each series we calculate daily log returns for the current period and shift the forward to get the return for the next period. Then we calculate realized volatility (RV) and split the data set into "low volatility" and "high volatility", by looking at median realized vol for the whole series.

Then, for each series, we use bootstrapped samples to simulate a number of trajectories/equity curves for each strategy (MR/FT) under the two classes of RV. Finally we take an average of the total return of each trajectory to get a ballpark idea of how they went.

Results

The data is from the start of 1999 to the present, so roughly 15 years. Each run generates 1000 trajectories with a sample size of roughly 950.

For the low vol case, the results are unfortunately ambiguous. Follow through in a low vol environment seemed to do well for NDX and RUT, but the opposite was the case for SPX.

The TR column is the sum of the series over the whole period for the volatility class (i.e. a simple long only strategy), giving an idea of a directional bias that may be present in the sampling.

In the high vol environment, mean reversion was a clear winner, and consistent over the different underlyings.

The results seem relatively stable across trajectory size/sample size.

Outro

I'm not really sure what is going on SPX. My intuition was that FT would do well in low vol environments, but that doesn't seem to be the case, at least not for SPX.

I was actually getting consistent votes for FT in the low vol case, then restarted R to run with a clean environment and started getting the above instead. You can't spell argh without R it seems.

Source is up here. As always you can find me on twitter here. Thanks for stopping by.

The case for data snooping

When we are backtesting automated trading systems, accidental data snooping or look forward errors are an easy mistake to make. The nature of the error in this context is making our predictions using the data we are trying to predict. Typically, it comes from a mistake with our calculations of time offsets somewhere.

However, it can be a useful tool. If we give our system perfect forward knowledge:

1) We establish an upper bound for performance.

2) We can get a quick read if something is worth pursuing further, and

3) It can help highlight other coding errors.

The first two are pretty closely related. If our wonderful model is built using the values it is trying to predict, and still performs no better than random guessing, it’s probably not worth the effort trying to salvage it.

The flip side is when it performs well, that will be as good as it will ever get.

There are two main ways it can help identifying errors. Firstly, if our subsequent testing on non-snooped data provides comparable performance, we probably have another look ahead bug lurking somewhere.

Secondly, things like having amazing accuracy yet still performing poorly is another sign of a bug lurking somewhere.

Example

I wanted to compare SVM models when trained with actual prices vs a series of log returns, using the rolling model code I put up earlier. As a baseline, I also added in a 200 day simple moving average model.

(S) Indicates snooped data

A few things strike me about this.

For the SMA system, peeking ahead by a day only provides a small increase in accuracy. Given the longer-term nature of the 200 day SMA this is probably to be expected.

For the SVM trained systems, the results are somewhat contradictory.

For the look forward models, training on price data had much lower accuracy than the log returns, and the log return model performed much better. Note that both could have achieved 100% accuracy by predicting its first column of training data.

However, when not snooping, the models trained on closing prices did much better than those trained on returns. I’m not 100% sure there isn’t still some bug lurking somewhere, but hey if the code was off it would’ve shown up in the forward tested results no?

Feel free to take a look and have a play around with the code, which is up here.

A volatility filter using historical vol

We have been looking at a way to improve risk adjusted returns by using a volatility filter. Although we could use VIX or equivalent, it turns out that historical volatility will work just as well, if not a little better.

You can see part 1 here Digging into the VIX, and part 2 here What can we use VIX for?

Although the mean return of how we slice things is zero, the distribution of returns is wider for higher readings of our relative measure of volatility. High volatility begets high volatility, at least for our purposes.

By staying out during periods of higher relative volatility, we aim to reduce drawdowns and the volatility of our returns, leading to better risk adjusted results.

A plus of using HV over some external measure like VIX is that it is readily available for any underlying. This means such a filtering technique can be applied to whatever it is we are trading.

Performance

Below is a table with two comparisons, the first compares the HV filter to buy and hold.  Although performance is generally better, we still get some pretty big drawdowns.

The second adds in a 200 period moving average, which is a reasonably strong way of protecting against downside. Again we can see lower volatility and smaller drawdowns with the addition of a vol filter.

I used a 3 month/63 day look back for our relative volatility measure. I haven’t really dug in to what happens when volatility remains elevated for extended periods of time.

I also did not experiment much with the threshold for where we draw the line on ‘high’ relative volatility. I use 0.6 as the cutoff because I originally split things into quintiles when making the first charts.

I also ran this for RUT and NDX over the same period

These results are all frictionless, don’t factor in dividends, return on cash, etc, etc. I don’t consider this viable as a standalone indicator, but something that can be used along side other factors like rotational strategies, or as a potential tool if you are looking for lower volatility.

The source is up here, feel free to have a play around with it and see how you go.

Thanks for reading, 'till next time.

What can we use the VIX for?

In part 1, we took a look at VIX and the relationship it had between historical volatility and realized volatility.

Continuing on, I thought I would take a look at next day returns and the VIX. There is a relationship between SPX and VIX in that when SPX drops, VIX typically rises. This leads the question: does a high VIX serve as a useful indicator of what might happen to SPX the next day?

A problem in answering this is quantifying what we mean by a "high" VIX. A VIX of 25 might be high if it has been under 20 for the last 3 months, but might be considered low if it has been around 30.

To deal with this I use a proportional measure based on its previous trading range. If it is at the top of its range, it gets a 1, and at the bottom it gets a 0. Anywhere in between it gets a number between 0 and 1.

I used data from 2000-2013 and a 252 period lookback, which is roughly 1 year, recording the next day return of SPX. I split this data into 5 groups, the first with relative VIX readings up to 0.2, the second between 0.2 and 0.4, and so on up to 1.0. Then I plotted the next day returns as a boxplot, giving the distribution of returns for a VIX reading in its relative quintile.

The box on the far left corresponds to the lowest relative VIX levels, while the box on the far right corresponds to the highest.

One thing that jumps out to me is that the mean of all these returns is more or less zero. Using this relative measure of VIX has no value as a predictor of next day returns, as whatever measure we use, a mean return of zero will see us end up flat (or in practice, probably down a bit). If there was a real edge, we would expect the mean return to be somewhere other than zero.

The second thing that I notice is the distribution of returns is a lot wider for the right most two boxes, which corresponds to relative VIX measure above 0.60, i.e. relatively higher volatility. Now this makes sense, as we would reasonably expect periods of higher volatility to have higher volatility.

It seems that high readings of VIX can serve as a useful predictor of higher periods of next day volatility, and for someone trying to implement a volatility filter, that could be quite useful.

Conclusion

To recap, we have seen that:

1) VIX is not very good at forecasting 30 day realized volatility.
2) Historical volatility can be a reasonable proxy for IV/VIX.
3) Higher periods of volatility do portend continued higher volatility, at least for the short term (1 day).

Note that 3 does contradict the mean reversion of volatility, because it is looking at the next day returns only. Over a longer period we would certainly expect it mean revert, otherwise we'd be experiencing some very, very bumpy rides ...

Also, I did similar analysis using RUT & RVX and NDX & VXN, both of which had very similar results. We will take a look at them in the next post where we use the findings from this post and the previous to implement a volatility filter.

Code for this post and the previous is up: here

Another cut at market randomness

I have some background in computer security and one day found myself tasked with assessing the quality of randomness for session id tokens generated by popular web frameworks (namely Java and .NET). As it turns out, NIST have developed a series of tests for just this purpose detailed here.

As a non-believer in the absolute randomness of markets, I thought I might take a look at the series of returns from SPX/GSPC to see how they held up.

About the tests

The NIST tests are designed to be run on bit streams from a random number generator, so some liberties had to be taken. I am mainly interested in the predictability of up days vs down days, so I first generated a series of returns using quantmod then encoded them using a 1 for an up day and -1 for a down day and run the tests on the encoded stream. The NIST tests use -1 in place of a 0.

The main test is called the monobit test, to quote from the document:

The focus of the test is the proportion of zeroes and ones for the entire sequence.  The purpose of this test is to determine whether the number of ones and zeros in a sequence are approximately the same as would be expected for a truly random sequence.  The test assesses the closeness of the fraction of ones to ½, that is, the number of ones and zeroes in a sequence should be about the same.  

The tests are done using a significance level of 0.01, so come with a good degree in confidence assuming the underlying method is sound. 

One caveat is the length of the runs compared and how it relates to the distributions used to model the results. For the monobit test, the suggested input size is 100, requiring 101 days of data to determine 100 up or down days. If we were looking at 32 bit integers, 100 bits would only be 3 "full" random numbers, so arguably we would want to look at shorter time periods (e.g. 3-5 days of data). Given the difficulties around distributions which require a large n, I thought I would vary the significance level instead of a lower n, as our requirements are not as stringent as those for cryptographic random numbers.   

Results

At a basic level, this series does appear to be random, at least the vast majority of the time with n = 100 and alpha = 0.01. My confirmation bias was very upset with this. 

However, if we plot the proportion of runs deemed random vs the significance level, we see the proportion rising as one might expect. One thing that remains unexplained is why this appears to rise in steps rather than something more linear, though I expect this to be a side affect of either methodology or the normalisation done by the tests. I also took a look at the weekly data, which tends to a greater proportion of non-random runs quicker than daily data.

I am interested in the applications of machine learning to financial markets. Close to close returns that we have been looking at here are not the only information we have available, nor are they what I trade on a personal level. Also this is only one price series, and one could argue that in practise it is not actually a tradable series. 

Close to close returns are very useful in lots of applications, but if we are trying to build some predictive model we might need to look for more predictable pastures. Machine learning algorithms are great, but can only do so much. Finding some better potential inputs is what I will take a look at next.

The test code is available on github here: R Monobit test. Would be very interested to hear if anyone else takes a look.

I also took a visual and binomial look at randomness of the series in this post: A visual look at market randomness.

Oh and in case you were wondering, the web session tokens all turned out to be very strong.

A visual look at market randomness

I recently did some statistical testing to see if markets were random (details in the post Another cut at market randomness). It turns out they were, at least close to close returns for SPX/GSPC. My confirmation bias wasn't going to stand for that, so I thought about taking a different look.

Two things interested me. Firstly, I am looking at up vs down (i.e a higher or lower close than previous), rather than trying to predict an exact price. If markets are random then up or down have a probability of 0.5 each and are independent. A run of 5 consecutive ups or downs has a probability of 0.03125 or roughly 3%. How would that pan out looking at historical data?

Secondly, how could one visualise seemingly random data without it ending up looking like noise?

I came up with the following chart:

Each square represents one week and each line represents one year. If the close was higher than the previous week, it is blue, otherwise it is red. As the count of successive higher or lower weeks rise, the boxes get deeper in colour, up to a maximum of 5. As a side effect of date calculations and the definition of "week" some years have 53 weeks, which is why some lines are longer than others.

In total there were 138 runs of 5 weeks in the same direction out of 2208 samples, or around 6%, roughly double what we might expect.

Looking at that, I wondered what it would look like comparing weeks across years, comparing week 1 of year n with week 1 of year n + 1. That lead to the second chart:

This time we had 175 runs of length 5 out of 2208, just under 8%, again quite a bit more than the 3% we were expecting.  

That is all well and good, but these charts only represent the direction of the week to week moves, not the magnitude of the moves which is probably more important. Finally I took a look at the return over 5 periods. 

Again if it is positive the squares are blue, negative they are red. The colours are scaled as a proportion of the largest positive and negative returns for blue and red squares respectively. The very pale squares are where the returns were proportionally so close to zero they would not otherwise be visible, so I set a minimum level to ensure they displayed.

We can see that positive returns tend to follow positive returns and vice versa, at least for this 5 week look back period. This is somewhat deceptive as a negative return, though negative, may still be higher than the previous one implying a loss. 

What does all this mean? Not too much in practise, as it is another thing to know in advance if a series will have consecutive up days or down days. In this case a tradable edge is not so easily won.

However, it does reflect my understanding of how prices move a little better, in that they trend for a while then range for a while and vice versa, and things may not be as random as we might expect. My confirmation bias somewhat sated. 

The charts were done in Processing using the free weekly data from Yahoo! finance for GSPC. If you would like a chart for a given ticker, let me know.