RELATED TOPICS --> | The Piotroski Strategy | The Piotroski Blog | Piotroski JSE Backtests |

Since October 2008, after we had documented our first Piotroski medium and long term back-tests on the JSE, PowerStocks Research have been running and maintaining a live "active" Piotroski investing strategy. This report summarises the results of what we have achieved during the year so that subscribers can mirror the strategy successfully for themselves using our JSE scans.

The "Active" Strategy ejects shares from the portfolio when the company announces results which give them F scores below 8. This evaluation happens twice a year when companies announce their Finals and their Interim results. The "Passive" strategy (depicted in our JSE backtests) sells the shares after 12 months regardless of results that are announced during the holding period. The active strategy holds shares for shorter periods (average 180 days) than the Passive strategy and obviously the passive strategy attracts less tax because of the 12 month holding periods.

ACTIVE STRATEGY
The Piotroski "Active" Strategy works as follows:

1. Every Monday we scan the JSE for Piotroski Candidates who have F-scores of 8 or 9
2. The strategy BUYS shares with F>=8 and who have Price/Book ratios of less than 1
3. If candidates have market capitalisation > R1Bn, we relax the Price/Book criterion to 1.2
4. The strategy SELLS a share its F score drops below 8 after results are announced.

So we have 4 simple mechanical rules for acquisition and selling with which we built and ran a Piotroski Active Model Portfolio, commencing right in the depths of the Great 2008 crash. One point to note regarding the scans - they used the last published financials of the companies concerned, regardless of if they were interims or finals, to always ensure freshest fundamental data is being used in the screens. Piotroski compares current financials to the previous financials and since we are using finals or interims, we compare them to the previous corresponding finals/interims when computing our F-scores. This also means shares are evaluated twice a year for inclusion/exclusion (hence the "Active" version of the traditional "hold for 12 months" strategy in the backtests.)

TOTAL RETURN PERFORMANCE
The daily calculated TRI of the Active Piotroski Model Portfolio appears below. It is inclusive of brokerage charges, interest earned on cash and dividends paid by shares in custody.



Since six months ago we commented on the peculiar nature of this strategies' TRI curve versus all the other strategies we track. It has a wiggly look to it, with occasional big spurts upward, but maintains an incredibly straight upward march versus the JSE's gyrations. Since the outset it more or less doubled the ALSH buy and hold returns. Many of the shares are smaller cap and what we see happening is a share releases results, we see value (low price/book) and great financials (F score) and acquire the shares and then soon after the market catches on and the share shoots up to get re-rated and then sits there (as they are not massively liquid stocks). As we can see the strategy outperformed an ALSH buy-and-hold strategy rather convincingly.

STRATEGY WIN/LOSS RATIO AND EXPECTANCY
The strategy made 31 trades since Oct 2008. For the purposes of this report we have ignored the last two recent trades as they are less than 30 days old and won't be used in the calculations. So this leaves 29 trades with 20 winners (growth greater than zero) and 9 losers (growth less than zero). We conclude a 69% win rate (or a 2.22 win/loss ratio)  which is slightly higher than when we did this exersise 6 months into the program.



Of more importance is the average growth of the winners and losers. Winning trades pocketed on average a 55.2% gain whilst losing trades only resulted in 17.1 average losses. So not only do winners outnumber losers 2-to-1 but winners provided returns that exceeded losses by over 3-to-1. This gives a TRADE EXPECTANCY of (69% x 55.2%) + (31% x -17%) = 32.6%. That is incredible.

INDIVIDUAL SHARE PERFORMANCE
The shares that are currently still been held in the portfolio are shown below. DAY is how many days they have been sitting in the model portfolio since acquisition and DIV is dividends received.



Shares that have been disposed from the portfolio over time are displayed below:


Winning trades averaged 55.2% with a standard deviation of 42.5% giving a Sharpe of 1.29.
Losing trades averaged -17.1% with a standard deviation of 9.0% giving a Sharpe of -1.88.

ACTIVE OR PASSIVE?
Which method is better - active or passive? Sell when the results show a lowered F score or just hang on for 12 months before selling? Assuming the shares currently being held in the portfolio would be the same for both strategies (since they have all been held for less than a year) we decided to make a comparison of returns gained from the disposed shares to those if we still had them in our possession as at 3 Nov 2009. The results are below:



Clearly we sold SLO 34 days too early! Whilst the above exercise ignores interest earned for cash by the Active strategy after it has disposed of shares and the higher dividends possibly paid by shares held in the passive strategy, it seems on the surface that a "buy and hold" Piotroski strategy provides higher expectancy in terms of 42.65% per trade versus the active strategies 37.96 per trade (for the 20 trades under consideration). Of course the passive strategy would only be liable for capital gains tax as opposed to the active strategy that is likely to attract personal income tax for higher frequency of trading (lower holding period average of 186 days versus 287 days). This shows how "time in the market" works for you, by raising your win rate and overall expectancy.

OBSERVATIONS
While win rates seem high, there is no doubt that making purchases in trough periods significantly enhances returns and this strategy benefited by commencing purchases on 20th Nov when TroughFinder had signaled a major market bottom.

Large cap shares that qualify for model portfolio inclusion perform very well.

Small to micro-cap and illiquid shares that have not shown price improvement for some time seem to perform less well (SPS etc.) and should probably be avoided.

Performance could probably be enhanced by reducing losing trade losses through implementation of stop-losses. The model portfolio did not attempt to limit losses and held on until a shares exclusion through announcement of results that decreased its F score.

As the bull market progressed, it started getting harder and harder to find qualifying shares. The Price/Book ratio criterion could probably get relaxed to 1.2 for small cap shares and 1.5 for large cap shares.

Additionally as the recession progressed, more and more shares in the portfolio fell out as results were announced. This is likely to improve as the economy returns and better financial reports start getting published.

Subscribers could probably lower the loss rate by conducting further fundamental and external inspection of small cap candidates to detect problems with the share the screen might not detect.

The strategy produces candidates that offer high probability of success for 6-12 month holding periods. Forcing investors to not overpay with the strict Price/Book < 1 criteria certainly helps in this regard but this might need to be relaxed as the economy recovers and the market becomes more expensive.

It seems a 12 month holding period as opposed to disposing of shares when their results drop their scores, offers better win rates and higher overall expectancy (returns per trade).

Subscribers can run their own Active Piotroski portfolios by following the Piotroski Blog where candidate lists are published each Monday.

RELATED TOPICS --> | The Piotroski Strategy | The Piotroski Blog | Piotroski JSE Backtests
|
 
Make a Free Website with Yola.