SimCon

[boots]

SMASAR

The chart in the previous post shows buying and selling based on a single chart using simple support and resistance lines. Add to this a simple moving average for trade confirmation and you have a trading method.

The entries in the direction of the SMA would be taken and the ones not in the direction would become the stops. The stops follow along at the last pivot until hit. An SMA of 17 was a popular choice but a longer one may give better results. Also the longer the chart the more reliable the signals will be ( usually always the case ). Any trigger chart over 30 minutes could use the Daily/Weekly trend confirmation spoken of in earlier post.

Good luck

boots

[Phoenix]

Would also be interested in how it would act if it was an adaptive EMA. Maybe I can get Loomis to change the code so we can try that.

The Adaptive Moving Average is an EMA. The smoothing variable is based on the Efficiency Ratio, which determines whether the current bar's value should get more or less weighting in the EMA.

Anyway, I have done some backtesting of 21AMAX vs 21SMAX with mixed results. I tried it on hourly as well as 15 minute charts. On the NQ, the AMA gave much better results on the hourly charts over the past three months, but was not as good on the ES. I decided I will leave both on my hourly charts with a plot of the Efficiency Ratio and maybe throw on a Bollinger Band and watch it for a while.

I still have not found a trading system that I like and believe that I can follow.

[boots]

>The Adaptive Moving Average is an EMA.<

Well that answers that then..........hehe. Was not aware of that.

I have been using the EMA as a line in the sand. Not using it in itself as an indicator of trend or trigger but as a relative indicator to market position. Simply put, if the market is above or below then the trend is up or down. Using the same length on different time spans has worked well for me in determining when to trade long, short or not at all. It appears very good at keeping you out of the market during chop.

The EMA was not one of my favorites until I started using them in this way. The EMAs ability to change direction and follow the market on the very bar that crosses it means it will always be moving closer to your position regardless of market conditions or EMA length. This can be a handy tool and a good "line in the sand" for determining market direction.

For my purposes I continue to return to the simplest relationships. Finding them the best and Truest of all. Using the markets relationship to a "point" on the chart gives you two options that are true: market is Up or Down in relation to that point. Use this point on two different charts and you now get three options that are true: market is Up (above both), Down (below both), or Sideways (between both).

This can then become a reliable base to build rule sets from.

[boots]

I would like to recommend a book I found very interesting and much different than most books about trading. The book is "Fooled By Randomness" by Nassim Nicholas Taleb.

It has little to do with systems and trading but a great deal to do with what you should be looking for in a method and how you should approach the markets.

[Phoenix]

Confirmation of Trend

Boots: Your ideas on using emas in multiple timeframes to confirm trend led me to another idea: multiple length regression channels in a single timeframe. In my work in the actuarial field, regression is relied on heavily to determine and predict trend, so why not in the markets as well?

Here I have put regression channels of 13, 21 and 34 periods (using fibonacci numbers) on an ES hourly chart. If all the channels are pointing in the same direction, it indicates the market is trending that way. I also have the Kaufmann Efficiency Ratio as an indicator here,matching the middle length of 21 periods. In my usage, if ER is above .15 (blue line), the market is "efficient" enough to be tradeable. Below .15, no trades.

http://share.esignal.com/groupcontents.jsp?folder=&picture=Trend.png&groupid=49

Let me know what you think.

Phoenix

[boots]

Hey Phoenix,

Sorry for the delay. Am playing with your idea. That ER thingy is real cool. Have not seen that before. Are you using the 3 channels as trend indicator AND trigger or what would be the trigger for the trade, what would get you in the trade?

Will be playing with this some more.

Thanks for the idea.

boots

[Phoenix]

Regression Channels and ER Thingy

Hi boots:

I was thinking of the regression channels just as trend confirm (maybe augmented by the Efficiency Ratio). This was all still in my mind in the context of a moving average crossover trade signal. The trend confirm would be a trade filter, telling one when to stand aside.

The ER is something I find very useful, not only as a trade filter, but when it hits high-end extremes it is a good signal for the end of trends and imminent reversals. You can see a couple examples of that on the chart I posted this weekend. The 21 periods may not be the optimal setting, however.

Not too many charting packages include it. I got Jerry Medved to code it into his QuoteTracker program a couple years ago, but I haven't seen it anywhere else. I can post the .efs file for it if you want.

A good example of the value of ER as a trade filter was today. I had short daily signals on the NDX at yesterday's close for an EOD daily system that I developed based on moving average crossovers, DMI crossover and a close below the 20 day ema. But the 10 day ER was at .03, so no trade could be taken. Good thing, since NDX rallied 48 points intraday today!

[vladar]

Regression Oscillator

Boots,

You may also want to take a look at the regression Oscillator.
It has some very nice aspects to it that, thus far, seem to be somewhat better than moving averages.

http://share.esignal.com/groupcontents.jsp?folder=&picture=Regressions%20.png&groupid=48

I got the code for it from Esignal:

////////////////////////////////////////
function preMain()
{
setPriceStudy(true);
setStudyTitle("tsf");
setCursorLabelName("TSForecast");
}
function main(nInputLength,nInputBarPlus)
{
var len = 25;
if (nInputLength != null) len = nInputLength;
var BarPlus = 7;
if (nInputBarPlus != null) BarPlus = nInputBarPlus;
var vPrice = getValue("Close", 0, -(len+1));
var Num1 = 0.0;
var Num2 = 0.0;
var SumBars = len * (len - 1) * 0.5;
var SumSqrBars = (len - 1) * len * (2 * len - 1) / 6;
var SumY = 0.0;
var Sum1 = 0.0;
var Sum2 = 0.0;
var Slope = 0.0;
var Intercept = 0.0;
var i = 0;
for (i = 0; i < len; i++)
{
SumY += vPrice[i];
Sum1 += i * vPrice[i];
}
Sum2 = SumBars * SumY;
Num1 = len * Sum1 - Sum2;
Num2 = SumBars * SumBars - len * SumSqrBars;
if (Num2 != 0) Slope = Num1 / Num2;
Intercept = (SumY - Slope * SumBars) / len;
var LinearRegValue = Intercept + Slope * (len - 1 - BarPlus);
return LinearRegValue;
}

[boots]

Re: Regression Channels and ER Thingy

Hi Phoenix

> I can post the .efs file for it if you want. <

I have not been able to find this, can you post it please or load it in the SimCon file share. I would like to play with it some and see the code.

Also............did you see the Valdar post with the "Regression Oscillator". I think that might work real well with what you are doing. It is very responsive but smooth. I can put a 50,4 on a chart and it is as smooth as my 34 EMA but much more responsive. Your cross over idea might benefit from that.

boots

[Phoenix]

Efficiency Ratio.efs Posted to Simcon File Share

Posted to SimCon File Share. I couldn't find it either, so I "wrote" it. (The code is embedded in the KAMA.efs file.)

I looked at Vlad's Oscillator. My observation to him on the E-mini Pit was that the 21 period could be very closely approximated with a 15sma, but I haven't watched it real-time to note the responsiveness you mention.

[boots]

This is a re-post of some work by dloomis.


function preMain()
{setCursorLabelName("Spread"); setStudyTitle("Spread")}
function main() {var BL=0;
return new Array(getValue("Close", 0, -1,"$spx") -
getValue("Close", 0, -1,"es z2=2"),BL);}


Thanks Dave

[vladar]

ES 13 min chart with Triggers

My neverending analysis of ES Renke charts with indicators as triggers........trade is considered successful if moves at least 4 points, when stop loss is moved from 4 points to break even, and is continuously trailed at 4 points thereafter.

http://share.esignal.com/groupcontents.jsp?folder=&picture=ES%2013%20min%20Renke%20chart.png&groupid=48

[boots]

Vladar

These Renke charts are interesting to say the least. Can you explain what it is that you like about them, what you see as an advantage?

Could you also explain the "False" indication on the chart.

Thanks

boots

[vladar]

Boots,
The False Lables show the one false signal in that list.
For me, a false signal is one that stops out the trade, ie, price changed enough to have the indicators cross over, but it was short lived.

What I like about the renko charts is the effect they have on the indicators. Looking at a standard chart, those same indicators will whip around and creat extremely jagged lines as price attempts to find a direction.

The renko charts add a 'brick' (candle) only when price has moved a certain amount, so it sits quietly until the delta is reached, then prints a brick. This smooths out the indicators, which now react with more orderly oscillations.

I think Phoenix tried trading them for a day, but hated the drawdown. I'm trying to filter my backtest to give me more information regarding the renko signals.....my first tests showed that for a size-2 (the standard) renko chart, trading ES gave at most a four point drawdown on all successful trades, anything more than 4 would most likely lead to a losing trade, so I've been watching (papertrading) with a four point stop.

As posted on SI, I've been watching these renko charts for about 6 months with excellent results....don't know why I don't have the hoo-ha's to actually trade them. I think this is the kind of thing I may try first with automated trading through IB. Just let the program sweat it out....I don't think I could watch something that sits static for up to an hour at a time before printing anything out.......I've just spent too many years staring at candles that give immediate feedback (gratification).

[Phoenix]

Renko Chart Backtesting

I have found that standard backtesting on Renko charts gives misleadingly good results. Because there is no time dimension to these charts, and if they are based on closing prices, they may not show you the price swings either side of the closing price. If anyone has written some backtesting code that works for Renko charts, please post it. Thanks.

Phoenix