PLEASE REMOVE THE CONSTRAINTS ON TECHNICAL INDICATORS!

Research
1

I chimed in with David (Primus) in a different thread in which David was requesting that you open up the limitations on the parameters of technical indicators such as simple moving averages, RSI, and ATR. However, I think this issue is exceptionally important and justifies its own thread with a more targeted headline.

A HUMBLE PLEA TO MARCO and the P123 TEAM: [color=red]PLEASE REMOVE / OPEN-UP THE CONSTRAINTS ON THE P123 TECHNICAL INDICATORS![/color]

BUSINESS JUSTIFICATION: Opening up these parameters could help to cement P123’s position as a market leader for strategists in the fast-growing Exchange Traded Fund (ETF) industry (which is now growing by 25% per year - with more than $1 TRILLION in funds added to ETFs each year) – not to mention keeping long-term stock investors like David happy with the P123 product.

Reasoning: At present, there are a limited number of sites that offer the ability to design, backtest, and implement ETF-based investment strategies, but more are being launched each year. At this time, P123 has an advantage in the capabilities it offers to ETF-strategy designers, but with the unnecessary limits imposed on parameters for technical indicators and other legacy factors in P123, many ETF-strategy designers who rely on technicals will go elsewhere.

Here’s Why: Because ETFs do not have the fundamental ratios that are inherent to individual stocks, ETF strategists frequently rely on technical indicators for timing signals to reduce market exposure or change positions. But P123 has imposed unnecessarily short limits on technical indicators such as Moving Averages, EMA, MACD, CCI, Chaikin, OBV, RSI, Bollinger Bands, Stochastics, and more…

[color=red]P123’s BOLLINGER BANDS are limited to just 100 days! [/color]

[color=red]P123’s RSI is limited to 200 days![/color]

[color=red]P123’s STOCHASTIC OSCILLATOR is limited to just 50 days![/color]

StockCharts allows 600 days for each of these indicators! But no one is asking you to compete with StockCharts or any other sites that specialize in technical indicators –– just to increase the existing limitations on your existing technical indicators!

Many of the P123 settings are only appropriate for DAY-TRADERS, not the longer-term investors and investment professionals that P123 attracts. I would like to promote the P123 affiliate program and send traffic your way, but I haven’t because of these limitations, which stop me from teaching others how to implement many of my best ETF-based strategies using the P123 platform.

If you cannot open up these parameters, I’m sure many P123 members would appreciate if someone could please explain why these very short limits remain in place – in some cases, 15 years after they were first added to the P123 factor library. I know that many P123 members have tried the technical indicators with poor results – but that’s probably because they are set-up for used by short-term DAY TRADERS!

Opening up the limits on technical indicators for investment (LONGER-TERM) purposes would probably surprise every P123 member with what is possible. I will even offer to write and publish for P123 members a comprehensive, detailed, chart-filled report that shows how the longer-term technical indicators can be put to practical use as accurate components of a stock or ETF-investment system! When you combine stock fundamentals with technical, price-based indicators, the results can be astounding (and far more reliable)!

If there is a good, practical reason why these parameters must be so limited, it would be very valuable to know. If not, why not increase them ASAP? It seems like it would be a 15-20 minute job to set those numbers higher within the P123 admin interface. Changing a “50” => to “500” or “1,000,” for example. I’ll be glad to provide free advice, after consulting with other key P123 members for reasonable limitations for each indicator, while still making them practical for investment purposes. Just let me know!

Thank you,

Chris – and many other P123 members who would love to spend all our money with P123, not be forced to supplement our P123 work with less limited technical indicators elsewhere

2

I second that. This would be a massive selling point for me

3

We always place limits to ensure that power users or bad parameters don’t affect the rest. Then we adjust accordingly.

What are reasonable limits ? Are these OK ?

RSI from 200 - > 500
BBUpper BBLower from 100 → 300
StockK StochD from 50 → 200

Also RSI has a smoothing period of 100. So if you choose 200 as the period you need 300 bars otherwise it will produce N/A. Should smoothing period be more forgiving?

The above is very easy and we can release this week. Just chime in if the new limits are not ok.

I think though that limits are not the real issue. We need the ability to use choose daily or weekly series by adding an additional series parameter #weekly. So for example using RSI(100,0,#weekly) would use about 500 bars. I’m not too familiar with how Technical Analysts build weekly series. Are they calendar weeks or 5 bars? Does an RSI using 100 weeks basically give the same number as RSI using 500 days ?

It’s been a while since I heard requests for weekly series so probably not all that important.

Thanks

4

Marco,

I’m totally agree with Chris. It would be awesome.
In addition to #weekly, I’d let the user decide - something like: RSI(iterations[increment,offset,series])

Thanks

5

This is a brilliant point, though I think the suggested bump up in lookback periods couldn’t hurt.

Technical indicators in EasyLanguage are built with barlengths of varying discrete intervals (second, minute, hourly, daily, weekly, monthly, yearly, etc) I.e., they use discrete time interval vice straight bars. Moreover, I think that some generic logic to help us deal with discrete weeks/months/years would be more helpful, especially because bar intervals can already be defined for the loop functions (which are themselves useful for building custom technical indicators).

For this case, I wouldn’t think the result should matter much. But I think that this is inherently a qualitative distinction that should really be left up to the user. For example, I might think it’s important to distinguish between the volatility of weekly bars and estimated weekly volatility of daily bars. But it is really that important? ¯_(ツ)_/¯

If I could, I’d like to revisit a question from the related post on analytics from custom time series. The question: Could you guys allow us to access the value of a custom series using a key-value pair? For example, we can currently access Portfolio holdings using the portfolio id# (vice a quoted string). Could this functionality be extended for the [font=courier new]GetSeries(“ticker/series”)[/font] function?

6

Hi Marco,

Thank you so much for your generous and constructive response to my original post in this thread. You have made my day – better than that, you’ve made my era! haha

While I think it would be very advantageous to have longer limits on the technical indicators in daily time frames, your proposal for adding weekly price series is ‘windfall profits’ to my request. I had considered bringing up weekly price series in my post but didn’t want to push it too far. In the 14-years I have been a member of P123, from time-to-time I have posted about technical indicators and have always advocated that my fellow members use weekly charts over daily charts.

The reason is that daily charts/price points are rife with meaningless noise caused by investors testing to see how far they can push the upper and lower limits of price in an equity or ETF. However, the significance of an asset’s price at the close of the week on Friday is always a far more accurate signal of future prices, when put in context with previous price points.

Not only do the P123 Sims and Ports naturally rebalance/reconstitute using fundamentals on weekends, but having weekly price points for technical and momentum factors can make the accuracy of fundamentals far more improved. Technicals based on ‘investment’ time periods can be used to identify the optimum entry and exit points for a stock – and you can eliminate stocks you might have bought if technicals were not there to show you that its price is overextended (using Bollinger Bands, stochastics, or MACD) and is destined to decline first.

Here are my individual responses to questions in your post:

Although those are much better than current settings, I’m wondering if you can open them up a bit more during the interim – at least until the addition of #weekly price series is being implemented? For example, is it possible to use a minimum of 600-days on all technicals? The reason I propose 600 periods is because StockCharts has a maximum of 600 days/weeks/months/etc. While that may seem like a large number for P123 members who have not used technicals, there are several indicators where a parameter this high would be appropriate…

For example, I can use MACD on a monthly chart on the StockCharts.com website and get very accurate IN-MARKET or OUT-OF-MARKET signals that are appropriate for a long-term investor. In fact, on StockCharts, the MACD indicator can be applied to a monthly chart of a full century (100 years) and an investor would have accurate signals of when to exit the market and avoid significant downturns that anticipate recessions and are 100% accurate.

However, since P123 only works with daily prices (at this time), to make the standard settings for MACDD(12,26,9,#Bench) emulate monthly signals, we would have to adjust those settings to accommodate the average of 20 trading days in a month. This would convert to MACDD(240,520,180,#Bench) on a daily chart. However, with P123, our MACDD settings are limited to periods of 200 or less! Hence, I can’t use MACDD in P123, and I’m forced to rely on other websites.

If you can increase the settings to 600 bars, I’m sure no one will ever complain about the daily limits being too constrained! If you would like made to propose the maximums that might be applicable for each technical indicator, just let me know. I’ll put together a list. Attention P123 members: Let me know what you would like to see!

[b]While you asked about settings for just three indicators (RSI, BBUpper/BBLower, StochK), I hope you will open up the bar limitations on all of the technical indicators.

This would include: SMA, EMA, MACD, CCI, Chaikin, OBV, RSI, ATR, all Bollinger Bands, and all Stochastics.[/b] If P123 members have any other recommendations, I hope you will add to this list.

Yes, it does. For my current work with ETFs, I strictly use weekly prices in technical indicators, which are established at the close each Friday. To try to emulate this in P123 I do exactly as you propose; 100 weeks equals 500 days.

Weekly technical charts always use calendar weeks. The closing price of each security on Fridays is used for establishing the weekly price of each security or index.

Marco, thank you for the unexpectedly generous response to a long-time customer (14 years!) request. I was hoping you would respond favorably to the request to open up the bars on various technical indicators, but the proposal of adding Weekly Price Series is an unexpected gift!

  • Chris
7

I have asked for #weekly and #monthly. This is standard fare in the advanced scan workbench in stockcharts. I am particularly interested in rsi. So max could be 14 training months. This is defined here:

https://stockcharts.com/school/doku.php?st=rsi&id=chart_school:technical_indicators:relative_strength_index_rsi

Note that rsi (14,0,#monthly) is different than rsi (14x30,0) ie. rsi(420,0) as per the formula in the link above

Thanks for considering this multiple time frame analysis.

8

Rohit (calif2),

There are an average of about 20 trading days (aka, ‘bars’) – in a calendar month. So you would actually use RSI(280) if you wanted to emulate signals from a monthly RSI(14) – but on a daily chart. If you wanted to emulate a weekly RSI(14) on a daily chart, you would use RSI(70) – i.e., 14 x 5 bars in a week = 70.

/Chris

9

Chris,
Per my post and link I referenced you might see (and I believe) that rsi(280) and rsi(70) is not an equivalent to #monthly or #weekly. Correct me if I am wrong but don’t think the math works out like that for this indicator based on the formula for rsi calculation. A quick spot check on stockcharts would support my interpretation.
Thanks.
Ro

10

Yes, you are right Ro (calif2), extrapolating those settings with the RSI oscillator and other oscillators may not work appropriately. However, they do work for a plethora of other technical indicators, such as Simple Moving Averages (SMA), Exponential Moving Averages (EMA), Bollinger Bands, MACD, CCI, and many, many more.

It will be fantastic to have the flexibility that comes with setting the indicators to ‘investment’ time periods – rather than the ‘trader’ time periods where they currently stand.

/Chris

11

Chris, I totally agree and hope that weekly and monthly indicators are added since the source data is all there. Thx.

12

Ro (calif2) and all,

I was checking this thread today and re-read my last response. I don’t think I answered the question about the proper RSI settings in the best possible way. It may have left some users with the impression that we can’t use many technicals as I said in my first post, and I would like to amend my answer.

Here’s what I said in my last response:

In fact, there is a nearly-perfect 5:1 and 20:1 extrapolation that is possible for most price-based technical factors (and most fundamental factors too, for that matter) when going from effective weekly or monthly settings to daily chart settings.

Marco has said that he could add a weekly pricing option (I don’t think he has mentioned monthly settings, but you are right, those prices are already available in the system also. Weekly prices would be ABSOLUTELY FANTASTIC, and monthly prices are also desirable! However, I know some projects can take longer than expected, so it’s excellent that Marco said that by the end of this week he would incorporate the expansion of the setting limitations on daily, price-based, technical factors and formulas.

I’d like to give an example of how accurate the longer-term weekly and monthly settings on technical indicators can be extrapolated and used on a daily chart. There also some oscillators – such as RSI – where this extrapolation doesn’t work as expected. That’s because of the formula used to establish those settings are not easily transposed when going from one timeframe to another. However, they will still work with longer-term, ‘investment-suitable’ settings.

Here is an example of how the MACD indicator can be used in both monthly and daily settings. (It can also be used in a chart with weekly prices.)

I applied the standard MACD settings that are typically used by short-term traders, i.e., MACD(12,26,9) to a MONTHLY chart of the S&P 500 and was surprised to find that these settings were very accurate in identifying timely entries into the index and exit out of the index just before a significant downturn. These settings will not capture smaller corrections, such as the 2011 -19% correction, but they do accurately identify the deterioration of momentum well before the index itself begins turning down. If you used these MACD settings with a monthly chart of the S&P 500 from1993 to present, when the indicator crosses its signal line, you would have precise demarcations to exit the market or reenter.

Note that I also included a 10-month moving average in the upper window of this chart, which investors can use as a confirming signal of the MACD indicator. Both the 10-month SMA and the MACD indicator can have inappropriate, whipsaw signals that occur, caused by severe corrections. The SMA(10) could be used to help determine if the MACD signal is accurate. For example, if the 10-month moving average is still climbing, you may wish to put off selling a stock or ETF. You could also require both signals for an even more accurate set of indicators.

MONTHLY CHART WITH MACDD INDICATOR

Next, we can try to apply those same settings to a daily chart. Of course, there are many more sessions in a daily stock chart than there is a monthly chart. In fact, there is an average of 20 market sessions (days) in the average month. So, if we multiply each of those settings (12, 26, 9) times 20, we get settings of (240,520,180). I have also converted the 10-month moving average to the 200-day moving average (10×20=200) in the upper window this chart.

DAILY CHART WITH MACDD INDICATOR

Does it work? You bet it does! The daily chart with these expanded settings is just as accurate as the monthly chart. It may, in fact, be more accurate because it’s possible you will see the signal occur earlier on a daily chart than you would at the end of each month.

Unfortunately, we have never been able to use settings higher than 200 with the MACD indicator on the P123 platform. However, that all may change soon if Marco implements an increase of the setting parameters in all the technical indicators. I have suggested 600 days as an appropriate, across-the-board setting for the daily sessions of technicals. After all, I needed 520 days in this MACD example, and users may find higher/different settings are even more accurate for investing purposes, and different indicators may inherently require higher settings.

If Marco implements weekly charts as he has also suggested, that will be all the better because stock prices are far more accurate when based on weekly signals than they are on daily signals. IMO, this could apply to virtually every factor or formula we might use in P123 (unless you are a very short-term trader).

I am keeping my fingers crossed that we will see technical indicators with higher parameters possible by this weekend. And the ability to use weekly charts (which are based on the price at Friday’s close) will be exceptionally welcome!

– Chris

13

Chris,

You are correct. The calculation of rsi on multiple timeframes is not a simple math problem though it is still math. As far as MACD I will study your example more closely and do a spot-check in comparison to the signals on stockcharts to see if they correlate. Elder’s impulse system is but one example of multiple time frame analysis also available on stockcharts but they have zero useful backtesting. I do think there is merit to making it possible on p123. Specifically rsi has some interesting possibilities if it can be made to happen. Thanks for the thoughts.

Ro

14

Marco - there is a fallacy in this argument. Power users WILL find a workaround and the workaround will usually be more resource-intensive than if you had given the power user the desired functionality to begin with.

There is a bigger issue at stake here, however. The bigger issue is the lack of universality amongst functions. So not only is there inconsistency between functions for time period but inconsistency in the actual calling parameters. Is there a reason why some function calls allow GetSeries and some don’t? The problem gets worse when one ventures into Hedge/Market Timing. Some GetSeries are allowed and some are not, and it is different than for Buy/Sell rules. And I’m sure I will find that it is different again for custom series.

I am speaking here not as a “power user” but someone who is concerned about the complexity of the website that is already difficult to use for regular folk, even without the inconsistencies of function parameters throughout the site. Standardization of function call parameters and making the same GetSeries available throughout the site is certainly one area that P123 could improve upon and make the site slightly easier to use for the average member.

Thanks
Steve

15

Marco,

I agree 100% with Steve (‘SectorInspector’) about the inconsistency of the allowed time-frame parameters and inconsistency of ‘GetSeries’ and many other commands. I would like to offer a little bit of unsolicited constructive criticism – from someone who, in a former life at Deloitte, was a consultant to large, midsize, and small companies on business strategy and market position.

While - for me at least - Portfolio123 is the absolute best platform for my objectives, there will be competitors in the future that will try to develop a superior product. It is a consistent trait of a free-enterprise, capitalistic system that competition will always arise to challenge the position of a market leader. There are already several that are nipping at the heels of P123.

If Portfolio123 will make a quality-control effort to standardize across the platform (the issues that Steve remarked about), as well as expand the time-frames available on technical indicators (the subject of this post), Portfolio123 will take a big step toward cementing their position at the top.

While there are probably many projects underway that will add new features and capabilities to Portfolio123 (there always are!), a quality-control project that will review and standardize each function and command across all the site, in the buy-sell rules, ranking systems, screener, custom universes, etc., will go a long way in making sure you keep your core customers far into the future.

It is HUGE (for me, at least) that you are going to expand the allowed time frames for technical functions, and it would be FANTASTIC if you can offer the ability to run the functions and formulas on weekly prices (close of market on Friday). These easy-to-implement improvements based on data you already have will make a huge difference in my satisfaction with the platform, and I’m sure it will help many other strategy designers. Please keep up the good work! …And thank you for the continuous effort you make to satisfy the myriad needs of your customers.

Chris

16

Hi Marco & Team,

I fully agree with Chris‘s plea and his suggestions, this would definitely be a very valuable improvement for the P123 community.

Many thanks for your great work and for consistently improving this platform!

Andreas

17

I am wondering what the status is on this, Marco?

You said you thought the parameters on technicals could be expanded by the end of last week (10/06/2018), but so far I don’t think any changes have been implemented.

Can you let us know where things stand in this? Much appreciated!

/Chris

18

Hopefully today. It wasn’t as simple as I first thought.

19

Most TA limits are 600 bars now. Thanks for reporting this.

20

Marco, are you thinking at all about adding #weekly or #monthly indicators, with special interest to rsi ? Thanks