How to improve my percent rate of profitable trades

Hello

My trading system give me arround 25% of trades with a positive return, meaning 75% failed

Even do, my system is nicely profitable because of my rules to follow my position. This is very nice on paper, but in real life, 75% of failure is to high.

This system is not designed to be a trend follower. So it is normal for me to have a higher rate of failure. so I take orders when they filled my setup rules.

I already use RSI and ADX as filter. But obviously, lots of falling entry got trougth

- What is your favorite filter. Can't use ema or other moving average as again, this sytem is not a trend follower

Any input appreciated

Martin

It all depends on the type of system. Depending on your rules for entry/exit and your trading style, having more losses than wins doesnt necessarily matter.

What kind of system are you trading?

Hello ItsMeTyler

A bollinger band and 4 setups rules from L.I.V.E. developped by Ivan Krastins. It include money management, time management.

I'm looking for reversing setups that occur in the Bollinger band area. So i'm mainly counter trading.

My main gain is done by the strategy that i have developed to manage my position. Average 5R seen 35R( 5 time the risk). My risk is known, but not my return. Since counter trend trading is more risky, I need to profit as much as it is possible from a position.

Now, my goal, reduce my fault entry. A technique to go over ( not take) entry that has less chance to reverse the trend.

Martin

Hi Martin,

If your system is not a trend follower and produces on average a 75% failure rate, then what is it?

No one can suggest any filters for you to improve the performance without knowing some basic information or the concept behind your system...


Regards,

Nader

You need a system with an edge. Now I am not exactly sure if bollinger bands provide an edge, since the market is definitely not random and thus not normally distributed.
You can apply as many filters as you want, it will not increase the profitability, if the system as a stand-alone is not profitable.

Totally agree....If the engine of the car is not working, adding new rims, spoilers, hoods and other stuff will never make the car move an inch...


Regards,

Nader

Nader, Nicely put. Totaly agree to.

System seem to work as the 25% return more than 3 time my initial capital. The picture include is for 1 year. Not 1 month, the whole year

System look for a setup after a bar as touched the opposite Bollinger band.
- 250$ per trade.
- Slippage of 1.5 pip and commission of 2.5$ / trades included.
- Minimum 25 000 and Maximum 250 000 units.
- No scalping. Stop caculated ahead of time and adjust when needed
- No partial profit taken. All in all the time
- Over tuning ??? difficult to say. All parameters seems to make sens and as a logic behind

For sure, If I could reduce the rate of failure, the return will be better

My quest continu

I don't know why you wanna change anything. That looks like a pretty awesome equity curve to me.
However, I would suggest you to make a backtest for the last 10 years, to see how robust the method is over the long-term, also include other pairs.

Other than that, I don't see a reason to add filters. Mostly filters will just worsen a result, and not make it better anyways.

Also try changing the parameters slightly to see how much the results get affected.
If you want a method to work long-term, a variety of parameters should give you a positive expectancy, otherwise it's maybe just curve-fitting.

By the way, I just realized that you are changing the units you go in with from 25,000 to 250,000. If you wanna test a system and see if it has positive expectancy, you should just always use the same lot-size. So that the result doesn't get inflated.

test entries -> look after entry how often does price go up (or down for sell) after t time for a minimum of x pips ( x depends of the timeframe and t should be used as your average holding period). This will tell you if you have a lot of bad entries or exits.

Example if test shows that you have a 75% good entries (after you buy prices actually rises) but you win only 25% then conditions for exit are bad. If you have 30% good entries and it drops to 25% good trades the condition to enter needs to be evaluated if you want to perform better.

note: this test is pretty crude but it gives a good orientation when building MM within a sistem

Custos, nice point but:

This is the finess of my system. Positions size is adapted to the risk i take divided by the amount $$$ that i want to risk.

I know what I risk, but I don't know the result or gain from that risk. My rules allow me to stay as long as it is profitable.

Risk taken express in pip PAR= ( Entry - stop loss) + spread. ( PAR: Pip At Risk)
$$$ risked / PAR: 250$ / PAR will give me the Quantity of units to use

In my opinion, I need to know what i risk. I'm living in $, so I risk in $ Same position size with differents risk (# of pip at risk) is not for me

Do the calculation:
Risk: Where is your entry - Where is your exit ( stop loss) + spread

Same size with different risk
Size 100 000 Risk = 10 pip, you risk 100$
Size 100 000 Risk = 50 pip , you risk 500$

Now, If you calculate your position using $$$
PAR: Pip At Risk
$Risk 250$ PAR: 10 pip Size: 250 000
$Risk 250$ PAR: 50 pip Size 50 000

If your setup rules are accurate, then the logic behind it should work with any size.

Now, the question is: Witch is easier to reach after you go in the market with your setup: 30 pip or 150 pip. I will let you think about it and see what is the result of your position sizing

Are you living in PIP or with $$$.... That is the question

Martin

What currency pair(s) are you using? What tools do you use for your backtest?

If your system really has an R-ratio (avg. win / avg. loss) of 5 and Reliability (% wins) of 25%, then it's already an exceptional positive expectancy system, with a theoretical Kelly Optimal f of 10% risk per trade. (((5+1)x0.25-1)/5=10%.)

If that's really the case (I'm not convinced yet), you don't need to look for filters. You need to start forward testing it and verify that dreamboat backtest equity curve, while focusing on what drawdowns you could live with and still sleep well at night.

As it is, I would point out that your backtest is misleading. You are using a fixed $250 risk per trade, right? At the start of the test, that's 1% - 1.4% of your account equity; at the end, that's less than 0.3%. In the real world, that's actually a viable MM / position sizing strategy many successful traders with growing funds adopt, for sound psychological reasons. For the purpose of your backtest, though, there are a couple of related problems with that approach:

1. You assign less and less weight to trades as time goes on. You give up to nearly 5 times (1.4% / 0.3% = 4.7) greater weight to trades taken early in the year than to trades taken late in the year, based solely on when the trade occurred. That distorts the test results.

2. You equally dramatically minimize the relative severity of your drawdowns as time goes on. You start off with a fairly steep, although not unexpected, 30% DD, ~$8,000 right off the bat. That's not unexpected, given the initial 1% risk and your system's parameters.

Your backtest goes on to show a total of about 7 $5,000+ drawdowns, one every ~7 weeks and lasting up to a few weeks. Each DD is a result of a finite sequence of losses outnumbering wins by at least roughly a factor of 25. ($250x25-1,250x1=$5,000.) But thanks to decreasing % risk, each of the DDs gets smaller and smaller in % terms, making the equity curve look that much better.

By contrast, if you were to stick to, say, a fixed fractional MM with 1% risk per trade, all those drawdowns would be 20%+ and some would likely be 30%+. Could you live with such equity curve roller coaster? Whether your answer is yes or no, you want your testing (as opposed to real-world trading with growing funds) to treat all time periods equally in terms of relative, not absolute, degree of riskiness.

I'd run a batch of fixed % risk backtests, preferably over considerably longer than 1 year, if you can. Also might be useful to do the test by hand / manually for at least a sub-period and compare the two.

I'm using Multicharts and the pair is Eur.Usd

You are right about the comments about position sizing. I got my rules, but did not programmed it yet

Here is muy rule
Start with 1% or 2 % of the account Account: 25000 Risk:250$

When account as rised by 10% Account 27500 Risk:275$
Increased risk by 10%

When account as rised by 10% Account 30500 Risk:302$
Increased risk by 10%

Risk stay the same, even when account drop. Only modify if account drop by 20%. Then, Risk is lowerd by 10%

I will work on that position management soon and post the result
_______________________________________________________

I still need to concentrate on how to reduce my bad entry. As i often enter to go short on a rising market. Need to create rules using price action or indicator that would prevent that.

Forgot to show pictures

Increasing your % rate of profitable trades goes beyond just depending on technical, fundamental or automated trading system.

Years of live trading experience gives unflinching credence to the fact that what we see on chart are the after-effects of what we do not see on charts!

As such, to increase your % rate of profitable you must start trading like the banks do. You must take into account currency strength difference - the force behind profitable currency pairs!

This singular discovery has totally transformed my trading carreer!

Kind regards,

efxsignal

As mentionned in my post #13 , I did work on adjusting the Risk$ in relation to the new account Value

Result:
See the graph and compare to the previous one
Same rules, Only the $ risked is adjusted in relation to the new account value

Martin