VXX trading based on fundamentals

In a previous article we have presented a simple option strategy that can be put in place to capture the expected erosion of contracts like VXX or VIXY.

Just as a recap, in that article we have said that is wrong to believe that the daily rolling of VIX futures during contango, consisting in selling the front contract at lower price and buying the second month at higher price, is the reason for the price erosion that we observe in VXX (or VIXY). Instead, the erosion is coming from the contango itself, not the daily rolling. After all, by holding a unit of VXX/VIXY we are investing in a portfolio of long VIX futures and in contango these futures are “rolling down the curve” to settle at the spot VIX level on expiry date. So contango is the source of price erosion for these VIX-related instruments.

We also said that harvesting the VXX erosion by simply shorting the VXX is not a safe game. A safer way to capture VXX erosion due to contango can be obtained by using VXX options.

The strategy we presented was a simple unbalanced (or broken wings) condor with a max profit zone located where we expect the underlying to be by the time the options expire, with no risk on the upside and a defined risk on the downside.

In this article we want to look at this in more detail to get an idea of where the max profit zone needs to be set and how likely it is for our max risk to be hit by the VXX movement.

The strategy

Let’s start by showing the strategy on August’s monthly expiry. With almost 75 days till expiry, we have created a max profit zone that, at current levels, would cover a VXX movement from -3% to -29%. In other words, if VXX will decline between 3% to 29% the strategy will make the highest profit (in this case $123). The lower breakeven point is distant 30% and we will have the largest loss (in this case $277) if VXX declines more than 34% by August 20. As you can see, the strategy is fully hedged against spikes in VIX futures and will actually make some money even if VXX will rise.

The plots below are showing the strategy P&L profile at different point in time (i.e. today, in 45 days and in 60 days).

Needless to say that with options there are many possible alternative combinations and one could prefer to have no profit or even a small loss on the upside, in return for a higher max profit zone and/or lower max risk. An example below where we added a small put debit spread to the previous strategy to lower our max risk and change the max profit zone (sacrificing profit potential on the upside).

But let’s go with our first choice. Focusing on the red curve, we can see that delta is positive so an immediate rise in VIX futures would be beneficial. As time passes, and the VXX starts to erode, the startegy should have a positive P&L (depending on how low VXX has moved) and could be closed before expiry if needed. The only real risk is a sudden and drastic drop in VXX.

To understand if we can be comfortable with this profile it is worth looking at what has been the behaviour of VXX/VIXY over the following 75 days in the past.

Just looking at the expected erosion might give us an under estimation of the movement we can observe in the underlying. We can expect the underlying to decline at a faster rate due to many reasons. One could be the VIX level itself. While it is true that trading VXX/VIXY we are trading the expectation of where the VIX will settle on the expiry date of the contract (i.e. not the VIX index itself), we can see if the VXX/VIXY 75 days forward returns are somehow affected by where the VIX index currently is.

The plot below shows the median return in dark grey, while the shaded area corresponds to the range between the 25th and 75th percentile.

So, given that currently VIX is between 15 and 20 we can say that historically 75% of the time the return over the following 75 days has been higher than -35%, and 50% of the time it has been higher than -25%. Our max profit zone is currently set to include most of this range.

Before concluding that we are comfortable with this payoff, we can also look at a lower quantile. The plot below tells us that at current VIX levels (between 15 and 20), we observed a 75 days forward return higher than -41% for 90% of the time.

And our max risk area that is distant 34% would have not been reached 80% of the time, as per plot below.

We can also look at a different dimension. The ratio VIX/VXV gives us some information about the slope of the VIX term structure. In fact, while the VIX index represents the current market expectation about the volatility of the S&P 500 index over the next 30 days, VXV is a measure of expected volatility over 3 months.

Currently this ratio is below 1 and that is the usual configuration that we see most of the time (i.e. contango).

The median 75 days return when VIX/VXV is below 1 is -25% and 75% of the time the return is above -35%. The plot below shows that in contango the 75 days return is higher than -41% on 90% of cases.

Returns before expiry

So far we looked at the 75 days return given that the strategy expires in that timeframe. But it is also interesting to look at returns on intermediate timeframes to see how likely is for us to close the strategy at a profit before expiry.

In the charts showing the strategy (see previous section), we also display the P&L profile that we can expect in 45 and 65 days. So let’s look at the returns over these periods.

In 45 days we can expect the strategy to be in profit if the underlying has dropped less than 18%. Please note, we ignore here the impact of implied volatility but since vega is negative, any decrease in implied volatility would be beneficial, while sudden increases would hurt the P&L. Typically, VXX implied volatilities increase when VXX increases so we can expect that as VXX declines our negative vega could potentially add to the profits, although at a lower rate given our negative vega will also reduce at lower VXX levels.

From the chart below, we can see that if the strategy is initiated when VIX is lower than 20 it is more likely to see the strategy in profit already after 45 days. If the strategy is opened with VIX at higher levels, the median 45 days return is lower than -18% and the expected P&L would typically be negative (with the only exception of the range 30-55).

In 60 days we can expect the strategy to be in profit if the underlying has dropped less than 26%, again ignoring the impact of implied volatility.

Even in this case, if the strategy is initiated when VIX is lower than 20 it is more likely to see the strategy in profit before expiry. If the strategy is opened with VIX at higher levels, the median 60 days return is lower than -26% and the expected P&L would be negative.

The trading plan

The rationale of this strategy is based on the fundamentals of the VIX futures markets and their impact on the price of VXX/VIXY. A trader could start opening a strategy expiring in 2/3 months and then repeat that (likely on different strikes) for any subsequent month, so that in 2/3 months there will be a strategy expiring every month.

A way to potentially improve the payoff (i.e. reducing the max risk and/or enlarging the profit zone) is to build the strategy in more steps.

The simplest way to do so is by looking at the unbalanced condor as the combination of one bullish credit spread and one bearish debit spread. Indeed, the strategy is nothing more than a put bear spread funded by a 2x put credit spread.

Since Trading Mate actively trades SVXY/VXX (see all results here), one could use this system to decide which spread to open first. Please remember that while SVXY is our trading vehicle, similar results could be achieved by using VXX in the opposite direction, but remembering that SVXY is actually now trading at 0.5X leverage. So, results below show that the system could be used not only for directional trading in VXX or SVXY but also to assist in the building of our option strategy.

If Trading Mate’s signal on SVXY is short (hence long on VXX), one could start opening the bullish credit spread and wait for a flat or short VXX signal to add the bearish debit spread to complete the strategy.

By actively timing the entry, one could significantly improve the risk/reward of the strategy. Once the strategy is completed, one simply waits for the VXX erosion to run its course and hopefully reach the profit zone.

Conclusions

The startegy we have discussed has the potential to benefit from the VXX price erosion due to contango without exposing the trader to significant risks that, in the field of volatility, are typically on the upside.

This strategy can be structured to have a max profit that is at least 40% of the max risk, with a profit zone that covers a significant drop in the underlying.

Historical data suggests that the choice of the strikes (that will ultimately determine the profit zone) depends to some extent on the VIX level. With VIX lower than 20 it is likely that the strategy can be closed at a profit even before expiry.

Luckily VIX is lower than 20 at least 75% of the time so the best conditions to open this stategy can be met quite frequently.

While all these conditions can help setting up a strategy with higher chances of profit, opening the strategy in two moves by following Trading Mate’s signals could significantly improve the P&L profile.

Here you have it: a quantitative trading strategy to help you setting up a strategy to profit from the fundamental dynamics of VIX future markets.

Access the updated portfolio here

Understanding Leveraged ETFs

In this article we are discussing the mechanics of leveraged ETFs and try to understand if they are a suitable trading vehicle to replicate Trading Mate’s exposures.

Access the updated portfolio here

Leveraged ETFs are products that are designed to replicate X numbers of times the daily performance of an index, commodity, currency, etc.

An example of such product is the TQQQ which tries to replicate 3X the daily performance of the NASDAQ. This means that if the NASDAQ moves up 1% on a given day, the TQQQ should move up 3% on that day.

The way the fund achieves a leveraged exposure is typically via derivatives contracts like equity swaps.

While these products seem simple, there are few things that investors must take into account when deciding to use them in their portfolios. We will not focus here on aspects like the higher fees that leveraged ETF have if compared to their unleveraged counterparties, but we will mostly focus on one key aspect that goes often under the name of volatility drag.

This is related to the compounding effect and will result in the leveraged ETF’s returns to differ from the multiple of the market return they attempt to replicate, on a daily basis, even if they actually managed to offer a perfect daily tracking of the market returns. So, for example, even if a 3X product has always replicated 3X the market returns every day in the past year, it is possible that its yearly return will not be 3X the yearly return of the market.

To understand why this is happening we can look at the calculations below and use a 3X product as example:

From these calculations it is clear that the component highlighted in orange is the driver of the differences we observe over the long term between a leveraged ETF performance and the multiple of the market performance we expected to see.

It is also clear that this component will always be negative as long as the returns keep changing sign. This is what you would observe during a sideways market and it helps to explain why even if the market hasn’t moved much, your leveraged ETF exposure might have lost money.

On the other hand, this component will greatly help the leveraged ETF’s return during a trending market, i.e. when returns tend to have the same sign, since that component will add to the multiple of the market return we expected.

Daily rebalancing

The objective to replicate X times the daily return of the market requires the fund to rebalance its exposure to the benchmark every day by adding or removing exposure depending on the market movement. An example below, where we can see that, after a 5% gain in the benchmark on day 1, a 3X fund would need to increase its exposure by $30 in order to make sure that at the start of day 2 the exposure is still 3X, as desired.

Leveraged ETF vs Futures

The example below, will compare the performance of a 3X ETF vs a future contract to see how the daily rebalance to keep the leverage ratio fixed in the ETF will generate a difference in performance that, each day, is equal to 6 times the product of the 2 most recent consecutive returns (as seen in the table above).

Let’s assume we have 1 micro-future on NASDAQ. At a price of 14,000, this contract would have a notional exposure of $28,000. If we want to have the same initial notional exposure via a 3X ETF, we can buy $9,333 of this product.

As the market moves, each product will generate a P&L and while they are the same on the first day, they will start to differ over time. The amount (and sign) of the difference will depend on the sequence of positive/negative returns we see over time.

The examples below shows how a trending sequence of return will generate higher returns for the 3X ETF, while a sideways market will penalise the 3X ETF vs the future.

As we saw earlier, the fact that consecutive returns have the same sign (i.e. the market is trending in one direction) help the 3X ETF to out-perform the future. The opposite is true when markets frequently flip sign.

The role of volatility and strategy performance

In this context, volatile markets are directionless and experiencing frequent swings that keep eroding the performance of the instrument.

We can expect that a strategy like Trading Mate that tries to increase exposure to the markets only when a directional move is expected, will not suffer much from volatility drags.

Let’s see below how Trading Mate’s strategy would have performed by using the following 3X ETFs: UPRO, TQQQ, UDOW and TNA.

In the charts below the “market” (red line) is a buy & hold investment in that specific 3X ETF, while the strategy returns are shown in blue. 

While in all cases, the risk-adjusted performance of the strategy is higher than a simple buy & hold of the instrument, it is also interesting to note how TQQQ and TNA’s different behaviour has not impacted much on our strategy performance: in both case we have a 50%+ annualised performance, less than half the VaR and Expected Shortfall and a lower drawdown profile compared to a buy & hold.

A focus on 2020/21

In a long-term chart it is never easy to clearly see the 2020 performance so this section will show the performance, risk metrics and drawdowns since Jan 2020 by using 3X ETFs to follow Trading Mate’s strategy.

Again, in the charts below the “market” is a buy & hold investment in that specific 3X ETF.

If interested in current Trading Mate’s exposures, please click on the link below.

Access the updated portfolio here

Notional exposure is all that matters

As you can see in the results page, Trading Mate’s performances are shown without any leverage since they are based on the actual notional exposure of the portfolio.

This is, in our opinion, the best way to evaluate a strategy and we should consider this aspect when comparing different strategies in order to make sure we are comparing apples with apples.

Let’s use, as an example, our strategy on S&P 500. We show the exposures based on SPY but the same could be replicated with any other instrument that is tracking the S&P 500 index. Even if SPY is assumed to be the vehicle that Trading Mate uses to follow this strategy, in our actual trading we use futures/CFDs or options instead. This is mainly because SPY is not available for trading in Europe.

As you know, in order to trade futures/CFDs only a small margin is required by your broker to open and maintain the position. Let’s say that the S&P 500 micro future is trading at 4,170 points. Given that each point is worth $5, one micro future is effectively giving you a notional exposure of $20,850 (=5*4,170).

If we decide that our maximum allocation on S&P 500 is $100k, when Trading Mate has an exposure of 20% we could buy one micro future since this would give us the desired exposure of $20,000. Please note that futures cannot be traded in fractions so there will always be some rounding to accept. If you have access to CFDs, it might be easier to get the actual exposure you need. But for some futures you will end up rounding the exposure down and for others you will need to round up. Overall you might find that it won’t really make a major difference.

In the table below we assume that we are interested in trading these 3 US indices by using micro futures. For example, a micro future on NASDAQ at current prices would give a notional exposure of about $28,000 so if we have set our max allocation to this market to $100,000 we might need to buy/sell one micro future for each 25% of change in Trading Mate’s exposure.

With a margin of 5%, we could get our max allocation across all markets with less than 16k. This implies that each 1% of movement in the underlying will generate a 20% movement in the amount of capital we posted as margins.

Here is the tempting part. If we had 300k as initial trading capital and decided to allocate 100k to each market, that means that after posting the margins we now have 284k unused. Clearly we will be tempted to add more futures potentially up to a point where our notional exposure becomes simply too large to be acceptable.

As a rule of thumb, we can multiply our margins by 20 to have an idea of our actual exposure. That is all that matters since profits and loss will be calculated on that basis. As a consequence, performances obtained trading futures should be divided by 20 to have an idea of the equivalent performance on the actual notional exposure that was taken.

Leverage can be of great help as long as we keep track of how much notional exposure we are actually taking.

For example, if we know that 5 micro futures on S&P 500 will have an exposure of $100,000 and we stick to trading max 5 contracts since our maximum allocation on this market is $100,000, we don’t have to keep the whole $100,000 available on the trading account but maybe leaving just 3 or 4 times the required margins, in this case about $20,000 would be sufficient to cover our activity on S&P500.

How to compare strategies with different level of leverage

We often see strategies using different leverages being compared without considering the actual notional exposure (and hence risk) they take. A strategy that trades futures might have generated a higher P&L in dollar terms compared to a strategy using only unleveraged ETFs over the same period of time. But this does not necessarily imply that the first strategy is better.

We mentioned earlier that we should divide by 20 the performance of a futures portfolio to make it comparable to the performance of a non leveraged portfolio. This is a direct consequence of the assumption of a 5% margin on our future positions. But when dividing by 20 we are also assuming that we are investing our full capital in margins. In reality, a trader is not supposed to have more than 30/50% of the trading capital in margins at any given time otherwise that is a sign he/she is taking too much risks. So, maybe, the best approximation would be to divide the futures performance by a number between 5 and 10 instead, depending on how much capital is actually used as margin.

Let’s make an example. Say that our capital is 100k and we only trade NASDAQ futures (NQ) and S&P 500 futures (ES). If margins are 5%, with just 1 contract for each product we would post almost 25% of our capital in margins, with our actual notional exposure being 490k.

Now let’s say that we make a 100k P&L. This would be a 100% return on our trading capital, but just 20.4% on our actual notional exposure (=100k/490k).

Earlier we said we could divide by 20 the futures performance to get the equivalent unleveraged performance, but dividing the 100% return by 20 we would have only 5%, instead of the 20.4% we calculate above. This is simply because, as we mentioned above, that “divide by 20” shortcut assumes that we are using all our trading capital in margins. But here, as we saw, we are only posting about 25% of capital for margins. So the actual formula would look like:

where we can see that if instead of a 24.5% margins we have 100% posted as margin, we could actually divide by 20 to get our adjusted return.

If you use the adjusted returns, you can compare strategies using different leverages and understand how much alpha is generated by the strategist and how much is simply leverage.

How to use leveraged ETFs

We got some question recently about how to use leveraged ETFs in this context. Please read this article we wrote about leveraged ETFs to better understand how they work and how we can use them in our strategies.

Hopefully by now the message should be clear: if a future that has a 5% margin requirement has an effective leverage of 20x, when you are trading a 2x ETF you simply need to remember that this is like getting double the notional exposure that you bought. So buying $10,000 of a 2x ETF is equivalent to having an exposure of $20,000 to a non-leveraged ETF. At least on the first day (please refer to the article about leveraged ETFs to understand why this is not necessarily true on the following days).

So, in conclusion, always keep track of the effective notional exposure that an instrument offers and make sure that is in line with the desired exposure you want to have. A simple way to do so is looking at the margins and apply the formula above. A higher P&L in dollar terms is not necessarily exceptional if it is only achieved thanks to high leverage and the high risk taken. A strategy using no leverage could have achieved a much lower P&L (in dollar terms) but be far superior.

Risk-based allocation of Trading Mate portfolio

When a portfolio invests on different markets there is always the problem of deciding how much capital to allocate to each given market. Most of the time, this is done based on personal preferences or sometimes based on performance.

We will discuss here a risk-based approach that could also be followed to decide how to allocate your trading capital.

As a reminder, Trading Mate invests in the following markets:

The metrics we will use to decide the allocation is the Value at Risk (VaR). If you go to our results page you will find this value reported in the title of each plot. It represents the loss that you expect will be exceeded on the following day only with a very small probability (in our case 1%). In other words, you are 99% confident that during the next trading session your strategy won’t lose more than the VaR.

In the plot title we also show the Expected Shortfall. This number is very important because it tells you: ok, I’m 99% confident that I won’t lose more than the VaR during the next session, but how much can I lose in that 1% of cases when the loss will be exceeding the VaR? The answer is that your expected loss in those cases is equal to the Expected Shortfall. So, really, VaR and Expected Shortfall are two important risk measures that should always be read together.

In the following we will use only the VaR numbers of each markets to decide our allocations, but the same approach could be followed by using the Expected Shortfall. You will see that with some simplifying assumptions this approach will easily give us a VaR for the whole Trading Mate portfolio. The next step then is to set our own risk tolerance and change the allocations to maintain the portfolio VaR within the risk limits we set to ourselves.

An example

Let’s assume that our trading capital is 100,000. This is the amount of money we are willing to risk in the market in order to have the opportunity to achieve higher returns.

In the example below, we are assuming that the trading capital is equally split across all markets we trade. This can be seen in the “max allocation” column. For example, we will allocate 7.7% of our trading capital, so 7.7k, to our trading on S&P500. In this example, we assume the same allocation is given to all other markets.

The column “VaR” shows the market loss that we expect will be exceeded over the next 1 day only with a 1% probability.

Finally, the column “exposure” shows the suggested exposures that we publish everyday in the portfolio page. But for allocation purposes, let’s assume that we have a 100% exposure on each market.

Given the VaR, the expected loss in the last column will be obtained by multiplying our exposure to a given market by the corresponding VaR. So, for example, our 7.7k exposure on S&P500 would experience a loss higher than $455 only with a 1% probability.

By simply summing all the expected losses we make the assumption that there is a perfect correlation across these markets, at least when things go bad. This is clearly a conservative assumption but non necessarily a stupid one.

In this example, our VaR (in dollar terms) over the next 1 day is $5,983.

The VaR limits

The table above also shows a VaR limit. This is the part where personal risk tolerance will need to be factored in. For example, I might feel comfortable with a loss of 5% that I expect to experience only with a 1% probability. More risk-adverse investors might prefer to have a VaR limit of 3% instead, i.e. risking more than 3% only with a 1% probability.

In the example above we have set the VaR limit to 5% but given the equal allocation across all markets, we see that our portfolio VaR is 5.98%. This tell us that this allocation is too risky for our risk tolerance.

To reduce our VaR we can start re-allocating some of our capital from higher risk markets to lower risk ones.

We can, for example, reduce our max allocation to VIX, XLE and Silver from 7.7% to 3%. This would result in a portfolio VaR of 4.66% (so within our limits).

Having reduced the allocation on these 3 markets, we have now part of the capital that is not allocated. This is displayed above as “buffer” and, in this example, is about 14k.

To stay within the VaR limits of 5%, we could allocate part of this buffer to markets with lower VaR. Below we decide to increase our max allocation to Treasury Bonds and Gold.

Portfolio VaR is now 4,94%, so within the VaR limit of 5%. We would have about 6.4k as cash (buffer).

Market VaR vs Strategy VaR

The goal of Trading Mate is to generate superior risk-adjusted returns so it is natural to expect that risk metrics like VaR and Expected Shortfall will be lower when computed based on our strategy returns rather than market returns.

The example below shows our portfolio VaR assuming, again, an equal allocation of the trading capital but using the VaR of our strategies rather than market’s.

Instead of a loss of 5,983 we now can expect not to lose more than 2,478 with 99% confidence. This is well below our VaR limit of 5% so we have more freedom to increase our exposure to more risky assets, if we want to.

So what VaRs should we use, market or strategy? Well since Trading Mate actively manages the exposures based on these strategies it would make sense to use the associated VaR when deciding how to allocate the capital. Clearly, using market’s VaRs would be a more conservative choice in this case.

To understand why using the strategy VaRs makes more sense it is useful to think about each strategy as a single actively-managed fund investing exclusively in that particular market. So, for example, our strategy on S&P500 can be seen as an active fund investing on S&P500 and dynamically changing exposure between cash and index based on perceived risk in the markets. By using the strategy VaRs we are basically assuming that we are allocating part of our trading capital to this active fund which will then change the exposure day by day as the market moves.

How to use the file

At the link below you can download the file that has been used for these examples. Feel free to use it to see how the allocation would change based on your risk tolerance.

Download the file from here

Please note that not all the markets have to be used (set the allocation to 0% if you want to ignore that market) and that the assumption to have an exposure of 100% is only made to understand what could be the portfolio VaR should we be fully invested in those markets.

In reality, Trading Mate only has full exposure for limited periods of time as long as favorable conditions persist. So, to have a more realistic portfolio VaR on a given day it is better to update the column “exposures” with the actual exposure we have published in the portfolio section.

Access the updated portfolio here

One last point to remember is that VaR measures change everyday. We provide updated figures on a weekly basis in the section where we publish recent results. Please monitor this section and update the “VaR” column should the VaR of a given market have changed significantly.

How do probabilities change based on model exposures – an example on EuroSTOXX 50

When repeating a given experiment over time, we can count the number of times a given event occurred. If for example we toss a coin 100 times and we saw that the event “head” occurred 51 times, we will say that the relative frequency of the “head” event is 51/100 (i.e. we observed “head” for 51 times out of the 100 trials) , or 51%.

Now, let’s bring the number of trials to infinity and the concept of relative frequency can be used to define the probability of that event happening. To understand why this is not valid when the number of trials is low, think about tossing a coin for just 10 times and observing a “head” for 6 times. The relative frequency is 60%, but can you really say that there is a 60% probability associated to a “head” event? The answer is clearly no.

We will use these concepts below by looking at the trading days since January 2008. Given the number of “trials” is not infinity, we will just pretend to interpret the numbers below as probabilities but we want to see if the fact that Trading Mate has a given exposure is affecting the relative frequency (or “probability”) of specific observations over the next n days.

In other words, we compute the “unconditional” frequencies of market moves over the following n days and then start conditioning by Trading Mate’s exposure to see if those frequencies change significantly.

Knowing how many times a given market move has occurred over the next n days can be very useful when deciding what strikes to choose in an option strategy (especially when approaching maturity) but it is also important to know when people want to decide where to put the stop loss.

So let’s look at the numbers for EuroSTOXX 50. Under the column “unconditional” we can see the relative frequencies of a given market move over the following 2 days, irrespective of Trading Mate’s exposure. Please note that when reading the table, the market move should be read as “higher than” (if positive) or “lower than” (if negative). So, for example, the unconditional frequency of a move higher than 3% over the next 2 days is 4.61% (third row in the “unconditional” column). Similarly, the unconditional frequency of a move lower than -3% over the next 2 days is 5.78% (third last row).

What is clear from the table above is that when Trading Mate’s exposure is negative (so the portfolio is net short) the frequencies of negative market moves over the next 2 days are significantly higher than the unconditional ones. For example, when our portfolio is short with an exposure between -50% and -100% at market open on a given day, the relative frequency of a market move lower than 3% is 20.73%. This compares to a 5.78% unconditional frequency. Even when the exposure is less decidedly short (i.e. between 0% and -50%), the relative frequency of a move lower than 3% over the next 2 days is 9.61%, hence higher than the unconditional frequency.

Similarly, looking at the positive returns, we can see that when Trading Mate has a short exposure, the frequency of positive market returns (i.e. losses for our portfolio) are lower than the unconditional. When the portfolio is long, these frequencies are higher than the unconditional, even if the differences are less strong.

It is important to note that we are looking at the next 2 days, assuming that the exposure is not changed. The table below assumes that the position is kept unchanged over the next 5 days.

In general we observe a similar behaviour: the relative frequency of lower returns is higher when Trading Mate has a short position, while the frequency of higher returns is higher when Trading Mate is long.

To summarise, by knowing Trading Mate’s exposure, the relative frequency of a given move over the next 5 days is different than what you observe unconditionally (i.e. ignoring the portfolio exposure).

If you need to choose how distant to set your stop losses or what strikes to sell, it is useful to know how frequently a given market move has been observed over a given number of days. But we just saw that knowing Trading Mate’s exposure you have an additional level of information that gives you a better view of how frequent a given move can be seen so that you can make a better informed decision.

Say, for example, that you have sold a put and there are 5 trading days to expiry. The strike you sold is 4% away from current levels and you want to keep the position unless the chance to observe a market move lower than -4% is higher than 10%. If you read the unconditional data, you see that the relative frequency of a move lower than -4% is only 8.14% so you might be happy to keep your short put till expiry.

But now assume that you also know that Trading Mate is short with an exposure of -70%. The relative frequency of a move lower than -4% conditional on Trading Mate’s exposure being between -50% and -100% is 17.07%. Being higher than your tolerated 10%, you will decide to close the position or at least hedge it until expiry.

Time to become risk-free on this strategy – a follow up

Few weeks ago we opened a bull vertical spread on EuroStoxx 50 in line with our bullish exposure in Trading Mate. In this post we discussed a possible adjustment to make in case we needed to increase our delta exposure following a signal from Trading Mate.

Since our models didn’t suggest to increase the exposure since then, we never had to put in place that adjustment. But we are now approaching option maturity and it is important to decide what to do with the position.

Our delta on this position has become very low as the market is trading inside our maximum profit zone. A clear choice would be to simply close, take profit and move to next month with another strategy with positive delta (given Trading Mate is still long on this market).

We already saw the possible adjustment we wanted to make to increase our delta in case of an increased exposure in Trading Mate.

The plan we had in mind for this position if Trading Mate would have turned short, instead, can be seen in the image below. The adjustment in green would have provided a nice opportunity to profit from the movement until options expire next week in that scenario.

You can also see another adjustment in blue, which was our plan B in case nothing relevant was going to happen and models would still keep a slightly long position. A simple addition of a put bear spread built between the strikes of the original call spread is providing a nice reduction in risk (effectively leaving the position entirely risk-free) without sacrificing much of the maximum profit.

So here we have it, it is just one week till expiry and it makes little sense to make any other adjustment on the upside now since premiums are very low. Rather, this adjustment puts us in a risk-free position where we can now simply forget this strategy until expiry since no harm can be done here. Meanwhile, we will position ourselves on the May expiry following, as usual, Trading Mate for directions.

This is the final adjustment, with prices for the bear spread filled this morning.

Access the updated portfolio here

Trading Mate with options – March 2021

Trading Mate is a portfolio that dynamically adjusts the risk exposure based on a mix of quantitative trading rules, risk management and money management. This portfolio tries to minimise the number of transactions by avoiding intraday trading and increasing the portfolio exposure only on days when it has higher confidence.

Access the updated portfolio here

The system can be used to trade any instrument that gives exposure to a given market. So, for example, when looking at the system on S&P 500, the actual trading can be performed by using futures, CFDs, ETFs (i.e. SPY, etc), options and more.

This article tries to offer an overview of how Trading Mate uses options on SPY to follow the signals and how the suggested exposure can be translated to delta exposure for a given strategy.

Let’s say that our maximum exposure on SPY is 100 units. At current levels, this means that for every 1% move in the underlying we would have a P&L of about $380. One option controls 100 units so if we buy an options with delta 0.5 this will have approximately a P&L of $190 for every 1% move. We will ignore higher order effects here.

The chart below shows the last month’s of trading in SPY together with the exposure that Trading Mate had at the market open of each bar.

This exposures have generated the following P&L profile against the market.

How should we use this information when trading options? There are many ways, from a simple purchase of one single option (put or call depending on the direction of the exposure) with a delta equal to Trading Mate’s exposure, to more complex strategies.

Access the Option Strategist section here

The images below summarise our activity on SPY options (expiry: April 16) during the month of March with a strategy that has changed its delta over time based on the exposure in Trading Mate. Please note, the last few days of rising markets haven’t required any major adjustment and the position has been kept unchanged even if it would have been more convenient to increase further our delta in the morning of March 25, as suggested by Trading Mate. But the strategy was already providing enough exposure.

The date each plot refers to can be read on the legend. Also, greeks (in particular delta) can be read below each plot and we can see how the delta exposure has been managed (approximately) in line with Trading Mate’s exposures.

This is just one of countless ways to use options with Trading Mate. We will add more examples in future.

More delta for “free” – a possible adjustment to increase exposure with little impact on the downside

All those who are following the Trading Mate portfolio know that we are currently long on EuroSTOXX 50. A bullish vertical spread has been opened in 2 moves, first buying the call 3775 and then waiting for the up move in the market to complete the structure by selling the call 3850. Both expiring in April.

Clearly the profit potential now is capped so what if there is a signal to further increase our exposure on EuroSTOXX? There are many alternatives, but most of them would involve increasing the risk on the downside.

Since we are confident in the models but we also know that anything can happen, an alternative to add positive delta adding little or no risk on the downside is preferred.

Our position looks like the spread below (in dotted line) while the proposed adjustment consists in adding a call backspread at almost no cost to add positive delta and vega to our strategy.

So, if Trading Mate will increase the exposure, this might be a good solution to ride the expected move in the market without having material new risk on the downside.

Profiting from VIX futures contango with low risk VXX and VIXY options strategies – theory and practice

The VIX index represents the current market expectation about the volatility of the S&P 500 index over the next 30 days. Being a volatility measure, there are features like volatility clustering and mean reversion that make VIX index relatively easy to predict. Volatility clustering means that high volatility tends to be followed by high volatility, hence creating a cluster of high volatility. But over a relatively longer period of time, volatility tends to revert towards a long term average level.

The fact that VIX index is relatively easy to predict might sound very promising but, unfortunately, it is not possible to trade the VIX index directly. The market offers instruments like VIX futures and options that can be used to express our view on the VIX index but by trading them we are actually trading the market expectation about where the VIX index will settle on the expiry date of the contract.

Making profits with VIX futures requires different set of strategies like the ones we use in our portfolio, so if interested you can access them below.

Access the updated portfolio

The results we obtain with such strategy can be seen below. Here we indirectly trade VIX futures by using SVXY, but similar results would be obtained by taking the opposite position on VXX or VIXY.

So we can successfully trade VIX futures (or related instruments like SVXY) but in this article, instead, we want to trade based on the fundamentals of the VIX market. Let’s focus on VIX futures. These contracts are already pricing the expected value that the VIX index will have when the contract expires. Even if the VIX index is easy to predict, that prediction is essentially already priced in the future price. The consequence is that predicting the VIX future prices is not as simple as predicting the VIX index.

There are other VIX-related instruments that retail traders might be more familiar with as they belong to the family of ETF/ETN. Popular examples are those we use in our trading portfolio mentioned above, like VXX, VIXY, SVXY, etc. All these instruments can be seen as portfolios of VIX futures so all we said for VIX futures is valid here.

This article will focus on VXX because of its high liquidity and characteristics that make it an interesting instrument to trade.

For those who are not familiar with this instrument, it is sufficient to know that it is a product that holds a daily rolling position of the first two monthly VIX futures. Every day, in order to maintain a 30 days exposure, the note will sell some units of the first month contract and buy some units of the second month contract.

One critical aspect to notice is that VIX futures are in contango most of the time. Under this configuration, futures that are closer to expire are trading at lower prices than contracts with longer expiries, i.e. the future curve is upward sloping.

Even if contango is the most common shape, VIX futures will sometimes be in backwardation. When this happens, we see future prices that are close to maturity trading at higher prices compared to longer term contracts, i.e. the curve is downward sloping.

Two examples are show below. The spot VIX is shown as red point, while the first 2 monthly contracts that are used by VXX are shown in green.

Clearly, in contango the future that is the closest to maturity (M1) trades at lower price compared to longer maturities (for example the second month, M2). The opposite is true when the curve is in backwardation.

It is a common belief that, during contango, the VXX daily rolling consisting in selling the front contract at lower price and buying the second month at higher price is the reason for the price erosion that we observe in VXX.

The truth is that the erosion is coming from the contango itself, not the daily rolling. After all, by holding a unit of VXX we are investing in a portfolio of long VIX futures and in contango these futures are “rolling down the curve” to settle at the spot VIX level on expiry date.

This is evident if we look at it in a static way. If VIX spot is not moving until M1 future expires, this future will eventually go down to converge to the spot value on expiry date. Given that VXX hold a quantity of this future, its value will go down accordingly. The same process is valid for M2.

So we know two things:

• VIX futures are in contango most of the time
• Contango is the source of erosion in VXX

Many investors generate regular returns by selling VXX to capture this price erosion. But shorting this contract can be extremely dangerous since VIX index might move up significantly driving up also the VIX term structure and hence the VXX.

If you look at the previous chart, it could easily be that we moved from the blue curve to the black curve in a matter of days. If that’s the case, a short VXX exposure that was held to capture the regular erosion due to contango will suffer significant losses.

So even if contango is eroding the VXX value, the VIX index might move significantly while a trader is trying to capture this price erosion. When the spot VIX moves, the futures will move as well (even if to a lesser extent) and a sustained rise in the spot price might lead to backwardation. When the curve is inverted, also the eroding forces we saw earlier during contango are inverted and contributing to VXX rising further even if spot VIX stays flat at that higher level for a while.

So harvesting the VXX erosion by simply shorting the VXX is not a safe game. A safer way to capture VXX erosion due to contango can be obtained by using VXX options.

One of the simplest strategies one could put in place is a broken wing condor or a broken wing butterfly. Possible strategies are shown below.

While the green one is a broken wing butterfly, the other two are broken wing condors. Let’s say we want to use the blue strategy. Assume VXX is trading at 20, the unbalanced condor in blue could be obtained by:

• Buying 2 puts strike 12 maturing in 75 days
• Selling 2 puts strike 15 maturing in 75 days
• Selling 1 put strike 17 maturing in 75 days
• Buying 1 put strike 20 maturing in 75 days

So, essentially we are buying a put bear spread (strikes 17 and 20) funded by a 2x put credit spread (strikes 12 and 15).

Most of the risk is on the downside so this position should not be taken when we expect VIX index to drop significantly. But this strategy is well positioned to benefit from the normal price erosion in VXX due to contango. This erosion is not fixed and can change depending on the VIX term structure but on average it is expected to be between 5% and 8% monthly. So there are few important points that we need to remember:

• We want to capture the price erosion in VXX due to contango
• We want to be hedged against possible spikes in VIX
• We want our profit zone to be the highest in the price region where we expect the VXX to be (by the time options expire) due to erosion

The unbalanced condor above is ticking all the boxes as by the time the options expire we expect the erosion to have driven the VXX price down by more than 15%.

During the life of the trade the VXX price will move driven by changes in the VIX term structures but we have hedged our risks on the upside and the only real risk is that VXX will move down more than what we expected due to erosion. Since this is certainly possible, it is important to leave a bit of buffer to accommodate a higher drop maybe due to an unanticipated decrease in VIX.

The chart below shows our P&L profile after 45 days in the trade.

Greeks profiles at two different point in time can be seen below.

So this is a low-risk strategy to profit from the typical shape of the VIX term structure and its impact on price erosion of VXX. The strategy can be put in place in one time or the trader can first add one spread and then the other. So, in the example above, if there is a long signal on VXX it is better to start with the put credit spread and then add the put bear spread once there is the short (or flat) signal. For more guidance in terms of expected direction you can access our updated portfolio page below. 

Access the updated portfolio