The vexed question of "Filters"

trader.trends

Well-Known Member
#1
As traders when we want to check a system we start taking trades when the system gives the signal. The system can be Breakouts, fakeouts, gaps, trendline breach, myriad ema cross overs or whatever. As we start taking the trades at the signal we realise that we are getting caught in the false trades. When we take a NR-x day breakout on the long side, the trade just went a few points up before collapsing into the range triggering our stops. The trendline breached in an illusory manner before breaching the other way. The EMA cross over was just a whipsaw. Irritation and frustration builds up. Then we hear the word 'Filter' whispered in our ear.

We start looking at the history of our trades and realise that if we had a filter of X points all those damned false trades which hurt our bottom line could be annihilated in one stroke!! What a discovery! We build X points as filter into our trades. As we go along we find that instead of X points X percentage is a wiser move. We pat ourselves on our back and start relaxing with our trades. As we total up our trades at the end of the month, it is interval time and SRK in the background is hamming, "Picture abhi baki hain mere dost"

Post interval we discover the darker side of the Filter's character. The under belly gets exposed. The losses making trades of the Pre-filter era are giving bigger losses and the profits of the pre-filter era have shrunk because of the filter. For those who think "filter" is used for making coffee just a little detour.

Suppose the ema cross over is the strategy. The long should have triggered at 5000. We start keeping a filter of say .2%. So instead of taking the long trade at 5000 we will take it at 5010 and if it is a short trade it is taken at 4990. For the longs if the exit was at say 5050, our profits after filter is 40 reduced by 10 because of the filter. If SL was at 4950 our losses are 60 instead of 50. Both profits have decreased and losses have increased. Double whammy. The larger the filter factor, the greater the difference. Damn!

So what do we do? Each of us have to find the balance. Do we use the filter or do we not?

I looked at my trades and concluded filters add one more complication to the strategy. One more factor to keep track of. Since I like to keep it simple I have dropped filters from my trades. In the long run I feel filters don't really make a difference. We would be naive to believe that by using filters we are escaping from the tribe of stop-hunters. All this comes with the usual disclaimer of being just an opinion.

Peter Clemenza tells Rocco in Godfather, "Leave the gun, take the cannoli". If he traded perhaps he would say, "Leave the filter, take the trade"

How do you deal with filters?
 
Last edited:

enygma

Well-Known Member
#2
Ah, a subject close to the heart ...

My own analysis of my trades and also from back testing, has lead to my filters being trimmed down substantially, more so for intra-day trades. For positional trades, I still use filters. On intra-day charts, patterns succeed or fail almost instantly and we are much better off taking the trade with negligible filters. Filters often just delay the inevitable ...

For positional trades in which we are sitting on good profits, trailing stop losses with filters ensure we stay with the trend. Otherwise, we would run the risk of cutting our winning trades too quickly. On daily charts, we need to give some amount of leeway to the stock while it still sticks with the pattern or the trend.

Just my 1.285 cents ...

Regards,
Enygma.
 

linkon7

Well-Known Member
#3
Filters are something very useful for every trader. But a fixed point filter or a percent one tends to dampen the spirits a lot.

If i were to trade breakouts, i would use another complimentary system to confirm the breakout and then take the trade. I normally use regression channel and boilinger band to give me the entry point.

If using ema cross over, then high/low of the last 2 bars is a good filter.

trendlines break can be confirmed with a break of trendline on the PDI / MDI or RSI

The main purpose of filters is to avoid the false signals and if we have complimentary indicators for the same trade then confidence to pull the trigger increases. Just adding a 2-3 point filter or a percent filter solves nothing other than adding to the frustration.
 

Amanc

New Member
#5
Filters are something very useful for every trader. But a fixed point filter or a percent one tends to dampen the spirits a lot.

If i were to trade breakouts, i would use another complimentary system to confirm the breakout and then take the trade. I normally use regression channel and boilinger band to give me the entry point.

If using ema cross over, then high/low of the last 2 bars is a good filter.

trendlines break can be confirmed with a break of trendline on the PDI / MDI or RSI

The main purpose of filters is to avoid the false signals and if we have complimentary indicators for the same trade then confidence to pull the trigger increases. Just adding a 2-3 point filter or a percent filter solves nothing other than adding to the frustration.
Rigtly said, whipsaws - false signals- etc........can be reduced using the indicators... But i personally feel , in the end these indicators are also mathematical farmulae and use historical data to predict future trend, All we should concentrate on the buying or selling pressure for that respective time frame what we are dealing with. That would be more reliable.
 

linkon7

Well-Known Member
#6
Rigtly said, whipsaws - false signals- etc........can be reduced using the indicators... But i personally feel , in the end these indicators are also mathematical farmulae and use historical data to predict future trend, All we should concentrate on the buying or selling pressure for that respective time frame what we are dealing with. That would be more reliable.
There is a story of 5 blind men feeling the elephant with their hands and each giving a different version of what they think the elephant looks like. Thats what indicators do to price. No one can predict the next bar on the right and indicators make assumptions based on the bars on the left and tell us a half true story. combine a lot of indicators, and you'll have a confusing story. Best is to understand how each of them reacts to price movements and then take a decision.

When market is trending, use one set, when its sideways, use another set.
 

Amanc

New Member
#7
There is a story of 5 blind men feeling the elephant with their hands and each giving a different version of what they think the elephant looks like. Thats what indicators do to price. No one can predict the next bar on the right and indicators make assumptions based on the bars on the left and tell us a half true story. combine a lot of indicators, and you'll have a confusing story. Best is to understand how each of them reacts to price movements and then take a decision.

When market is trending, use one set, when its sideways, use another set.
Yes agree with ur story but wish it could be applied in the market.............. 5 blind men - touching elephant, they can be sure whether its an elephant or not if they have touched that in past coz elephants have certain defined structure,so prediction is highly accurate coz elephants can not have different structure what those blind men have learnt in past.
Whereas markets can not be predicted to same level of certainity, coz it does not have a specified structure what u can learn. In markets elephant and horse are just differ by a very minute, almost invisible difference which decides your success or failure. And when your money is at stake you cant afford to take chance on the basis of some historical events.

Technical analysis is based on historical events, prices have memories, follow trend but each time is so minutely diiferent compared to past that can actually decide your success or failure- John Murphy
:cool:
 

linkon7

Well-Known Member
#8
Yes agree with ur story but wish it could be applied in the market.............. 5 blind men - touching elephant, they can be sure whether its an elephant or not if they have touched that in past coz elephants have certain defined structure,so prediction is highly accurate coz elephants can not have different structure what those blind men have learnt in past.
Whereas markets can not be predicted to same level of certainity, coz it does not have a specified structure what u can learn. In markets elephant and horse are just differ by a very minute, almost invisible difference which decides your success or failure. And when your money is at stake you cant afford to take chance on the basis of some historical events.

Technical analysis is based on historical events, prices have memories, follow trend but each time is so minutely diiferent compared to past that can actually decide your success or failure- John Murphy
:cool:
You misunderstood the elephant story. Indicators are derivative of price action. based on the method of calculation, they give you a reading which can never be substituted for price. But definitely serves as a early warning bell for us to take action.

success or failure is just a product of proper risk management and discipline. There are thousands of ways to play this market and only this 2 dictates the profit / loss. Nothing else matters.
 

rajputz

Well-Known Member
#9
One thing worth understanding is that "supply and demand in accordance with price provides us indicators signals. Indicators Does not predict future movement, rather price make the future movement of Indicators."
 

AW10

Well-Known Member
#10
Interesting topic.. In my view, What is your belief about FILTER ?
If you think that it can reduce your wrongly stopped out trade and increase win %..then most of the thought given by others are right.

But, if you think that market produces lot of redundant noise in all timeframe and the intention is to filter that noise, then YES, filters do help.
That will still give u stopped out trades, but frequency will be far less..
And who stops us from re-entering the position, even when we are stopped out but all other criteria of being in trade appear on next bar ?

Their size do vary with the basic concept of system. Say for contrarian entry, it needs to have more buffer, For trades done during opening hours due to high volatility, it needs higher buffer,
but a trend following system that uses 2/3 confirmation, you can afford to be conservative in filter size.

In my trading, I classify "noise for a timeframe" using ATR of that timeframe and then use some factors of it as filter. Eg - for day trade, 5 min ATR is
8 points on nifty, so 0.6*ATR ie. approx 5 points is good enough to absorb the noise of 5 min TF. For a swing trade on 60m chart where ATR is 35 points, 5 points of filter is too small..
The filter there turns out to be about 20 points. And similarly, for daily TF, the number could be different.

And my systems do have rule for re-entry so that I don't keep crying for being stopped-out and miss a great trade on next bar.

One can also take help of statistical methods to find it.. or just analyse prev trades and set a limit that would have avoided atleast 70 to 80% of false stop-outs.

To me, Filter design is the ART of trading and no way it can avoid stopping out. So handle it with reality of your belief about market. At the end of the day, what we trade in the market is our belief.
If someone thinks Filter is useless then he is right and that's what he is better of trading. Using filter will only give him -ive experiences. And if someone believes that no, we need some buffer to absorb the mkt shocks and give time for our entry to workout, then he can't trade without filter.. And both traders are right in their own way.

Happy Trading
 

Similar threads