Article Categories

Financial Concepts

Performance Metrics

# INTRODUCTION

The HIIT strategy aims to increase the number of trades per year while also maintaining a relatively low temporal exposure to the market. In back testing HIIT performed roughly 10 to 12 trades per year with a Pwin (Probability of Win) percentage above 80%.

HIIT adheres to the MCI principle that it is best to invest into market weakness in an otherwise up-trending market. It buys into weakness (oversold) and sells into strength (overbought). HIIT has proprietary algorithms and metrics that are designed to determine the strength of current market conditions and dynamically adjust itself accordingly. The algorithm adjusts its level of aggressiveness by scaling the size and number of trades from high to low as the market moves from a robust bull market to a fragile bull market. As the market becomes less bullish and more bearish, the algorithm begins to mix conservative long trades with conservative short trades. During bear market conditions, the algorithm only conducts short trades. The opposite transition also takes place as the market transitions from a bear market back to a bull market.

Each HIIT trigger is made by assessing short term weakness based on the magnitude of its pullback and the rate in which it falls. If the magnitude and rate of the market drop are within acceptable parameters, a buy trigger will be made. Each trigger is based on the concept of “Averaging Into” a position. A position may open with a fraction of available capital. If conditions become more favorable after opening the position, the algorithm will indicate allocating additional capital to the position. This process is known as position sizing or scaling into a position. Because our algorithms can’t guarantee the most optimal time to open a position, this process is used to potentially capture many small while initially committing less capital with the idea of increasing the position at a more favorable price, thereby reducing the average cost of entry and also reducing the risk of a losing trade.

averaging into positions is done to buy into market weakness and sell into strength. However, the difficulty lies in knowing where the bottom of the weakness occurs. By increasing the number of possible entries, a trader can decrease his/her margin for error and increase the probability of a winning trade (Pwin). Our algorithms evaluate multiple metrics to determine how to average into a position. These metrics provide measures of various market conditions including short and long term trending, market strength, volatility, etc. When the conditions are considered safer (i.e. the probability of a winning trade based on historical performance exceeds a predetermined threshold), the algorithm becomes more aggressive and scales in with larger positions. Conversely, when conditions are deemed riskier (i.e. the probability of a winning trade meets a lower threshold) the algorithm scales in more conservatively.

# SCALE-IN POSITION

This algorithm trades SPXL on the long side and SPXS on the short side.

SPXL is the Direxion Daily S&P 500 Bull 3x Shares, which seeks daily investment results, before fees and expenses, of 300% of the S&P 500 index. Conversely, SPXS is the opposite of SPXL and seeks 300% of the inverse (or opposite) of the performance of the S&P 500 Index. When conducting trades, both are long purchases, the algorithm does not actually enter short positions on any ETFs.

It is possible to utilize the HIIT triggers with other ETFs of varying leverage (1x or 2x) on the S&P 500 index. Limited back testing on these lower leveraged products shows reduced equity volatility but at the expense of greatly diminished returns over time.

Below are illustrations (not real trades) explaining how a **HIIT** scale-in position may occur.

In each of these examples, it is assumed that the trading account (or the capital available to trade this strategy) is $10,000. The percentages illustrated will be against this$10,000 amount.

Figure 1 provdes an illustration of possible scale-in positions where the algorithm has chosen a 10% / 20% / 30% / 40% allocation based on determined market conditions.

FIGURE 1: Example showing a long scale-in position using 10% / 20% / 30% / 40% allocation.

• Example A: shows a single trigger for a 10% allocation followed by a market rebound before a second scale-in allocation could be made.
• Example B: shows a trigger for a 10% allocation at position (1). The market then moved down and a 20% trigger allocation occured at position (2). The market then rebounded before a third level of scale-in could be established.
• Example C: shows a trigger at position (1) for a 10% allocation followed by a 20% allocation at position (2), a 30% allocation at position (3), and a final allocation of 40% at position (4). at position (4) the algorithm has signaled that allavailable capital should be allocated (10% + 20% + 30% + 40% = 100%). These additional positions reduce the cost of entry relative to the first trigger.

The second example in Figure 2, below, illustrates a similar possible scale-in position where the algorithm determined that a 50% / 50% allocation is warranted.

FIGURE 2: Example showing a long scale-in position using a 50% / 50% allocation.

• Example A: shows a single trigger for a 50% allocation followed by a market rebound before a second scale-in allocation could be made.
• Example B: shows a trigger for a 50% allocation at position (1). The market then moved down and an additional 50% allocation occurs at position (2). The market then rebounded and all shares were sold at position (S).

There are also times when the HIIT algorithm indicates a single scale-in position of 100% of the available funds, and this is illustrated in Figure 3.

FIGURE 3: 100% allocation trade setup.

• Example A: shows a single scale-in trigger for 100% followed by a market rebound resulting in a positive trade.

The HIIT algorithm attempts to conduct quick trades with open positions that last on the order of a few days. However, under some conditions the algorithm will stay in the trade for a longer period as it attempts to ride a longer-term uptrend. These longer trades can last for several weeks or even months These trades are relatively rare and occur roughly 10% of the time in the back testing. Often they are far more profitable than are the smaller, quicker trades. Figure 4 illustrates how such a trade may manifest.

FIGURE 4: 50% / 50% allocation trade setup.

• Example A: illustration of a two scale-in triggers followed by a longer term uptrend. Conditions in this example are right to allow the algorithm to stay in for a longer more profitable trade.

# BACK TESTING RESULTS

## REAL SPXL DATA 2008 TO 2016

Below is the back testing report on this strategy going back to the inception of SPXL and SPXS in the year 2008.

TABLE 1: HIIT results from the inception of SPXL to Dec 2016** (NOTE: HIIT data is not correct. New data will be posted soon. The following discussion may reference the data that will be showing up in the near future.)

The results in Table 1 show that HIIT is a high performing strategy that significantly outperforms the benchmark of holding the SP500. Over the 8+ year period it had a Compound Annual Growth Rate (CAGR) near 50%. Along with a high CAGR, this strategy also provided a very high Pwin of 88%, thereby providing a high level of confidence in its outperformance over the benchmark. Note that a single trade is determined between the first allocation until the time all funds are sold. So, whether 10% or 100% was allocated, the win assessment for a particular trade is determined by assessing whether money would have been made at the time all funds were sold. In short, if the account has more money after selling all funds than it did prior to the purchase of any shares, then it is a win.

This strategy is not perfect, as nearly 12% of the trades resulted in a loss. The average recovery time from a losing trade, however, is very reasonable. Over this time it was just 32 bars where a bar is a trading day (i.e. weekends and holidays are not counted). This indicates that when a losing trade occurs, there is a high probability that subsequent trades can recover in approximately six weeks on average. With the high CAGR, a high Pwin, and a higher average profit vs loss per trade, the occasional losing trade is tolerable given the overall strategy performance.

The remainder of this discussion is being reviewed and is not accurate at this time. Updates will be coming soon.

Another very important statistic is the exposure to the market. This performance metric measures the percentage of trading days that this strategy has an open trade. Generally, with a given CAGR and Pwin, the lower the number the better. This measures the efficiency by which the strategy is able to achieve its results. It also is a measure of risk, the less amount of time that a trader is in the market the less likely he/she will be exposed to potential unexpected market movement that may result in a equity drawdown. So, while one needs to be in the market in order to make money, it is best to be in the market at the most optimal periods to make that money, and to sit in cash otherwise. The results in Table show that HIIT’s exposure is just 20%. That means that this strategy is exposed to the market only 20% of the time, and yet is still able to achieve a outperformance of 50% on the CAGR. Compare that the buy and hold benchmark of the SP500 where it is always exposed and achieves a CAGR of only 12%.

EXAMPLES OF ALGORITHMIC TRIGGERS

Below is a price chart of the SPXL from 2013 to 2016 showing a sample of HIIT buy and sell triggers.

The green up arrows depict buy triggers, while the red down arrows depict when all shares were sold.

FIGURE 5: Buy and Sell trigger examples from the HIIT strategy between 2013 and 2016

EQUITY GROWTH OF BACKTESTED RESULTS

Below is the equity growth over time from the inception of SPXL in 2008. This is a linear plot illustrating what would have occurred by reinvesting all profits right back into the strategy.

FIGURE 6: Linear equity plot of the HIIT strategy from 2008 to 2016

Looking at figure 6 above, it can be seen that the HIIT strategy far outperforms the buy and hold of the SP500. It does so, so much that the red line depicting the buy and hold on the SP500 is barely visible.

Keep in mind that we do not expect anyone including ourselves to take $10,000 and trade it to tens of millions in a mere 8 years’ worth of trading. This is for illustration purposes only, to illustrate the potential for this strategy to boost our existing investment portfolio. To learn more about how our trading strategies are not ‘get rich quick schemes’, please click here. Below is a log plot of what is seen in Figure 6. This is a useful plot because it allows the reviewer to see variability throughout the entire period of testing. It basically linearizes the exponential effects of the compounded growth rate, thereby allowing the reviewer to observe the consistency of the strategy over time FIGURE 7: Log equity plot of the HIIT strategy from 2008 to 2016 Figure 7 above shows that the strategy is very consistent over the 8 year back test period. The dotted line is an exponential regression through the data based on a compounded growth of 50% per year. The R2 value is 0.98, meaning that the line does a very good job of describing the data. A perfect score would be 1.00. This means that the strategy performs consistently well year after year. INCLUSION OF SIMULATED SPXL/SPXS In our opinion, 8 years of back testing data is not enough for us to gain trust in the algorithmic trading strategy. To increase the size of our SPXL/SPXS back testing dataset we generated simulated these from the underlying issue SPX. SPXL is the Direxion Daily S&P 500 Bull 3x Shares where it seeks daily investment results, before fees and expenses, of 300%. Conversely, SPXS is the opposite where it seeks 300% of the inverse (or opposite) of the performance of the S&P 500 Index. We took the known statistics between the relationship between SPXL/SPXS and SPX and generated a matching profile that could be applied to SPX data that runs from 1991 to 2008 to create a complete dataset of 25 years’ worth of back testing data. We then verified that the simulation matched well against real data from 2008 forward. The table below is a summary of the results over the last 25 years of back testing data – which includes both simulated and real SPXL/SPXS data. TABLE 2: Summary of HIIT results from 1991 to Dec 2016** (NOTE: this is only a placeholder at this time actual HIIT statistics will be updated at a later date) The results of this back testing are very encouraging. It generated a CAGR of over 50% over that period with a win rate over 90%. It was able to recover quickly from a losing trade when they did occur, and had a relatively low exposure rate in conjunction with a high growth rate. The results below provide further evidence that this is a high performing trading strategy. EQUITY GROWTH OF BACKTESTED RESULTS USING SIMULATED SPXL/SPXS Below in Figure 8 is the 25-year linear equity results of what occurs with a starting value of$10,000 invested into the HIIT strategy. It also includes the SP500, provided as a reference. Please note that the SP500 profile is plotted using the secondary right hand axis. This is done so that the reviewer can easily compare the shape and volatility of both to see how the HIIT strategy performed in various market conditions.

FIGURE 8: Linear equity plot of the HIIT strategy from 1991 to 2016. Also provides the SP500 on the right hand axis as a reference

When looking at the SP500 profile, notice the amount of volatility that exist in the broader market as well as the inconsistent direction of the market over the 25-year period. During this time period there were two severe bear markets that resulted in a 45% drawdown from 2000 to 2003 and a 55% drawdown that occurred between 2008 and early 2009.

One area that we were especially interested in assessing was how the strategy would perform as the broader market transitioned from a long term bull market to a severe bear market, and then back into a bull market. As evidenced in Figure 9 below, the HIIT algorithm performed very well during the two bear markets and transitioned gracefully as the bull market picked up.

The log plot in figure 9 linearizes the exponential effects of the compounded growth rate, thereby allowing the reviewer to observe the consistency of the strategy over time. This can be seen by looking at the dotted red regression line through the data below in figure 9. Notice how consistently the data follows this linear regression straight through the bull markets as well as the bear markets. This means that the strategy was consistent in maintaining a CAGR of 50% over the entire 25 year period.

FIGURE 9: Log equity plot of the HIIT strategy from 1991 to 2016. Also provides the buy and hold SP500 results on the right hand scale.

CONTINUOUS TWO-YEAR PERFORMANCE

Lastly, we want to provide further evidence of the consistency that the HIIT algorithm has been able to perform over a long period of time.

Figure 10 represents the absolute 2-year return that we could have expected when beginning the strategy on any given trading day over the past 25 years. It takes the 2-year return beginning on the first day 25 years ago and then rolls it forward to attain thousands of 2-year samples. The results show that regardless of the starting point over that 25-year period, the return 2-years later would have been very good.

The lowest performing 2-year period was in the 40% range with the average over 150%.

FIGURE 10: Histogram on the continuous two-year performance

# CONCLUSION

Each HIIT trigger is made by assessing short term weakness based on the magnitude of its pullback and the rate in which it falls. The algorithm assesses the magnitude and rate of the market drop to determine if a buy trigger will be made. This strategy identifies when the market is oversold and then averages into the position as it becomes more oversold.

We feel the HIIT strategy is a high performing strategy that can be used to amplify our existing investment strategies. It has a Pwin of 90% and a CAGR of over 50% through 8 years of testing on SPXL/SPXS.

The addition of simulated data going back 25 years suggests that the trading algorithm is a high performing strategy able to benefit from a large diverse range of market conditions that include various levels of volatility in both bear and bull markets.

Disclaimer: Market Chronologix, Inc. makes a good-faith effort to accurately convey the performance metrics of our strategies, but we assume no liability for incorrect information, or for any losses that may be incurred as a result of using these strategies. Past performance of our strategies does not guarantee or imply that they will continue to perform at the same level in the future. All investing involves a degree of risk. You may have a profit or a loss when you sell shares of an investment, and you should carefully consider what level of risk you can accept before investing any money. We are not registered financial advisers and do not offer investment advice, nor should any of our written materials or services be construed as such.