Regardless of timeframe, I always find it interesting when SPY and IWM are trading on opposite sides of a seemingly important moving average. Currently, this situation exists for the 370SMA (without a doubt the longest moving average I use). Below I show both ETFs on 1yr and 5yr timeframes. I’ll certainly be paying attention to see how this situation is resolved.
The correction over the past few months has reinforced the importance of having a hedging strategy in ones investment approach. The difficulty comes into determining when to take a short position and how much short exposure to accumulate. In addition to developing rules-based trading models at AlphaLab Capital, we also build proprietary indicators that help us determine downside potential of the overall market. For our Select Equity model, we use two independent quantitative methods to determine when to take a short position in the market.
The two indicators we use are called the Volatility Oscillator (VO) and Momentum Indicator (MI). The VO measures extreme overbought or oversold conditions and the MI measures the short-term trend of the overall market. When one of these indicators signals for a potential downside move in the market, we take a 50% short position in the Russell 2000 ETF (IWM). When both indicators are triggered simultaneously, we take a 100% short position in IWM. These short positions are typically bought on margin, since the model will always have a certain amount of long exposure.
Below is a graph showing historical data for the MI since April of 2010. For this indicator, we take a short position in IWM at every ‘High Turn’ data point and cover the short position when the model signals a new ‘Low Turn’ data point. We use a similar methodology for the VO indicator.
These two indicators have proven to be successful at consistently predicting market selloffs. Since April of 2010, the IWM ETF has gained approximately 8.5%. Over that same time, VO would have produced a theoretical net gain of 51.2% and MI would have produced a theoretical net gain of 46.1% by shorting IWM. I say ‘theoretical’ because these indicators where developed after years of data mining and they weren’t implemented into the trading system until their effectiveness was verified.
I started using both indicators in my daily trading strategy approximately six months ago. Since implementation, they have helped protect against market selloffs and been timely in going long. For example, after holding a short position in IWM for a few weeks, the indicators flipped bullish around the market bottom on July 5th as was detailed by the following tweets:
Right now, both indicators have a bullish bias; however, that can change quickly if the market trades down from here.
Go out and make you some coin.
I had an interesting dialogue with ZenHunter throughout the day today. After disclosing my positions this morning, he was quick to point out that I was holding quite a loser position in KERX. He was absolutely correct; I was holding a loser position in a bad stock. He was curious why I was holding onto a stock that was down 18% (down an additional 8%+ after today). The answer was simple: because I’m supposed to hold the position. I only sell positions when my model triggers a sell signal. The sell signal isn’t based on individual stock performance, but rather my models trading rules. Currently, they all say to hold the shares…
I don’t see many performance charts on this site, which is something I wish people were more willing to share. I think transparency is something to be respected. Regardless, I’d like to share my models performance over the past two years to provide a little clarity regarding the success that comes from following the model I’ve developed. Here’s the chart:
In just over two years of trading, the SEM has gained 146% (compared to a 15% return in the SPX). As you can see, it’ll have it share of drawdowns, but that’s just part of the game. The model consistently outperforms during market upswings.
Here are the backtested results for the model going back to the year 2000:
With my job search coming to a close, I now have more time to provide detailed updates regarding my Select Equity Model and its holdings.
Below is a summation of the current model open positions:
Indeed, the past few days have had there impact on my long positions. Losing money never feels good; however, our relatively large short position in IWM has lessened the pain.
Ever since my first trade, I’ve been interested in how one properly determines a stop loss level. After all, cut your losers quick and let your winners run was one of the mantras drilled into my head from other traders. I tried using average true range (ATR) multiples, moving average convergences, and historical resistance levels. Regardless of method, I was never satisfied with the result. Too often I would watch a stock trade down to my stop loss level just to reverse to the upside and start a new leg higher.
For those of you who don’t know, I’m a system based trader. I have a system that tells me what to buy, when to buy, how much to buy, and when to sell. Using a trading system such as this allows me to collect and analyze data most discretionary traders probably don’t track. For example, since each trade in my system has a distinct holding period, I can monitor the price action for each stock and develop rules around establishing stop loss levels.
When it comes to determining stop loss levels, the most obvious metric to track is holding period low versus the trades resulting gain or loss. Below is a graph showing all the trades for my Select Equity Model from 2010 to date (597 trades in total) with the holding period low on the y-axis and the trades resulting gain or loss on the x-axis:
For an example on how to read this graph, take a look at the data point highlighted by the red circle. For that trade, the stock dropped -14.4% from the buy signal then reversed +23.6% higher where the sell signal closed out the trade for a +9.2% gain. The other item to point out is the diagonal line extending from the origin to the lower left corner of the graph. Data points on that line simply represent trades that closed out on the lows for the holding period.
Okay, so we have all this data. Now, what to do with it? Let’s start by applying theoretical stop loss limits to the data and take a look at the results. First, let’s use a -5% stop loss limit. Filtering the data for a -5% stop loss limit gives the following results:
All of the data points with holding period lows above -5% were removed from the graph. Also, I’ve added a vertical red line that represents the -5% loss that would result if a -5% stop loss was used for each trade. All data points to the left of the vertical red line represent trades that would have sustained additional losses (i.e., the stop loss ‘worked’) whereas all data points to the right of the vertical red line represent trades that would have closed above the -5% stop loss level (i.e., the stop loss ‘didn’t work’). Let me state here that I am using the term ‘didn’t work’ very loosely. Clearly, stop losses are designed to protect against additional downside. I’m simply stating that for the purpose of these results, there are several examples where the stop loss would have closed the trade too early. Looking at these results, using the stop loss would have resulted in a -5.0% loss for each trade whereas not using the stop loss would have resulted in an average loss of only -4.4%. Meaning, a -5% stop loss (although an excellent mental safety net) would have actually had a negative impact on the model results by an average of -0.6% per trade. Some may think a -5% stop loss is too tight. Okay, what about using a stop loss of -10%?
Applying a theoretical stop loss limit of -10% produces the following graph:
Once again, all of the data points with a holding period low above -10% were removed from the graph. The vertical red line was also adjusted to represent the -10% stop loss. As was the case for the -5% stop loss, there are several trades on each side of the vertical red line. In this case, using the stop loss would have resulted in a -10.0% loss for each trade whereas not using the stop loss would have resulted in an average loss of only -9.0%. So, a -10% stop loss would have actually had a larger negative impact on the model results compared to a -5% stop loss (by an average of -1.0% per trade). What about using a stop loss of -15%?
Applying a theoretical stop loss limit of -15% produces the following graph:
Once again, all of the data points with a holding period low above -15% were removed from the graph. The vertical red line was also adjusted to represent the -15% stop loss. For this example, it can be seen that there are more data points to the right of the vertical red line than not. Meaning, most of the stocks that dropped -15% from the buy signal ended up reversing and closing higher at the sell signal. In this case, using the stop loss would have resulted in a -15.0% loss for each trade whereas not using the stop loss would have resulted in an average loss of only -11.7%. So, once again, a -15% stop loss would have actually had a larger negative impact on the model results compared to a -5% or -10% stop loss limit (by an average of -3.3% per trade).
Compiling the data for any whole number stop loss limit provides the following graph:
This graph shows the performance spread between using a stop loss limit and letting the trades unfold without stop loss limits. Most interesting is the red line, which represents the delta between the two methodologies. The data shows that letting the trades unfold without stop losses actually outperforms any system that employs stop loss limits (i.e., all the red data points are above 0%). Also, the spread in the performance actually increases as one loosens the theoretical stop loss limit (i.e., the more a stock drops from the buy signal, the more likely it is to reverse and close above the stop loss limit). That second point is counterintuitive and difficult to accept. When I’m watching one of my stocks down -10% or -20%, the last thing I’m thinking about is how likely it is to reverse higher. That being said, I’ve seen far too many examples of my -12% or -15% stops being taken out just for the stock to shoot higher (which is consistent with these results).
I find it interesting that this data supports the mantra of cutting your losers quick and letting your winners run. In this case, cutting your losers quick (by using a tight stop loss) has the lowest impact on the models trading results. As a result of this work, I now employ an -8% stop loss on most of my positions. I feel as though that gives enough room for the trade to work while protecting the downside. That’s the whole point, isn’t it?
After running approximately five points this week, the model closed out its position in IOC at $55.44 for a +3.1% gain. Also, as a discretionary trade, I covered about a third of my IWM short at $81.00 for a +1.4% gain. I decided to hold the remaining IWM short in case the market decides to reverse over the next few days.
The model is spitting out long recommendations at its fastest pace this year. Although not perfect, this model was designed to hold more long positions on upswings and fewer positions during market corrections. Considering the model now holds 23 long positions, it appears as though the model is expecting the market to wake-up on the right side of the bed over the short-term. The model opened another five new positions today and added slightly to the JAKK position. All of today’s trades are shown below:
Our Select Equity Model gained 0.75% today, which underperformed with the 1.36% increase in the SPX. As expected, today’s trading session was quite different than yesterdays. Several of my positions swung wildly through the day. For example, AN opened up about +3% after reporting earnings only to reverse down -9% and close the day only down -1%. Also, EEFT ramped up +6% after the open only to fade the rest of the day and close basically flat.
My two biggest losers today were AN (-1.0%) and IVN (-2.6%). Although a small position, IVN is turning out to be quite the loser position being down -7.8% since being added it to the model. The best performing positions in the model were FCS (+3.4%), PII (+3.7%), LF (+3.9%), and LGF (+5.4%). The model was held back quite a bit by the still large IWM short position, but that was the result I was willing to accept.
The model is currently positioned 118% long and 35% short (leverage = 1.53). This is the longest the model has been this year. Even though the averages tacked on some impressive gains today, I still don’t trust the market enough to cover my short position in IWM. Both proprietary indicators are still looking for higher prices. Here are the current values of each indicator:
Volatility Oscillator (V.O.) = +17.4 (short market if below 10 or above 90)
Momentum Indicator (M.I.) = 37.4↑ (long market until local maximum below 37.4)
Similar to early April, IWM has bounced solidly off the 150EMA (solid light blue line). A decisive move above $81.60 would prompt me to cover the remainder of my short position in IWM. Until then, I plan on holding the shares for some downside protection.
‘Destiny is calling me.’ – The Killers
The model closed out the position in LAD this morning at $25.15 for a -3.2% loss. The shares traded up most of the morning before fading and closing slightly under our exit price.
In addition to the seven new positions the model opened yesterday, the model opened another seven positions today. It is extremely rare for the model to initiate this many new positions over a two day period. Also, despite my comments last night, the model added to both FCS and PACW this morning. All of these trades were made based on the trading rules of the model. All of today’s trades are shown below:
Our Select Equity Model gained 0.12% today, which was inline with the 0.37% increase in the SPX. Altogether, the trading session was relatively uneventful with most of our positions vacillating in a relatively tight range. Clearly, the market was holding its collective breath for the big AAPL release after the bell. I suspect tomorrow will show more movement compared to what we saw today.
My biggest losers today were PII (-1.1%), IVN (-1.3%), and LGF (-3.2%). Some of the long positions that stood out from the crowd were FAF (+2.1%), JAKK (+2.7%), STKL (+4.4%), and IOC (+5.1%). The rest of the model performed inline with the market.
The model is currently positioned 98% long and 50% short (leverage = 1.48). Today, my Momentum Indicator flipped bullish. So, for the first time in three weeks, both my proprietary indicators are looking for higher prices. The SPX has lost approximately 40 points (-2.9%) since the last time both indicators had a positive bias. Although those timing indicators don’t always work, it’s nice when they do. Here are the current values of each indicator:
Volatility Oscillator (V.O.) = +21.4 (short market if below 10 or above 90)
Momentum Indicator (M.I.) = 37.7↑ (long market until local maximum below 37.7)
With both my indicators now solidly bullish, the rules of the model dictate that I close my short position in IWM; however, I didn’t cover a single share today. I didn’t cover any of my IWM short because I wasn’t comfortable being 98% long with no hedge against a slew of big-name earnings reports after the bell. Considering the market reaction to the earnings reports out of AAPL et al, it appears as though being 98% long would be a very profitable posture.
This is an excellent example regarding one of the difficulties in trading a rules based model. One can create a model with trading rules that result in significant outperformance over time; however, putting those rules into practice is much more difficult than it sounds. It is imperative to follow the rules in our trading model; however, it is more important to protect capital. From time to time, I will deviate from my model to do just that.
‘Rules made up by you, the only rules you should live by.’ – Pennywise
I sold all my shares in MKTX, SABA, and TITN this morning. Although TITN served me well over the past few weeks, it was time to close the position and reinvest the profits elsewhere. The details are below:
I opened several new positions today (seven total). Each position is relatively small, being between 5-7% of total capital. The model is not looking to add shares on any additional weakness; however, it may initiate new positions over the coming days. All of today’s trades are shown below:
Our Select Equity Model lost 0.79% today, which was inline with the 0.84% decrease in the SPX. Considering the model came into today with a 48% short position in IWM, I liked seeing the large gap down this morning. Unfortunately, the long positions in the model counterbalanced the benefit obtained from the IWM short, which ultimately resulted in a push against the averages.
My biggest losers today were FAF (-2.0%), SCVL (-2.7%), and LAD (-3.4%). Both DOV and IOC posted modest gains, while the IWM short provided the largest upside for the model being down 1.4%. I mentioned last week that I wanted to see a break below 1366 in the SPX, which we did; however, the bulls showed their resiliency and were able to reclaim the 1366 level by the market close.
The model is currently positioned 64% long and 49% short (leverage = 1.13). Currently, only one of my proprietary indicators is signaling additional downside risk. The Volatility Oscillator turned bullish late last week. Here are the current values of each indicator:
Volatility Oscillator (V.O.) = +17.8 (short market if below 10 or above 90)
Momentum Indicator (M.I.) = 31.9↓ (short market until local minimum above 28.8)
The M.I. is trending down and will more than likely flip bullish with any higher close in the market tomorrow. If that happens, I will cover my entire short position in IWM.
I suspect we will see higher prices tomorrow morning and into the FED meeting announcement on Wednesday. As such, I took my 401(k) long exposure from 66% to 80% on today’s close. If we sell off from here, I still have 20% cash to put to work.
‘I can’t ignore you anymore.’ – Red
I sold short more shares of IWM at $80.60, just before the economic data release at 10:00am EST. I covered that portion of my short position at $79.60 for a small 1.2% gain. I’m still holding a sizeable short position in IWM (approximately 48%) and looking for additional downside.
I opened new positions in DOV and RJF today. Dover Corporation (DOV) reported earnings yesterday and raised the lower end of their 2012 EPS forecast $0.10, from $4.70 to $4.80 while keeping the high end of the range at $5.00. The stock reacted poorly to the report and traded lower over the past two sessions. I have a stop loss at $57.69, slightly below 150 and 200 day SMAs. The shares of Raymond James Financial (RJF) have been trending down for the past month. I expect the bulls to defend the $33.50 level and I have stop loss at $32.69. All of today’s trades are shown below:
Our Select Equity Model gained 0.91% today, which outperformed the 0.59% decrease in the SPX. Although the model performed well today, it is still lagging the SPX for the week. The price action during the first few hours was extremely interesting. The market opened and ramped up ahead of the economic releases at 10:00am EST. The data came in soft and the market sold off, based for a few minutes, made new intraday highs, and then sold off the most of the day. It was pretty interesting price action to watch, especially if you’re holding a large short position like myself.
My two biggest losers today were LAD (-1.1%) and SCVL (-1.3%). The rest of the model performed fairly well with the largest gainer being TITN (+4.0%). It looks like TITN has found solid support at the $32 level and is now working to the upside. The IWM short position certainly worked well today; however, I’d like to see the SPX break below the 1366 level. Until then, we’re looking at higher lows on the index.
The model is currently positioned 65% long and 48% short (leverage = 1.13). Both my proprietary indicators are still signaling additional downside risk. Here are the current values of each indicator:
Volatility Oscillator (V.O.) = +9.3 (short market if below 10 or above 90)
Momentum Indicator (M.I.) = 34.9↑ (short market until local minimum above 28.8)
The V.O. is getting close to flipping bullish, but it’s not there yet. The M.I. is trending up; however, it needs to make a higher swing low (i.e., higher local minimum) before I’ll cover my short position in IWM.
On Tuesday, my relatively large short position in IWM felt like an anchor. Today it felt like a warm hug.
‘You used to smile at the thought of me failing.’ – Chevelle
Although I didn’t make any sales today, I will be selling my entire position in both BPZ and FELE sometime tomorrow (regardless of price action or market backdrop).
I opened new positions in FAF and LAD this morning. First American Financial (FAF) shares have been rising steadily over the past five days and I’m hoping to see that trend continue. Lithia Motors (LAD) shares have been on a nice steady climb since mid-2010. I mentioned previously that I would buy back some of the TITN shares I sold last week if the stock hit the $34 level, which is what I did toward the end of the day today. All of today’s trades are shown below:
Our Select Equity Model lost 1.07% today, which significantly underperformed the 1.55% rise in the SPX. This was one of those days where my long positions failed to outpace my relatively large short position in IWM. On days like today, I tend to find myself walking around the office channeling Too Short and thinking ‘I wish I was a little bit taller.’
My two biggest losers today were TITN (-4.4%) and BPZ (-5.7%). It seems as though TITN still had some work to do on the downside after ripping 30%+ after earnings. I added to the position near the close. As for BPZ, the selling started early in the morning and persisted through most of the day today. Both SABA and FELE posted mediocre gains.
The model is currently positioned 58% long and 49% short (leverage = 1.07). Both my proprietary indicators are still signaling additional downside risk. Here are the current values of each indicator:
Volatility Oscillator = -7.6 (short market if below 10 or above 90)
Momentum Indicator = 30.3↓ (short market until local minimum above 30.1)
The details behind each indicator are a bit complicated; however, I wanted to provide a little more insight into what I’m looking at and what it would take for me to cover my short position.
Also, it’s important to for me to restate the purpose of my short position. Whatever I take a short position in the market, it is not a prediction that the market will proceed into a steep decline. Rather, when I take a short position, it acts more of a hedge against my long positions. I hold a short position quite often. As a matter of fact, I hold a short position hedge about 75% of the time. However, on days like today, the short position can feel like an anchor.
‘I wish I was a little bit taller, I wish I was a baller.’