Wednesday, December 31, 2008

EURUSD Chart with commentaries

A EURUSD commented chart has been posted on MarketSnapshot :


(NB 1.15PM UK: the 60mins target predicted about 4hours ago has been reached already)

Tuesday, December 23, 2008

Volume chart with commentary.


Here is a commented chart on a ES volume chart. Again no change in settings whatsoever, but since the high volatility environment kind of "compresses" time, a "time-less" chart is ideal.

Good trading to all

(posted 3AM Eastern)

Thursday, December 18, 2008

DLL is now protected against hacking

Each DLL is now 'crafted' manually to make it a bit more difficult for hackers to crack. Nothing short of full decompiling to assembler is needed now. I really doubt a hex reader or debugger can do it. If you manage to crack it, at least please let me know :)

Wednesday, December 10, 2008

DLL now adapts calculation depth to CPU speed

One size doesn't quite fit all so I have added a routine which sets a Population Multiplier in our Genetic Algorithm according to CPU speed. Of course, CPUs nowadays are multi-core, but since it is not clear how TradeStation utilizes processing power, it was decided to keep this simple.

if ( nCPUSpeed > 3000 ) nPopMult = 12;
else if ( nCPUSpeed > 2000 ) nPopMult = 9;
else if ( nCPUSpeed > 1500 ) nPopMult = 6;
else nPopMult = 3;

In other words, if the CPU is over 3GHz (3000) it will look 4 times deeper than an old or low entry processor running under 1.5GHz.

No other functionalities added in this release, except for Swing2 which calculates Fib automatically (hence eliminating synch issues). The RetrievePivotInfo indicator is left for tutorial purposes.


DLL version is 1.21 (Dec 10th 2008)

Friday, December 5, 2008

Entropy indicator : minor update


The Entropy indicator has been marginally recoded to better reflect "realtimeness" exactly the same way as MTFS. However, the "Realtimeness" factor does exarcebate dramatically movements of the indicator, so it is still possible to set the parameter to -1 to eliminate its influence.

Tuesday, November 18, 2008

Updated pivot status extracted from the API

A few keen users may have desired to venture into the API to discover that some pivot points are sometimes not plotted with the Swing indicator. Indeed there is a "hesitation" factor built in the algorithm, but only on the last pivot point. All older pivots stabilize after a while.

In order to better detect and use that information, the API now also outputs a pivot status e.g. peak/trough/hesitant. With experience, we can notice that the "hesitant" pivot is often a "glass ceiling", "glass floor" i.e. some sort of invisible support or resistance which is just being tested. It generally preempts a reversal, and can be used to update money management ahead of the reversal.

The new API call is : BV_GetPivotDir which returns -1 for a peak, 1 for a trough, and 0 (again on last pivot only) if pivot is hesitant.

The TS_TT DLL (version 1.2) is currently sent to all users, and updated EL indicators will be be posted here soon.

Tuesday, November 11, 2008

Swing2 Update

A new version of the Swing2 indicator is ready for release. The convergence code of the genetic algorithm has been vastly improved by implementing a full search on condition only. This implies that normal swing search will always favour pivots detected on previous bars.

The next release will therefore calculate pivots on ALL bars. This is a major improvement this will allow a lot more support/resistance information to be derived from the indicator.

This work has been motivated by a request to finally turn our trading method into a mechanical system, and Swing2 was a major hurdle. A TSTT toolset update should be available within a few days.

Good trading to all,
bv

Monday, October 6, 2008

Swing Indicator (Genetic Algs)

I remind users of the toolset that the Swing2 indicator now takes a Search Depth value of 2 (previously 0 or 1 in most cases). It can help in very volatile environments.

Secondly, the genetic algorithms have internal settings which are adequate for recent PCs (Core Duo 2 ~2Ghz). I can on request recompile custom versions with deeper search abilities for users of faster PCs.

Wednesday, October 1, 2008

TS-TT Google Group

A new dedicated Google Group is now open to channel questions on our topic and make discussions available to all.

It is a public group for now.


Good trading to all,
bv

Monday, September 22, 2008

A few words on MM Stall level


Just took a quick snapshot on EURUSD to illustrate the MM stall level.

On two recent occasions, EURUSD reached MM stall level, bounced and came back to hit the major MM support/resistance level. This is a particular setup quite common in fast markets, especially when the pivot level seems unusually soft. Note however that prices sometimes do not come back to the major levels and bounce directly from stall level (H&S, C&H etc.).

How to approach a volatile market?

I have repeated again and again that our toolset is self-adaptable and universal. However all physicists will tell you any filter has a suitable frequency spectrum and you have all seen your usual oscillator going haywire on a spike (high freq.) or on a standstill (zero freq.).

What happened last week is unique, and warrants looking at a few settings, which will revert back to normal in a few days or weeks.

MM ShowMe: We generally use a Frame64 value of 2 i.e. a 128bars lookback period. I recommend temporarily a value of 1 (64 bars).

Swing2: We generally use a value of 0 (for regular swing traders) and 1 for scalpers or fast swing traders. We now increase it to 1 for regular or 2 for scalpers. Note 2 is the max. value for this indicator.

MTFS (aka AdLag64StoK_MTF) increase acceleration and realtimeness factor. Your discretion is needed here. A value of 0.25 for acceleration should do fine in many cases. Realtimeness can be upped to 2 or 3, but one should not overdo it as this can induce a Dorie effect (cf. Nemo) i.e. neglect the historical part of the pattern a bit.

The rest remains unchanged.
Note however that these proposed changes are 100% optional. I personally haven't changed anything to my own charts. Let's not forget that one can also move to a faster time frame set, or stay on the sideline for a while. Nothing with taking a break...

Good trading to all,
bv

Friday, August 8, 2008

August Update.


Another minor update to the DLL and API.
LEntropy is now ... LLEntropy with the application of the same zero-lag smoothing algorithm. MTFS (a.k.a. AdLag64Stok_MTF) has also got an extra optional parameter to tune the "realtimeness" of the indicator (see screenshot).
Market reading is simpler particularly around extreme peak/troughs of the Entropy indicator ( EntBin around 4 or 5 resp. -4/-5).

Another current development is the implementation of the SQL DB support. First release will be for a MS Access database, but it should apply all the same to SQL Server. We shall first record "pattern depth" i.e. the last 64 or 128 bars for each realtime dynamic pattern, i.e. Swing, Smooth, MTFS, Fib, Recurrence Plots, and now Entropy as well.

I'll publish more details by the time first release is available later this year.

So more to come soon and in the meantime, enjoy your holiday!
bv

Saturday, July 19, 2008

Blog Update

The Market Snapshot blog now publishes weekly posts on ER2 and ES.

A new private admarkan.blogspot.com takes over with daily snapshots. Contact me for access details.

Friday, July 4, 2008

Minor Update

MM lines have now been recoded inside the DLL, not only to have more calculations available through the API and the shared memory segment.
There is at most a few EL functions left to be recoded in C++. Ultimately, everything will be portable to other data feeds.

DLL + new TS files available on request.

Next step is the recording of Recurrence Plots (as well as every other calc) in a database for neural analysis.
Expected release: Sep 08

Sunday, June 15, 2008

Recurrence Plots for TradeStation

The toolset is now slowly entering Phase 2 with a new tool soon to be integrated. Here is a short preview of Recurrence Plots for TradeStation.

Recurrence Plots represent a new and relatively easy way to detect patterns in chaotic environments. The principle is to determine a phase space representation and lay it out in such a way that salient features of the series appear as regular patterns and invariants on a 2D graph. Such picture can then be analysed graphically or analytically.

An interesting site on RPs is : http://www.recurrence-plot.tk/ (reading references below)

I have therefore implemented RPs in a VC++ program using the API provided by the toolset allowing extracting data and calculations in an external program. (Some minor updates were necessary, so a new version of the library itself will be soon released).

I here give you a preview of what RPs look like on a 60mins EURUSD chart, both using price (z-score) and a MTFS indicator. The price RP uses a delay based phase representation whereas the MTFS uses the built-in separation in 3 views of the stochastic indicator.


Z-Score Price RecPlot


MTFS RecPlot

We will now record, analyse first visually then with the addition of neural networks to better detect salient turning points in the market.

For more reading, see the links page on the above site. I particularly recommend reading Norbert Marwan publications.

Thursday, May 8, 2008

sFLC3 Fuzzy Logic for TradeStation

Although not yet integrated in the main toolset, the Fuzzy Logic API is now available for testing on the main ForeTrade web site:

http://www.foretrade.com/fuzzy_logic_for_tradestation.htm

I understand it may sound a little complex at first, but rest assured it is very simple to use, and all FL functions have been grouped in a single EasyLanguage function.

http://www.foretrade.com/fuzzy_logic__easylanguage_code.htm

Two sample indicators are also provided for testing purposes:

http://www.foretrade.com/fuzzy_stok_indicator.htm

For more info, please contact Support@ForeTrade.com

Monday, March 31, 2008

A closer look at EURUSD


Hello everyone,

I am often asked about the EURUSD forex pair. Will the US$ drop further or will it recover? Again, no crystal ball, but let's have a look at the current set of EURUSD charts.

On the 60mins chart, we first notice a significance level that is fairly weak, and prices certainly confirm that confusion. SQ (Swing Quality) is not too bad, so we'll take the current UP swing as the most likely direction again. When it comes to entries though, one should clearly look at a lower time frame.
MTFS is kind of "stuck" in overbought zone, hence does not provide us any significant information. Nor does Entropy hovering around the zero line.
We therefore have a remaining upward bias with momentum exhaustion. We can also notice that Swing has detected a potential pivot (number 3 on last high), but has not toggled down yet. Lastly, we see a strong support level at SS:1.5625, indicating that current strong resistance MM level at 1.5625 at higher time frames could be tested for support.
That will be all for now. We have an upward bias with increasing hesitation, and should you want to trade in this confusing environment, just change time frames.

On higher time frames, the Daily chart is also UP and has a higher significance level. It is obviously way too late to enter long right now but one may want to stay in if one is long already, just tightening stops a bit. MTFS is also very much stuck in overbought zone, bars are blue, and if we didn't have that strong resistance level and fast decreasing entropy, we would be happily bullish still.
Where should we put our stops? Well, it's up to each and everyone, and i give no advice in that regard. Money management depends too much on available capital and personal risk profile. It is also likely that the Swing indicator will toggle down at some stage. Those of you who run the toolset can see the Swing indicator sometimes hesitating and adjusting when the pivot point is not confirmed yet. A warning can however be a good indication of an exit point particularly on a strong resistance level. Swings being UP-UP-UP on all 3 time frames, there is NO reason to go short anyway.

Lastly, the Weekly chart: all charts look the same* so the interpretation is simple and provides context for decision made in lower time frames. All we can see here is that it will certainly take a long time before we see the US$ recover significantly.

Obviously one must always refrain from reading too much into charts. I all too often see junior traders "wishful reading" charts, i.e. reading where they would want the market to go. It can be a fatal mistake...
For the time being we maybe hesitant to take profits and we should possibly adjust stops, but that's it.

* Reminder: the only indicator with significant parameters is MTFS aka AdLag64StoK_MTF. Some may have other settings than mine, but most important, one must be comfortable with one's charts. I personally hardly ever change any chart settings.

Friday, February 29, 2008

The more I practise the luckier I get

Famous phrase from golf legend Gary Player... I believe it (modestly) applies here too. Indeed while I can add postings from time to time to better explain the technique, Phase 1 of the indicator toolset is quite stable now, and will only warrant very minor updates from time to time. On the to-do list still lies the removal of the last 2 parameters in the MTFS to make the toolset 100% self-adaptive, but that's about it, so no pressure, no rush...

Practise makes perfect.
Yes, there is no documentation except for a document on the ForeTrade website, and the rest is left to users to read and analyse daily market snapshots to better grasp the trading technique.
It may take a short while as it is relatively unconventional, but it quickly becomes very natural. A few beta testers did not quite get it or have not been patient enough. I understand but I certainly regret that. Beta testing for Phase 1 is now closed. I am just pushing the 'release candidate' a bit because of other commitments like the forthcoming release of the Fuzzy Logic library (both generic and for TradeStation).
FL is not quite a distraction, as it will be an essential part of the Phase 2 of the toolset, with better decision making tools up to full trading automation. Its release is still scheduled for later this year.

Happy trading,
bv

Saturday, February 23, 2008

Some common visual patterns

The trading technique being presented here uses a proprietary TradeStation toolset, but since it is still a discretionary trading method, why not adding some visual interpretation to it?
Here below is a description of very common patterns to look at, and one reason for posting this document today is that one may find an interesting pattern formation on NDX over the next few days.
PATTERN TYPES:
Wedges:

The wedge formation is also similar to a symmetrical triangle in appearance, in that they have converging trend lines that come together at an apex. However, wedges are distinguished by a noticeable slant, either to the upside or to the downside. (As with triangles, volume should diminish during its formation and increase on its resolve.).
A falling wedge is generally considered bullish and is usually found in up trends. But they can also be found in downtrends as well. The implication however is still generally bullish. This pattern is marked by a series of lower tops and lower bottoms.
A rising wedge is generally considered bearish and is usually found in downtrends. They can be found in up trends too, but would still generally be regarded as bearish. Rising wedges put in a series of higher tops and higher bottoms.

Price Structure Pattern Structure Result
Falling Wedge in Uptrend H, L, LH, LL, LH, LL Breakout to the Upside
Falling Wedge in Downtrend L, H, LL, LH, LL Breakout to the Upside
Rising Wedge in Uptrend H, L, HH, HL, HH Breakout to the Downside
Rising Wedge in Downtrend L, H, HL, HH, HL, HH Breakout to the Downside

Symmetrical Triangles:

Symmetrical triangles can be characterized as areas of indecision. A market pauses and future direction is questioned. Typically, the forces of supply and demand at that moment are considered nearly equal. Attempts to push higher are quickly met by selling, while dips are seen as bargains. Each new lower top and higher bottom becomes more shallow than the last, taking on the shape of a sideways triangle. (It's interesting to note that there is a tendency for volume to diminish during this period.) Eventually, this indecision is met with resolve and usually explodes out of this formation (often on heavy volume.) Research has shown that symmetrical triangles overwhelmingly resolve themselves in the direction of the trend. With this in mind, symmetrical triangles in my opinion, are great patterns to use and should be traded as continuation patterns.

Price Structure Pattern Structure Result
Symmetrical Triangle in Uptrend H, L, LH, LL Breakout to the Upside
Symmetrical Triangle in Downtrend L, H, LL, LH Breakout to the Downside

Ascending Triangles:

The ascending triangle is a variation of the symmetrical triangle. Ascending triangles are generally considered bullish and are most reliable when found in an uptrend. The top part of the triangle appears flat, while the bottom part of the triangle has an upward slant. In ascending triangles, the market becomes overbought and prices are turned back. Buying then re-enters the market and prices soon reach their old highs, where they are once again turned back. Buying then resurfaces, although at a higher level than before. Prices eventually break through the old highs and are propelled even higher as new buying comes in. (As in the case of the symmetrical triangle, the breakout is generally accompanied by a marked increase in volume.)

Price Structure Pattern Structure Result
Ascending Triangle in Uptrend H, L, H, HL Breakout to the Upside

Descending Triangles:


The descending triangle, also a variation of the symmetrical triangle, is generally considered to be bearish and is usually found in downtrends. Unlike the ascending triangle, this time the bottom part of the triangle appears flat. The top part of the triangle has a downward slant. Prices drop to a point where they are oversold. Tentative buying comes in at the lows, and prices perk up. The higher price however attracts more sellers and prices re-test the old lows. Buyers then once again tentatively re-enter the market. The better prices though, once again attract even more selling. Sellers are now in control and push through the old lows of this pattern, while the previous buyers rush to dump their positions. (And like the symmetrical triangle and the ascending triangle, volume tends to diminish during the formation of the pattern with an increase in volume on its resolve.)

Price Structure Pattern Structure Result
Descending Triangle in Uptrend L, H, L, LH Breakout to the Downside

Flags and Pennants:

Flags and pennants can be categorized as continuation patterns. They usually represent only brief pauses in a dynamic market. They are typically seen right after a big, quick move. The market then usually takes off again in the same direction. Research has shown that these patterns are some of the most reliable continuation patterns.
Bullish flags are characterized by lower tops and lower bottoms, with the pattern slanting against the trend. But unlike wedges, their trend lines run parallel.
Bearish flags are comprised of higher tops and higher bottoms. "Bear" flags also have a tendency to slope against the trend. Their trend lines run parallel as well.
Pennants look very much like symmetrical triangles but are typically smaller in size (volatility) and duration.
(Volume generally contracts during the pause with an increase on the breakout.)

Rectangles:

Rectangles should generally be traded as continuation patterns. They are indecision areas that are usually resolved in the direction of the trend. Research has shown that this is true far more often than not. Of course, the trendlines run parallel in a rectangle. Supply and demand seems evenly balanced at the moment. Buyers and sellers also seem equally matched. The same 'highs' are constantly tested as are the same 'lows'. The market vacillates between two clearly set parameters. (While volume doesn't seem to suffer like it does in other patterns, there usually is a lessening of activity within the pattern. But like the others, volume should noticeably increase on the breakout.)

Head and Shoulders:

The head and shoulders pattern is generally regarded as a reversal pattern and it is most often seen in uptrends. It is also most reliable when found in an uptrend as well. Eventually, the market begins to slow down and the forces of supply and demand are generally considered in balance. Sellers come in at the highs (left shoulder) and the downside is probed (beginning neckline.) Buyers soon return to the market and ultimately push through to new highs (head.) However, the new highs are quickly turned back and the downside is tested again (continuing neckline.) Tentative buying re-emerges and the market rallies once more, but fails to take out the previous high. (This last top is considered the right shoulder.) Buying dries up and the market tests the downside yet again. Your trendline for this pattern should be drawn from the beginning neckline to the continuing neckline. (Volume has a greater importance in the head and shoulders pattern in comparison to other patterns. Volume generally follows the price higher on the left shoulder. However, the head is formed on diminished volume indicating the buyers aren't as aggressive as they once were. And on the last rallying attempt-the left shoulder-volume is even lighter than on the head, signalling that the buyers may have exhausted themselves.) New selling comes in and previous buyers get out. The pattern is complete when the market breaks the neckline. (Volume should increase on the breakout.)

Inverted Head and Shoulders:

The head and shoulders pattern can sometimes be inverted. The inverted head and shoulders is typically seen in downtrends. (What's noteworthy about the inverted head and shoulders is the volume aspect. The inverted left shoulder should be accompanied by an increase in volume. The inverted head should be made on lighter volume. The rally from the head however, should show greater volume than the rally from the left shoulder. Ultimately, the inverted right shoulder should register the lightest volume of all. When the market then rallies through the neckline, a big increase in volume should be seen.)

Thursday, February 21, 2008

Fuzzy Logic for TradeStation

Some may have seen the "Simple FLC 3" library presented on our main ForeTrade site (http://www.foretrade.com/fuzzy.htm)

The sFLC DLL has now been ported to TradeStation and is ready for beta release (in fact RC status already). Please contact me if you wish to test it before it is released (March 2008)

A shareware version will be available (3 concurrent FL controllers on TS charts), then more FL controllers can be enabled in the DLL per pack of ten costing US$250 each. The DLL like the indicator toolset is linked to the unique TradeStation customer number.

Wednesday, January 30, 2008

Update

Looking at the post below, we now know in hindsight that SPY dropped further to reach a bottom on Jan 22nd and 23rd... at least for now...

Changing subject, i have decided to postpone the commercial release of the toolset, not so much for technical reasons than because of the lack of support and training resources. I'll have to train a few more users then make sure the knowledge can be passed on etc. Phase 2 beta (signal generator) is still planned for 4th quarter 2008.

For the time being, market snapshots will continue being posted daily on http://marketsnapshot.blogspot.com

Saturday, January 19, 2008

Are we close to a bottom on SPY ?


Let's have a look the current SPY chart, and review our indicators.

On the 60mins chart, we have a clear divergence of the MTFS indicator ("failed recovery"), which combined with the clearly down swing and low MTFS significance level did noto fool us at all. However in case some would have taken a long position on the MTFS line crossover, the downturn of the white line looks like a pseudo 'bell shape pattern' which we always use for a 'return to the main trend' trade.

What about Entropy? Entropy is undoubtedly very low, and one may be tempted to look for a reversal point. When Entropy is indeed close to its historic lows (LEntBin = -5), a bottom provides good reversal signals. Having said that, this is only a support indicator. Swing is still down, bars are red, and we do not have a good looking MTFS.

How can MTFS look better if line crossover occured a few days ago? One must not forget that MTFS is also self-adjusting, so it will redraw itself as new events come along. MTFS is essentially used for the pattern shapes (line crossovers, parallel lines, line gradients) and NOT for the end point value like with most conventional indicators.

End of story: the 60mins chart tells you to wait, hang on short or maybe take profits. That's all for now.

The daily chart is the dominant chart here. We now know that there is no entry point from the 60mins chart, but what is the daily chart telling us?

Entropy looks sad, just like on the 60mins chart, and MTFS has tempted a (bearish) crossover. The good thing is that the indicator will adjust to a better looking crossover in oversold territory. At the moment, the line gradients are far too bearish. Then and only then will we look at a Entropy bottoming out. As usual, the picture gets clearer and clearer and we shall probably see prices stall at a Fib/MM target, and the Swing indicator (the earliest of all) attempting to turn up.

Obviously, each trade is to be considered within the context given by the higher time frame, and at this point in time, the weekly chart is also in correction mode (MTFS, Entropy etc).

That's all for now. We know that we must just be patient and wait at least for MM stall level ~128, or more likely right down to 125 (Fib/MM target).

Bears still have some appetite indeed...


NB: I added the SQ indicator (Swing Quality)

Friday, January 18, 2008

Swing Quality


I have added a simple yet interesting indicator called Swing Quality. It is a correlation between swing legs and actual prices. Although visually obvious, it is interesting to notice that swings can sometimes be out of tune for a while, until the indicator adjusts itself again in realtime. SwingQuality can therefore be used to filter those transient situations.

Here above, we have a QQQQ chart with swings and Fibonacci calculated automatically. Swing legs are not always suitable for Fibonacci calculations, and here they clearly are. A Swing Quality above 30 is generally more than good enough. Again, since the algorithm continuously adjusts itself, SQ will very seldom drop to low levels.

As a reminder, Fibonacci expansion levels are displayed from L1 to L5 at the right of the chart. L1 and L5 are seldom used, but L2 (Fib COP), L3 (Fib OP) and L4 (Fib XOP) are quite commonly used. Here QQQQ is sitting on COP. Quite an easy short signal it was, wasn't it...? :)

Tuesday, January 15, 2008

The Lomb-Scargle cycle detection indicator


Another lesser used indicator in the toolbox is the Lomb-Scargle indicator. There is well-known controversy regarding existence of cycles in markets, and even a good cycle detection algorithm like L-S admittedly has flaws, i.e. does not guarantee any stability in cycles. It is merely a market snapshot over a number of bars, and the algorithm tries its best to determine the underlying frequencies at a point in time. Markets are erratic, markets are chaotic, hence the internal dynamics constantly change.
Having said that, such information can probably be used to the trader's advantage. For instance, one can detect when there is a sudden change in dynamics or on the contrary look for periods of time where cycles are relatively stable.

Here above, one can see that a single frequency emerges from the spectrum. The indicator can plot the top 2 frequencies as higher frequencies are generally detected with ridiculously low amplitudes, i.e. noise... In the picture, the top frequency is represented by yellow dots, with its amplitude plotted as a histogram.

Although I do not use it often, i look forward to reading any feedback you may have with this Lomb indicator.

NB: there are a few variations on the algorithm available on the Internet. This one is the EXACT replica from the MIT PhysioNet laboratory.
http://www.physionet.org/physiotools/software-index.shtml

Wednesday, January 9, 2008

Where to start?

I am often told: your charts look great but what do I do with them? Where to start?

My answer is always the same though: Absorb the charts until you make them your own!

There are now a number of references on this blog for each tool including an audio commented chart. Of course a few prerequites like knowing about Fibonacci and some experience with markets are definitely necessary.

Now, if you want a few pointers to start:
1) Look at Swings to get the overall picture in 3 time frames
2) Check which time frame(s) is dominant and stay with it if it corresponds to your trading horizon (don't forget that a lot of traders get overweight sitting and waiting because there is often nothing to do but wait...)
3) With experience, you'll pick up easly recognisable MTFS patterns, i.e. decision points.
4) Fibonacci and MM are essential tools. In most market situations, one or the other will give you clues, and ambiguity almost disappears when they coincide.
5) The more indicators you get, the clearer the picture will be of course. Try and figure out what you need for your trading style. The Swing indicator is usually the earliest of all.
6) Do not enter counter-trend. Follow swings and bar colours.

Lastly, and this is maybe the trickiest part for beginners: MTFS is the only indicator which requires setting 2 parameters: length and acceleration.

Here is the way I do it:

I generally use 8, 16 or 24 bars (obviously the shorter lengths are more responsive). As far as Accel is concerned, there is a built-in double adaptive algorithm and adaptability to volatility is maximum around 0.1 or 0.2, and minimum at 0 or above 0.33
It is best to set both parameters to values you like and then leave them unchanged for good.

A good way to set those parameters is to find a configuration with a clear swing and try and fit MTFS to "tell you the best story" i.e. lines cross over near pivots points and significance level is high. Obviously MTFS will not always give you the most perfect pattern as the market is erratic in nature, but your indicator will work best when it is tuned in the 'frequency' of the market. You may want to fine-tune it again until you find it is in sync with the market. Note that such setting is only valid on clear patterns, and do not forget that the market most of the time carries no relevant or usable information. This may look like a tedious exercise at first, but it will make your life easier later on.

Feel free to send questions and screenshots anytime, which we can discuss via IM.

Happy trading day to all... :)