One of the most interesting findings dealt with in a previous posting (Pairs Trading (ETFs) was the RTHs (Retail HOLDR.) salient feature of being a favorable candidate for a potential mean-reversion strategy in conjunction with a major market or sector ETF. With respect to the primarily method used for cointegration (the augmented Dickey-Fuller test), the RTH showed a probability better than 90% of being cointegrated with the IWM (Russel 2000) and the SMH (Semiconductor HOLDR.), and missed being cointegrated with SPY and QQQQ by a hairbreadth only (two price series are called cointegrated if the pair has a consistent mean and standard deviation, both prices series never indefinitely wandering off in opposite directions and never drifting farther and farther away from its mean without eventually returning to the initial ratio or mean).

But the RTH doesn’t seem to be a favorable candidate for a longer-term (the half-life – the expected time to revert half of its deviation from the mean – is regualary measured in weeks or month) pairs trading strategy only, but may provide favorable short-term mean-reversion opportunities as well (market timing).

Table I below shows the performance metrics (since 06/01/2001 due to the RTH‘s inception in May 2001) for different pairs in conjunction with the RTH and – for demonstration puposes – different pairs of major market ETFs (SPY, QQQQ and IWM) and sector ETFs (XLY – Consumer Discretionary – and XLP – Consumer Staples -) based on an exemplary mean-reversion strategy, assumed one would’ve bought the pair (is equivalent to buying the first and selling short the second ETF in equal money amounts (number of shares in each ETF = 100% net asset value / share price)) on close of a session when the 4-day EMA (Exponential Moving Average) of the pair (the ratio of the closing prices) is less than the 4-day EMA of the ratio of closing prices for yesterday, and vice versa (selling short the first and buying the second ETF in equal money amounts in the event of a rising 4-day EMA of the ratio of closing prices).

Here is the link to the stats in a more ‘readable’, original size: Statistics 1

While – with respect to the specific setup defined – the SPY vs. QQQQ, the SPY vs. IWM, the XLY vs. XLP and the SPY itself (buy and hold) as a benchmark virtually went nowhere (or even closed in the red) over the course of the last 9 years – especially after accounting for fees and transaction costs -, the RTH as a pairs trading component in conjunction with the SPY, the QQQQ, the  IWM and the SMH not only easily out-performed a (S&P 500) buy-and-hold approach by a wide margin (and almost year by year as well, see ‘Periodic Returns‘ in the stats above), but comes up with a smoother equity curve as well, meaning there are much less dramatic departures from a gradually/geometrically increasing trendline (R-squared, maximum drawdown, maximum sessions in drawdown) in comparison to a SPY‘s buy and hold approach.

Interestingly the SPY vs. RTH and SMH vs. RTH pairs trading strategies and a S&P 500 buy-and-hold approach do NOT differ with respect to the probability of a winning trade (the probability is almost always slightly above 50% only). The reason for the deviation in total returns is the fact that – in contrast to the SPY‘s buy and hold approach – the median winning trade (+0.51%) now equals or slightly exceeds the median losing trade (-0.50% ), significantly improving the respective expectancy (probability of winning * average gain – probability of losing * average loss).

But a SPY vs. RTH‘s pairs trading strategy has another advantage as well: Chosing a slightly different setup in order to especially exploit those reversal opportunities where the pair is (from a historical and statistical perspective) exceptionally stretched to one or the other side would be sufficient to not only surpass previous compounded returns, but to cut in half the time in market and the maximium drawdown as well. Table II below shows the performance metrics (since 06/01/2001 due to the RTH‘s inception in May 2001) for the same pairs, assumed one would’ve bought the pair (buying the first and selling short the second ETF in equal money amounts) on close of a session when the pair (the ratio of the closing prices) closed at least -0.50% below its 4-day EMA, and vice versa (selling short the first and buying the second ETF in equal money amounts in the event of a close at least +0.50% above the 4-day EMA of the ratio of closing prices).

Here is the link to the stats in a more ‘readable’, original size: Statistics 2

With respect to SPY vs. RTH (Strat. #1), time in market and maximum drawdown have been exactly cut in half (giving you the chance to earn an additional return on cash) while the geometric growth rate per trade doubled. Although the probability of a winning trade (again) only slightly improved (from 54.63% to 57.44%), it is (again) the effectiviness (doing things right instead of doing the right thing only, meaning increasing your gains when you’re right and cutting your losses when you’re wrong) of the strategy which makes for the improvement in key performance metrics.

But what about the robustness of a SPY vs. RTH pairs trading strategy ? It it works with a -0.50%/+0.50% level below/above a 4-day EMA, it should work with a -/+0.30% up to a -/+0.70% level and a 3-day and 5-day EMA as well showing some gradual – no radical -  changes with respect to the key performance metrics only.

Table III below shows the performance metrics for the SPY vs. RTH pairs trading strategy, assumed one would’ve bought the pair (buying the first and selling short the second ETF in equal money amounts) on close of a session when the pair (the ratio of the closing prices) closed at least

• Strat. #1: -0.30% below (long) and +0.30% above (short) its 4-day EMA,
• Strat. #2:-0.40% below (long) and +0.40% above (short) its 4-day EMA,
• Strat. #3:-0.50% below (long) and +0.50% above (short) its 4-day EMA,
• Strat. #4:-0.60% below (long) and +0.60% above (short) its 4-day EMA,
• Strat. #5:-0.70% below (long) and +0.70% above (short) its 4-day EMA.

SPY vs. RTH (Strat. #6) represents a buy-and-hold approach (assumed one would always be long the SPY and short the RTH).

Here is the link to the stats in a more ‘readable’, original size: Statistics 3

And last but not least, table IV below shows the performance metrics for the SPY vs. RTH pairs trading strategy, assumed one would’ve bought the pair (buying the first and selling short the second ETF in equal money amounts) on close of a session when the pair (the ratio of the closing prices) closed at least

• Strat. #1:-0.50% below (long) and +0.50% above (short) its 3-day EMA,
• Strat. #2:-0.50% below (long) and +0.50% above (short) its 4-day EMA,
• Strat. #3:-0.50% below (long) and +0.50% above (short) its 5-day EMA.

SPY vs. RTH (Strat. #4) represents a buy-and-hold approach (assumed one would always be long the SPY and short the RTH).

Neither a slight variation in the percentage level below/above the 4-day EMA nor a variation in the duration of the EMA itself affects any of the strategy’s key performance indicators in a significant way, except – but expectedly – the so called opportunity factor (total number of sessions and time in market).

Summary: A SPY vs. RTH‘s pairs trading strategy, assumed one would’ve bought the pair (buying the SPY and selling short the RTH in equal money amounts) on close of a session when the pair (the ratio of the closing prices) closed at least -0.50% below its 4-day EMA, and vice versa (selling short the SPY and buying the RTH in equal money amounts in the event of a close at least +0.50% above the 4-day EMA of the ratio of closing prices), historically provided a (consistently) profitable market timing strategy (a median annual return of +15.65%), (consistently, in 8 out of the last 9 years) out-performing a S&P 500 buy-and-hold approach, with a smooth equity curve (R-squared, maximum drawdowns on a week/month/year end basis, maximum time in a drawdown), meeting at least basic requirements for robustness and reliability. Unfortunately a shortcoming is the deviation in yearly returns (one standard deviation = 32.72%).

A favorable basis for some further investigations and refinements (accounting for return on cash, position sizing, and making the strategy adaptiv to changing market conditions – if necessary).

to be continued …

Successful trading,
Frank

Remarks: Due to their conceptual scope – and if not explicitely stated otherwise -, all models/setups/strategies do not account for slippage, fees and transaction costs, do not account for return on cash, do not use position sizing (e.g. Kelly, optimal f) – they’re always ‘all in‘ -, do not use leverage (e.g. leveraged ETFs) - but a marginable account is mandatory -, do not utilize any kind of abnormal market filter (e.g. during market phases with extremely elevated volatility) , do not use intraday buy/sell stops (end-of-day prices only), and models/setups/strategies are not ‘adaptive‘ (do not adjust to the ongoing changes in market conditions like bull and bear markets).

________________________________

If you might want to be instantly notified about what’s happening in the markets and at TRADING THE ODDS, I encourage you to subscribe to my RSS Feed or Email Feed, and (or) follow me on Twitter.

xx

Disclaimer: Long SMH and short XRT at time of writing.

The information on this site is provided for statistical and informational purposes only. Nothing herein should be interpreted or regarded as personalized investment advice or to state or imply that past results are an indication of future performance. The author of this website is not a licensed financial advisor and will not accept liability for any loss or damage, including without limitation to, any loss of profit, which may arise directly or indirectly from use of or reliance on the content of this website(s). Under no circumstances does this information represent an advice or recommendation to buy, sell or hold any security.

First of all thanks for your patience, and from now on I’ll be posting again on a more frequent basis.

And furthermore I’d like to advise those interested in quantitative research of a new blog I just came across: Engineering Returns by Frank Hassler.

____________________________________

Due to the fact that I’m a big fan of statistical arbitrage (and trading it for a living), I thought it would be interesting to check if – and to what extend – there are pairs of ETFs (Exchange Traded Funds) which – as always based on historical data, statistical anomalies, regularities and irregularities, … – would provide a favorable and tradable edge maintaining a market neutral position.

I personally prefer ETFs to individual stocks due to the fact that the latter are much more sensitive to unforeseeable events and/or outcomes like earnings, fundamentals, crew changes (CEO, CFO, …), rate disputes, strikes, take-overs, force majeure (casualties, disasters, …). And I speak from my own experience …

In conjunction with pairs trading, you’ll probably hear about two (quite different) concepts: correlation and cointegration.

Correlation states the degree to which the (daily, weekly, monthly …) returns of two series of prices (e.g. the S&P 500 and the Nasdaq 100) will move in the same direction most days/weeks/month over a period of time (but  probably drifting farther and farther away from each other due to deviations in the magnitude of daily returns), while a pair (being long one and short the other series of prices in the right proportion) is called cointegrated if it has a consistent mean and standard deviation, both prices series never indefinitely wandering off in opposite directions and never drifting farther and farther away from its mean without eventually returning to the initial ratio or mean (mean-reversion).

But – unfortunately – the so-called half-life (the expected time to revert half of its deviation from the mean) concerning a cointegrated pair of price series will regularly be measured in weeks or month (see stats below), and I’m more a high-frequency trader looking for opportunities on a day-by-day basis.

The following are the ETF’s I’ve been utilizing for my investigations, meeting the necessary requirements like adequate liquidity (daily trading volume), as low as possible transaction costs (narrow bid/ask spreads), adequate volatility (in order to justify the arising high transaction costs), among others:

• SPY: S&P 500
• QQQQ: Nasdaq 100
• IWM: Russel 2000
• SMH: Semiconductor
• RTH: Retail

Other ETFs may be subject to a follow-up posting.

To test for cointegration, the primarily method used is called the augmented Dickey-Fuller test. If two price series are cointegrated (with a probability of better than 90%), the Dickey-Fuller test would’ve to come up with a t-statistic exceeding the 90% critical value of -3.038 (in absolute terms), otherwise the hypothesis that those two price series are conintegrated would be rejected. The following table provides the respective t-statistics based on the augmented Dickey-Fuller test for the time frame between 01/01/2002 and 08/06/2010 (price series are adjusted for dividend and cash payments).

Interestingly there are only two pairs – IWM vs. RTH (gt. 90%) and SMH vs. RTH (gt. than 95%) – which are cointegrated with a probability of better than 90%, while the SPY (as a proxy for the S&P 500) and the QQQQ (as a proxy for the Nasdaq 100) show the least probability for being cointegrated. IWM vs. RTH shows a half-life of 194 sessions, and SMH vs. RTH a half-life of 92 sessions. Both pairs seem to be good candidates for a (longer-term) mean-reversion strategy.

A second interesting observation is that even in conjunction with SPY and IWM, the RTH (Retail HOLDRS) seems to be a favorable candidate for a potential mean-reversion strategy.

But fortunately cointegration is not mandatory in order to find a profitable mean-reversion strategy, and on a day-by-day basis even non-cointegrated pairs (like the SPY vs. QQQQ) may provide favorable short-term mean-reversion opportunities (better fitting my style of trading). So my next step was to check for the pair’s performance based on the easiest mean-reversion strategy:

Buy (on close) the pair in the event the ratio of ETF X and ETF Y closed lower (means ETF A under-performed ETF B on the respective session), and sell short (on close) in the event the ratio of ETF X and ETF Y closed up (means ETF A out-performed ETF B on the respective session). Due to the RTH‘s inception date in 2001 start date for the following stats is always Jan. 1, 2002.

“Buy” means buy ETF A and sell short ETF B (and vice versa), the number of respective shares specified by the ratio of closing prices (e.g. if the ratio of ETF A’s and ETF B’s closing prices is 3, one would sell short 3 shares of ETF B for every share bought of ETF A). A marginable account would be mandatory, especially due to the fact that it is assumed that one would invest 100% of the then current net liquidation value on both sides of the market (means 100% on the buy and 100% on the short side).

(FAQs and a glossary concerning the stats can be found at the FAQ/GLOSSARY page)

Here is the link to the stats in a ‘readable’ size: Statistics 1

Interestingly it is again the RTH in conjunction with every other ETF which delivers the best results, always exceeding the 200% mark for compounded returns (gross profits before applying commissions, slippage and fees). Unfortunately commissions, slippage and fees would regularly eat up a major part of the compounded return, due to the fact that one would always have a position in the market, with an exposure of 200% (100% on the buy and 100% on the short side), and reversing one’s position (switching from the long to the short side of the pair and vice versa) would quadruple the respective transaction costs in comparison to somone who simply closes a long or short position with an 100% exposure.

In a second step I utilized a little bit more sophisticated concept (Bollinger Bands %B with 4-days EMA and 1 standard deviation):

Buy (on close) the pair in the event the Bollinger Bands %B closed below 0.35, and sell short (on close) in the event the Bollinger Bands %B closed above 0.65.

For a detailed explanation of the Bollinger Bands %B concept see Stockcharts.com. In other words: Buy the pair in the event the ratio closed almost (< 0.35) one standard deviation below its 4-day exponential moving average (ETF A is short-term ‘oversold’ in comparison to ETF B), and sell short the pair in the event the ratio closed almost (> 0.65) one standard deviation above its 4-day exponential moving average (ETF A is short-term ‘overbought’ in comparison to ETF B). A classical mean reversion concept.

Here is the link to the stats in a ‘readable’ size: Statistics 2

Things are (partly) significantly improving: Compounded returns, t-score (vs. chance and benchmark) are increasing while transaction costs, maximum drawdowns are decreasing (now Time in Market is less than 100%, with a smaller frequency of closing or reverting one’s position), and especially the SPY vs. RTH and IWM vs. RTH pairs show promising results to be worth some further investigations.

More to come in a follow-up post (at time of writing it’s almost midnight in Germany) …

to be continued …

Successful trading,
Frank

________________________________

If you might want to be instantly notified about what’s happening in the markets and at TRADING THE ODDS, I encourage you to subscribe to my RSS Feed or Email Feed, and (or) follow me on Twitter.

xx

Disclaimer: No position in the securities mentioned in this post at time of writing.

The information on this site is provided for statistical and informational purposes only. Nothing herein should be interpreted or regarded as personalized investment advice or to state or imply that past results are an indication of future performance. The author of this website is not a licensed financial advisor and will not accept liability for any loss or damage, including without limitation to, any loss of profit, which may arise directly or indirectly from use of or reliance on the content of this website(s). Under no circumstances does this information represent an advice or recommendation to buy, sell or hold any security.

A few days ago Michael Stokes at MarketSci made an excellent post concerning RSI(2) readings, the changing frequency of extreme readings, its (the RSI’s) quality of forecast and efficiency of trading short-term mean-reversion in the markets over the course of time since 1950 (Extreme RSI(2) Readings Becoming Less Common, and he already covered the topic here and here), and based on his findings and conclusions -for exemplary purposes- I set up a static trading strategy (static means no adjustments of RSI time frame and break points) for the time frame since 1993 (SPY, see Trading the RSI as a Static Strategy) which would not only have been profitable over the course of time but would have been outperformed the index as well (and on top of that with a lesser maximum month-end drawdown).

But what would Michael’s finding that extreme RSI(2) readings currently become less common than in the past mean percentage-wise concerning a potential RSI(2) strategy ?

I conducted the same analysis for the time frames from 1950 until 1970, 1970 until 1990 and 1990 until 2009 for the S&P 500 with a 2-day RSI as I did for the static trading strategy with an RSI (2.5) for the SPY, meaning I determined the distribution of gains and losses on those sessions (the first and second day thereafter) immediately following a session when the S&P 500 RSI(2) closed between a lower and upper break point in order to evaluate an ideal exit point for a potential RSI(2) strategy (where the sum of all gains minus the sum of all profits = expectancy turns from positive to negative, which means -on average- the maximum possible gain on the upside would have been achieved).

For the three different time frames from 1950 to 1970, 1970 to 1990 and 1990 to 2009 , the following tables Table I to Table III show the respective distribution of profits and losses for the S&P 500 and the RSI(2) over the course of the then following 2 sessions, broken down by different ranges (potential upper breaking points) for the RSI (2), assumed one would’ve bought the S&P 500 on close of every session when the RSI (2) closed anywhere between the lower and the upper break point (no overlapping trades allowed). Please take a special look at the third last row (‘Profitability’).

Table I (1950 – 1970)

Regarding the time frame from 1950 until 1970

• all RSI(2) readings above 70 led -over the course of the then following two sessions- on total to profitable trades if one would have bought the S&P 500 on close of a session with an RSI(2) reading between the lower and upper break point,
• all RSI(2) readings above 70 show positive win/loss ratio, profitability and profit factor significantly above the at-any-time profit factor for the time frame from 1950 until 1970, and
• there were significantly more extreme RSI(2) readings between 90 and 100 than between 70 and 90 combined (the sum off all occurrences between 70 and 90),
• and RSI(2) readings between 90 and 100 show the highest profitability of all RSI(2) ranges.

So any RSI(2) strategy build upon a potential short on any RSI(2) reading above 90 would have been a clear receipt for disaster, and the upper break point would had to been set close to 100 in order to achieve maximum gains on the upside.

____________________________________

Table II (1970 – 1990)

Regarding the time frame from 1970 until 1990

• all RSI(2) readings above 70 led -over the course of the then following two sessions- on total to profitable trades if one would have bought the S&P 500 on close of a session with an RSI(2) reading between the lower and upper break point,
• all RSI(2) readings above 70 show positive win/loss ratio, profitability and a profit factor either slightly, partly significantly above the at-any-time profit factor for the time frame from 1970 until 1990, and
• the amount of extreme RSI(2) readings between 90 and 100 approximately equals the total amount of RSI(2) readings between 70 and 90 combined (the sum off all occurrences between 70 and 90),
• and RSI(2) readings between 80 and 90 show the highest profitability of all RSI(2) ranges.

So any RSI(2) strategy build upon a potential short on any RSI(2) reading above 90 would have been still a clear receipt for disaster, and the upper break point would had to been set close to 90 in order to achieve maximum gains on the upside.

But profitability and total amount of extreme RSI(2) were less extreme than the respective figures concerning the time frame from 1950 until 1970.

____________________________________

Table III (1990 – 2009)

Regarding the time frame from 1990 until 2009

• only RSI(2) readings between 75 and 80 led -over the course of the then following two sessions- on total to profitable trades if one would have bought the S&P 500 on close of a session with an RSI(2) reading between the lower and upper break point,
• only RSI(2) readings below 80 show positive win/loss ratios and a profit factor close to the at-any-time profit factor for the time frame from 1900 until 2009, and
• the amount of extreme RSI(2) readings between 90 and 100 approximately equals half of the total amount of RSI(2) readings between 70 and 90 combined (the sum off all occurrences between 70 and 90).

So any RSI(2) strategy build upon a potential short on any RSI(2) reading above 90 could have been profitable for the first time since 1950, and the upper break point would had to been set somewhere between 70 and 80 in order to achieve maximum gains on the upside

So I completely agree with Michaels bottom line (cit.) ‘… the markets are becoming more contrarian in the short-term. That means the market tends not to move in a single direction for as long, which means that the market tends to register less extreme readings on short-term indicators such as this one.‘

Successful trading,

Frank

P.s.: WordPress recently implemented a Twitter widget, so I’ll regularly make some intraday updates as well using Twitter (as I already did during the last couple of session, but unfortunately there seems to be a connectivity issue between WordPress and Twitter; hope that will be solved soon). If you’re interested in, please have a look at the blog during the trading session as well or subscribe directly to Twitter.

A few days ago Michael Stokes at MarketSci made an excellent post concerning RSI(2) readings, the changing frequency of extreme readings, its (the RSI’s) quality of forecast and efficiency of trading short-term mean-reversion in the markets over the course of time since 1950 (Extreme RSI(2) Readings Becoming Less Common, and he already covered the topic here and here).

Although I completely agree to his findings and conclusions which could be summarized as (cit.) ‘I don’t think this says anything about the effectiveness of strategies based on indicators such as RSI(2), but it does say that they might trigger a bit less over time.‘ and (cit.) ‘But I would be hesitant to trade the RSI(2) in the simple form I’ve described here as a static strategy.‘, I was incited by Bill Luby’s comments on Michael’s post concerning the usage of deviating break points (e.g. 95/5 and 98/2 instead of 90/10) and/or different moving averages (e.g. RSI (3) and RSI (4) instead of RSI (2)).

The RSI Relative Strenght Index was developed by J. Welles Wilder and introduced in 1978. The RSI (x days) oscillates between 0 and 100 and compares the magnitude of an assets (e.g. stock or index) recent gains to the magnitude of its recent losses, with low readings indicating oversold and high readings indicating overbought conditions.

My objective was to check if and to what extend the Relative Strength Index (not taking into account any additional indicator and/or condition, so in its pure form) could be utilized for a mechanical (and static) trading strategy, with investigations focused on the following questions:

• could an RSI based strategy stand the test of time without any adjustments (e.g. break points) and/or adaptions to the then current market conditions (e.g. bull/bear markets),
• its profitability year by year, in the long run, and its capability to deliver positive returns in bull and bear markets likewise,
• its capability to outperform the SPY as a tradable proxy for the S&P 500,
• the possibility to reduce the time in market in comparison to a buy and hold approach (which is always in the market) in order to reduce the risk of being hit by a potential ‘black swan’ event.

As an additional restriction I took into account and allowed for long trades only (the addition of potential short sales will probably be addressed in a future post), and no adjustements  (e.g. break points) are allowed for. No leverage is taken (no position sizing, money management and/or stops except the break points), but the strategy is always ‘all in’ (compounded returns, means any potential profits are always reinvested on the next trade in order to be comparable to a ‘buy and hold’ approach which is evenly always ‘all in’, assumed that no money is taken out). Due to the fact that this is a proof of concept/survey only, performance figures do not account for commissions, fees and slippage.

First of all I figured out that with respect to all the conditions listed above, it is not the 2-day or 3-day or 4-day RSI but the 2.5-day RSI which met those requirements best. Sounds surprisingly, but with respect to the computation of the RSI is makes no difference using 2.5 as a moving average instead of 2/3/4/… /x (even numbers for the number of sessions).

The next step was to evaluate the ideal lower and upper break points. From my perspective the optimal approach would be to determine the distribution of gains and losses (for the SPY) over the course of the then following first two days after a trade would have been entered at the lower break point, broken down by different RSI (2.5) ranges, and respectively at the upper break point in order to ride any upmove to its ideal extent (and before any potential profit would turn into a loss). For the time frame since 01/03/1993, the following tables (Table I for the upper break point,and Table II for the lower break point) show the respective distribution of profits and losses for the SPY over the course of the then following 2 session, broken down by different ranges for the SPY’ RSI (2.5), assumed one would’ve bought the SPY on close of every session when the RSI (2.5) closed anywhere between the lower and the upper break point (no overlapping trades allowed).

In order to maximize profits, the ideal exit is located between 67.5 and 72.5. That would present the ideal upper break point because any exit above (too late) or below (too early) would have reduced any already achieved or have missed any potential further gains due to the fact that the market’s tendency to reverse course significantly increased with an RSI (2.5) above 70. The same principle applies to the lower break point. The highest profitability (sum of all gains minus the sum of all losses, not the profit factor or anything else because the ‘opportunity factor’ plays a decisive role) would have been achieved with a lower break point of 18 (+179.06% during the then following two days if one woud have bought the SPY on close of every day when the RSI (2.5) closed below 18).

Strategy: Buy the SPX on close of a session when the RSI (2.5) closes below the lower break point (18); close the trade on close of a session when the RSI (2.5) closes above the upper break point (70).

(please note that trade performance figures were assigned to the date -and therefore regarded as achieved and realized- when the ‘buy’ was triggered, not on the date the trade was closed; that doesn’t make a difference concerning the total gains/lossed achieved, but due to the deviating distribution of gains/losses -not their total- the equity curve would look a bit different)

Table I

Table II

Table III shows the respective equity curve. Some additional stats:

• Maximum month-end drawdown: -11.22%
• % month positive: 74%
• Month outperformance SPY: 52.85%
• Time in market: 31.50%

So I completely agree with Michaels bottom line (cit.) ‘I do think RSI(2) has wings in today’s market. … But I would be hesitant to trade the RSI(2) in the simple form I’ve described here as a static strategy.‘, but not mainly concerning the point that it wouldn’t make sense to trade the RSI (2.5) as a static strategy (which would have been considerably profitable, less risky than a ‘buy and hold’ approach and would have almost always outperformed the SPY, but with hindsight only because we determined the ideal break points in 2009 and not in 1993) but in particular with respect to the fact that it will probably be possible to build a strategy around the RSI (2.5) in combination with one or more other indicator/conditions like Michael does in his State of the Market report which would be superior to any static RSI strategy alone.

Successful trading,

Frank

P.s.: WordPress recently implemented a Twitter widget, so I’ll regularly make some intraday updates as well using Twitter (as I already did during the last couple of session, but unfortunately there seems to be a connectivity issue between WordPress and Twitter; hope that will be solved soon). If you’re interested in, please have a look at the blog during the trading session as well or subscribe directly to Twitter.

The Market Shows it Pays to be a Contrarian

There are some often cited adages amongst investing commentators and traders, e.g. “Don’t try to catch a falling knife”, “The trend is your friend.”, “Nobody rings a bell at the market bottom.”, “Buy on strength” and “Sell into weakness” (the trend following approach), among others. But due to the fact that I’m one of those contrarians who see opportunity where others fear “disaster” (and vice versa), and being a ‘scientific sceptic’ (who regularly questions the reliability of those adages) I’m always eager to check if and to what extend those adages might prove true in the current (and past) investment cycle, and how to capitalize on any observations made during my investigations. To make a long story short: The current investment cycle  requires to -at least- question some adages and/or shows opportunity to add some fresh adages to the already long list.

Due to the recently often discussed short-term mean-reversion character of the markets, and in order to check if there is a way to increase the quality of forecast for the respective next session’s outcome (probabilities for a higher/lower open, higher high, lower low and/or higher/lower close), I took a deeper dive into the (trend following) “Buy on strength” and “Sell into weakness” adages.

For the time frame since 10/01/2007 (approximately the beginning of the current bear market), I checked for the SPY‘s outcome of the respective next session after the following setups had been triggered:

• at-any-time: Buy on close on every session regardless of any setup (no questions asked), sell on close the next session
• Survey I: SPY posted a higher high above the previous session’s high (as a proxy for intraday strength)
• Survey II: SPY posted a lower low below the previous session’s low (as a proxy for intraday weakness)
• Survey III: SPY DID NOT post a higher high above the previous session’s high (limited upside potential, as a proxy for some intraday weakness)
• Survey IV: SPY DID NOT post a lower low below the previous session’s low (limited downside potential, as a proxy for some intraday strength)

The following table shows -over the course of all 382 sessions since 10/01/2007- the SPY‘ behavior and the respective performance on those sessions immediately following the session when the respective setup was triggered. Odds (potential payout and expectancy, NOT the true chances that the event will occur) significantly above or significantly below their respective at-any-time odds (in this case +/-20.00%, but this percentage is up to everyone’s decision what may be regarded as ’significant above’ or ‘below’) are marked by a green (for a probable bullish or favorable outcome) and red (for a probable bearish or unfavorable outcome) background color. This should make it possible to catch on a glimpse if (any), where (e.g. EOD end-of-day change compared to the previous session’s close, or C-O close minus open for intraday strength/weakness) and to what extent (compared to historical odds) the respective setup out- or underperformed the market and if any tradable edge is provided.

(click on image to enlarge)

First impressions:

1. Survey (Setup) I -SPY posted a higher high than the previous session’s high- provides an unfavorable setup (concerning a bullish bias) due to the fact that the average profit on winning trades (+0.61%) is lower, and the average loss on loosing trades (-0.91%) is higher than the respective averaged at-any-time profits (+0,75%) and losses (-0,88%). In addition the profit factor of 0.66 (+104.85%/158,19%) is far worse than the even worse (because lower than 1) at-any-time profit factor of 0.85 (+285%/336,58%).
2. Survey (Setup) II -SPY posted a lower low than the previous session’s low- seems to provide a highly favorable setup (concerning a bullish bias) due to the fact that the average profit on winning trades (+0.90%) is higher, and the average loss on loosing trades (-0.78%) lower than the respective averaged at-any-time profits (+0,75%) and losses (-0,88%). In addition the profit factor of 1.15 (+185.99%/161,30%) is significantly higher than the respective at-any-time profit factor of 0.85 (+285%/336,58%). Concerning a potential mechanical trading system -buy the SPY on open, sell on close on a session following those sessions on which setup II had been triggered-  survey II would outperform a respective at-any-time trading system as well, means on average the market closed above it’s open more often and to a greater extent after setup II had been triggered than on an at-any-time session.
   
3. Survey (Setup) III -SPY DID NOT post a higher high than the previous session’s high- seems to provide a neutral setup in comparison to the respective at-any-time performance figures because it did neither out- nor underperform the market to any significant extent.
4. Survey (Setup) IV -SPY DID NOT post a lower low than the previous session’s low- seems to provide a highly unfavorable setup (concerning a bullish bias) -to say the least- due to the fact that the average profit on winning trades (+0.57%) is significantly lower, and the average loss on loosing trades (-0.99%) higher than the respective averaged at-any-time profits (+0,75%) and losses (-0,88%). In addition the profit factor of 0.57 (+99.17%/173,94%) is far worse than the even worse (because lower than 1) at-any-time profit factor of 0.85 (+285%/336,58%). The same applies accordingly concerning intraday trades following a potential mechanical trading system -buy the SPY on open, sell on close on those sessions following those sessions on which setup IV had been triggered-.

But how to capitalize on those observations ? A logical next step would be to simply capitalize on favorable and unfavorable (from a bullish perspective) setups by taking long trades only (on close of the day when the respective setup was triggered) concerning setup II -SPY posted a lower low than the previous session’s low-, and going short (in order to turn an unfavorable bullish setup into a favorable bearish setup) on close of those session when setup I -SPY posted a higher high than the previous session’s high- OR setup IV -SPY DID NOT post a lower low than the previous session’s low- were triggered.

Survey_ALL: “Buy the SPY on close of those sessions when the SPY had posted a lower low than the previous session’s low, and go short the SPY on close of those sessions when the SPY had posted a higher high than the previous session’s high OR the SPY had NOT posted a lower low than the previous session’s low; if both a long and a short signal had been triggered on the same day, take the buy signal only ; close the trade on close of the next session and enter into a new one.” (the last condition wouldn’t make sense in a real trading system, if no controversinal signal would be triggered you’d still hold on to your position) That is more or less the equivalent of  “Buy on weakness and sell on strength.” and contradicts the respective trend-following adage.

The following table shows -over the course of all 382 sessions since 10/01/2007 again- the SPY‘ behavior and the respective performance on those sessions immediately following the session when the respective setup was triggered, now including Survey_ALL in the last column and reflecting the potential performance figures of a combined trading system (Survey I up to IV are unchanged).

(click on image to enlarge)

Bottom line:

1. A mechanical trading system following Survey_ALL would have had (almost) always been in the market, there were only 2 sessions out of 382 when the setup would NOT had been triggered (05/22/2008 and 12/18/2008).
2. The system would have yielded a return of investment (not accounting for commissions, fees and slippage; not leveraged and not compounded) of 359.93%-260,47%=+99.47% compared to an at-any-time ROI of -51.42% (a ‘buy and hold’ approach would have yielded -47.08% since 10/01/2007, means the SPY has lost -47.08% during that time frame).

(click on image to enlarge)

Not bad for a pretty easy mechanical trading system with 197 winning trades and 184 loosing trades (but unfortunately with hindsight bias only), which from a win/loss ratio’s perspective only, would be nothing to write home about (but in this case not the probability of being right or wrong but the odds count). Up to now I haven’t calculated other important figures like max. drawdown and shape ratio.

So a new adage with respect to the current investment cycle might be: “If the market did not make a lower low today – it probably will tomorrow.” or “If the market posted a higher high today – it will probably post a lower low tomorrow.” (and therefore “Buy on strength” as well as “Sell on weakness” might not represent favorable guidelines at least for today’s markets)

But please keep in mind: These statistics are provided for informational and statistical purposes only, and there would be still a lot more work to do in order to check if this combined setup could be converted into a profitable trading system (e.g. if it would be profitable and to what extent in other times frames and markets as well).

Successful trading,

Frank