| Index | Recent Threads | Who's Online | User List | Search |
|
|
 |
Portfolio123 » List all forums » Forum: Screener » Thread: How does one find consistent performers? |
|
Total posts in this thread: 13
|
[Request a Feature]
[Post new Thread] [Add To My Favorites] [Watch this Thread] |
| Author |
|
|
KAR
Advanced Member
|
I am wondering if you may have some suggestions towards how to find stocks that move up consistently, with low volatility. I am experimenting with an interesting approach from Van Tharp, in his book, Safe Strategies for Financial Freedom. Here is my current version of screen based on his apprach. http://www.portfolio123.com/screen_summary.jsp?hist=no&screenid=9164 Using the rule: (close(0)-close(59))/ATR(60,1)>15 does a nice job of finding stocks that have moved up. But it gets dominated by ones that have sharp spikes up, the ones I want to avoid in this particular screening approach. So I added 2 more rules which split the 60 day period up in to 2 30 day periods: 5) close(0)-close(29)<.7*(close(29)-close(59)) 6) close(0)-close(29)>.3*(close(29)-close(59)) and that helped somewhat. But I was wondering if there is a better approach to find smoothly rising stocks. The adhoc addition I made, seems just that, a bit adhoc. Thank you in advance. Kurt ---------------------------------------- [Edit 1 times, last edit by KAR at May 27, 2007 11:21:42 AM] |
||
|
|
vladinvest
Advanced Member
|
Kurt, I had a ranking system that was trying to do that but, now trying to find it, it looks like I have not only left it unused but also deleted it. In general I was putting as separate same level nodes stock factors which were simply the MACD reading for let' say 15 ago, 30, ago, 45 days ago and so on. You can play with different intervals and different whole period depending on what you are interested in. It worked well. I encourage you to try it I will be willing to step in if you are not getting to sensible results soon. The main reason I left that system is I was looking not for just stock price performance (as capital appreciation) but ultimately total returns performance. To me – I was missing a lot in not accounting for stocks that pay good dividends and whose price graph performance will be degraded by the fact that it reflects company de-capitalization at dividend payout. In other post I have suggested brought up the need to have data about total returns performance in P123 in addition to price performance. That will help as identify companies in which at least till now investment returns have been of certain quality. The way I find such stocks now is by using morningstar.com stock screener. They have in their data base data about total return (capital appreciation plus any dividends) for all the stocks. I look for companies that have low reading on the “worst 3 months” indicator but with relatively high total return for the last let’s say 10 years. Not an ideal method but the closest to what I need. In ideal situation would should have return - risk measures (that include dividends) like Sharpe, Total Return, Standard Deviation, the way we have them for mutual funds for stocks too. As I have mentioned elsewhere part of my ingestible money I put in stocks who performed as good “strategy investment’ themselves. Not surprisingly such stock will turn out to be conglomerates, asset management companies, banks, investment banks, REITs, oil royalty trusts, utility or energy companies. This my other way to do some more timing and hedging. The stronger my feeling is the market is technically high the more of my funds will go to such investments, etc. Vlad ---------------------------------------- Mind is like a parachute - it only works when it is open. |
||
|
|
KAR
Advanced Member
|
Thanks Vlad, Your point about using Morningstar to get Total Return is subtle...and good. I had noticed in the past for Mutual Funds, that I needed to use Morningstar rather than Yahoo screener for that reason, but for stocks also, I hadn't realized it until you pointed that out. Thank you!!! Following up on your suggestion, I started exploring your suggestion of using Morningstar.com stock screener. I found that in Morningstar's universe, despite some obvious double counting here, the following: For Mkt Cap >300M there are 3558 stocks in total, any dividend For Mkt Cap >300M there are 773 stocks with dividend >3% For Mkt Cap >300M there are 129 stocks with dividend >6% For Mkt Cap >1000M there are 2222 stocks in total, any dividend For Mkt Cap >1000M there are 351 stocks with dividend >3% For Mkt Cap >1000M there are 69 stocks with dividend >6% So, your point is very important....a lot of stocks that are somewhat bigger are being disadvantaged in our SIM's and ranking systems due to not including total return. In thinking abou this more, it looks like you have identified an important part of reason P123 may not be as good at providing large advantage over indexes and mutual funds when picking Large Caps as it does for small and microcaps...eg. lack of Total Return information. I did see in P123 Factor Help, that there are a number of Stock factors, such as DIVLQ and DivPS5YAvg in P123 that might be helpful to mitigate this, but I havn't played with them so far. Vlad, I looked thru the P123 Feature Requests. I may have missed it, but I could not find a Feature Request for a factor giving Total Return. So, if indeed I missed it and if you want to put in a such a feature request, I will vote for it. ....So far so good... onward we go.... |
||
|
|
vladinvest
Advanced Member
|
Kurt, Absolutely. Considering dividends, and consequently total return, is of paramount importance especially when looking for investments that have proven consistent results and maybe not necessarily astronomic return percentage rates. It just so happens that stocks that have these qualities are predominantly ones with relatively high dividend rate. Kurt, I think there is a request for including total return in the performance readings for sims and port. In fact this is the request created by Marco: Include dividends in simulations and portfolios http://www.portfolio123.com/feature_request.jsp?view=open&cat=-1&featureReqID=79 I just created a request to include total return and risk metrics on it as stock factor. Feel free to vote for it. Add as Stock Factors Total Return (i.e. including dividends) Metrics http://www.portfolio123.com/feature_request.jsp?view=open&cat=-1&featureReqID=503 ---------------------------------------- Mind is like a parachute - it only works when it is open. |
||
|
|
KAR
Advanced Member
|
Thank you very much Vlad. I voted for your Feature Request. Kurt |
||
|
|
dcnelson
Advanced Member
|
I voted for it too. You are pointing out something very important. |
||
|
|
dwpeters
Advanced Member
|
Hi, I have been using efficiency as described by Tharp, and the Turtles before him in a screen in Stockcharts.com. To try to avoid the spikes I calculated the efficiency for each month and took an average weighted towards the most recent months. It still found stocks that spike and I avoided those based on a review of the charts, but it also found what you are looking for, stocks that have moved up very consistently. Unfortunately they don't always continue to do so. I feel that this method seems to work better with ETFs. I tested some rankings a while ago. This one was the best variation: http://www.portfolio123.com/rank_details2.jsp?rankid=36670 In my tests, the rankings based on this method tend to have a pretty nice profit curve but a disturbing, and often significant drop off at the very top 1 or 2 percent ranking. I tried sims that would buy and hold only while the rank was 96-98 or so but the performance just wasn't what I was looking for. There is a new Sortino function that may be useful in identifying consistent performers. Don |
||
|
|
vladinvest
Advanced Member
|
Don, I agree past performance, especially with stocks, should not be taken as an immediate indicator of future performance. I personallly, after whatever screeing I use to get to such stocks, look for companies that by the nature of their business indicate to me they may be able to demonstrate similar performance ahead. Examples are as discussed previously, asset mamanegement companies, financial companies of cetrain type, companies in which the price upward movement has always been well supported by a upward movement in shareholder's equity (i.e. relatively stable P2B ratio). I agree such data will be very important fopr research of ETFs and I will add CEFs. Again P123 can use this as an advantage even to morningstar since they do not have the advanced screen tools available for funds and stocks for ETFs and CEFs. Vlad ---------------------------------------- Mind is like a parachute - it only works when it is open. |
||
|
|
DennyHalwes
Advanced Member UNITED STATES Joined: Apr 28, 2004 Posts: 1532 Status: Offline |
About 1 year ago I set up a Ranking System that used an approach for consistent total returns in the Growth composite node. Since we don’t have total return except for the last 52 weeks the approach I used was to estimate total return by calculating the price gain for past periods and then add the dividends for that period. Most companies have a fairly steady quarterly dividend so I assumed that it was equally distributed in the different periods. This will obviously cause some errors, but was as close as I could get with the current tools. Below are the functions that I used in my growth composite node. Pr52W%Chg - Pr26W%Chg + Yield /2, This gives the 6 month total return 6 months ago. Pr26W%Chg - Pr13W%Chg + Yield /4, This gives the 1 Quarter total return 1 quarter ago. Pr13W%Chg - Pr4W%Chg + (((DivLQ * 0.666) / Close(20)) * 100), This gives 2 months total return 1 month ago. Pr4W%Chg + (((DivLQ * 0.333) / Close(0)) * 100), This gives the last months total return. My original Ranking System is private; however I set up a public system with just the Growth composite node and made it public so you can see how it works. I found that by setting the weighting of the oldest period highest, and the most recent period the lowest it gave me the best performance. http://www.portfolio123.com/rank_details2.jsp?rankid=45159 There are some interesting problems with adding the yield in this way. Since the yield is calculated by adding the dividends for 1 year and dividing by the last price, this can cause the total return to be calculated very high with stocks that previously paid a good dividend but later fall drastically in price. For example, NEWCQ had quarterly dividends of $1.70 to $1.90 for the previous couple of years with a price of $45.96 one year ago. However, it declared bankruptcy in February this year. The price then fell to $3.00 in March and is now trading at $0.46. The 3 dividends that were paid in the last year add to $5.50. This calculates to a yield = ($5.50 / $0.46 -1) * 100 = 1095%! (There obviously is a problem with the way yield is calculated in cases like this). The gain for the last year is ($0.46 / $45.96 -1) * 100 = -99%. So the total return is 1095% - 99% = 996%. An amazingly high total return considering that you lost 99% of your money! Another reason I hate dealing with these Q stocks! Denny  ---------------------------------------- "The significant problems we face cannot be solved at the same level of thinking that we were at when we created them". Albert Einstein |
||
|
|
KAR
Advanced Member
|
Hi Don, Thank you for sharing your ranking system. I need to spend some time playing with it. But in meantime, could I ask you, what part of your ranking system is what you called efficiency. Is it the part listed in composite node: TechRank, in your ranking system? Is that the formula for efficiency? In your ranking system, that looks like it is calculating on a 50 day interval. Also, your formula takes ratio of prices 50 days apart. The formula I put up above at start of this thread came from from Van Tharp a but used difference in Close, normalized to ATR, not ratio of Close. I am not sure what is differnce between these 2 approaches. You mentioned in your note that you computed on a month by month basis, but I did not see that in your ranking system, so I am thinking you did off-line perhaps? I looked in the 2 Van Tharp books I have, but could not locate discussion of efficiency or find it in index. Is it called something else by Van Tharp and Turtles? I have Safe Strategies and Trade your Way to Financial Freedom. Might you be able to indicate the page #'s. Thank you. I seem to recall that Steve (Stitts) found Sortino was quite useful in his Industrial SIM. Ah, here it is. Steve posted it here: http://www.portfolio123.com/mvnforum/viewthread?thread=2476&offset=0#10559 Steve's ranking system with Sortino is: Stitts High Liquidity Industrials Rev 1 Hi Denny, Thank you for your post. Now I see how one can use DivLQ. Very nice approach you developed. Thank you for sharing this approach incorporating yield. I think this will be very useful and go a long way towards approaching Total Return ultimate goal. I see what you mean about the issue you point out about using Yield. Today I look at your example bankrupt stock, NEWCQ, and its yield is even higher at 1,727.27%. See it here: http://www.portfolio123.com/stockratios.jsp?ticker_uid=NEWCQ+&show=Go Prompted by your post, I did some reading about Yield on the web. I learned that what you are warning about can be referred to as the Dividend Yield Trap. Here is a short descritpion of it: http://stocks.about.com/od/evaluatingstocks/a/Dividyiel021005.htm I thought of trying to avoid these kind of very high yield stocks. So I added a Boolean Filter to your Ranking System with a Boolean Rule of Yield<10%. Here is that modified ranking system: http://www.portfolio123.com/rank_details2.jsp?rankid=45418 That seemed to work to keep out the really crazy yields. Of course, there is the question of where to set the upper limit of acceptable yield to allow thru.... I also tried an alternative of using your ranking system without a Filter and made a sim with buy rule YIELD >10. Here is that SIM: http://www.portfolio123.com/port_summary.jsp?portid=277209 The curious thing about this SIM is it bought 2 stocks which currently have a Yield of <10. For one of them, DSX, this makes sense, since the price of the stock has increased since it was bought, so its yield has proportionally dropped. But the other stock the SIM holds, PTNR has almost same price, $17.14/share now as when it was bought 2 weeks ago on 5/14 at $17.11/share. (Very handy to look up these prices in P123) . And yet yield is now listed as 3.03%, which is significantly less than 10% of Buy Rule requirement of Yield >10%. That was puzzling to me, so I checked Yahoo Finance, and it lists the Forward Annual Dividend Yield as 2.7%. I am not sure what is used to calculate current yield in P123 of 3.03% and how that differs from Forward Annual Dividend Yield of Yahoo. Yahoo does list a dividend of 0.129 on June 1. But that does not seem to be enough to explain how the Yield is now ~3%, which is so much less than 10% and yet the SIM bought it 2 weeks ago with Buy Rule of Yield >10%. So this still seems puzzling to me. More generally,, it seems to me there should be some kind of more sophisticated way of calculating yield that does a better job of not being skewed by very large drop in price. I am thinking of something analagous to how IRR does a better job of calculating return except to handle yield. This is not my area of expertise, so I don't know if such a more robust yield formula exists. Kurt ---------------------------------------- [Edit 2 times, last edit by KAR at Jun 4, 2007 3:15:36 AM] |
||
|
|
[Show Thread Printable Version] [Post new Thread] |
Re: How does one find consistent performers?
