Biggest Lesson Learned So Far

[dla133 10]

I have been trying to become a more successfull trader for just over 3 years. I still consider myself a beginner in this fine art. I want to pass on to other guys the biggest lesson I have learned so far in my short trading career. Many more experienced traders have all been trying to help me learn this lesson, but it has taken me nearly 3 years to fully grasp it. To become a profitable trader you need an approach that gives you more winners than losers. Doh, how stupid is that! This is an obvious statement to make to anybody new or experienced in trading. Please let me explain myself. I have been looking for the holy grail because I believed that all successfull traders had a secret indicator or fancy system which I didn't know about. This doesn't exist. You have to learn to trade the markets in such a way that you will have 50 losing trades and 51 winning trades. This can be as simply as buying at support and selling at resistance CONSISTANTLY and not taking trades that don't follow this simple rule. Having the discipline to truely follow this is really the hard part. I can't tell you how many times I have struggled with CONSISTANTLY sticking to this simple rule. What I have now learned is that NOT taking the trades that don't match these rules, and not chasing trades that I have missed has improved my trading by a big margin. Following your trading rules consistantly can be all the advantage that you need to become succesfull trader rather than trying to find the next super duper indicator.

 

Good luck with your trading in 2010.

Lee

[sevensa 10]

I don't agree that you need to have more winners than losers. There are plenty of methods that have more losers than winners that are profitable. What is imporant is how much you win on your winning trades and how much you lose on your losing trades. I think a bigger and more accurate lesson is that you don't need to have more winning trades than losing trades to be successful.

 

If you have 6 losing trades each losing $50 and 4 winning trades each winning $100 then you are profitable with only 40% trades profitable.

 

On the opposite side, you can have 6 winning trades making $50 each and 4 losing trades losing $100 each and you will have a losing method even with 60% profitable trades.

 

If you want to really learn an important lesson, google expetancy and risk management.

[ekshay 10]

Congratulations Lee.

 

Just like you were so moved by this lesson that you had to start a thread to share it ...

... there are many who were so inspired, they wrote books explaining "how to be a profitable trader"

... unfortunately, even BEFORE they actually became profitable traders.

 

Many more experienced traders have all been trying to help me learn this lesson,...

 

I am not denying the value of your post. It is a very important step.

But, I hate it when experienced traders hide their secrets from new traders.

 

I must warn you ... your search for holy grail will NOT stop, even if you grasped this lesson.

Just because your Win% is high, doesn't mean it can't be improved. :haha:

 

------

 

Let me share the proper First Big Lesson...

 

To become a profitable trader you need an approach that gives you more winners than losers.

 

Correct version:

 

"To become a Profitable Trader, you need an approach that makes you more Profits than Losses-&-Costs combined."

 

This is actually universally applicable to all businesses. To become profitable, a business needs to make more profits than losses & costs combined.

 

(Now, we can really say) ... Doh, how stupid is that! This is an obvious statement to make to anybody new or experienced in trading.

Please let me explain myself.

 

In trading, your Win-Loss Ratio is HALF the equation. The other half is your Risk-Reward Ratio.

 

Or putting it in plain english ...

 

"Trading is not about being right or wrong ... but how much you make when you are right, and how much you lose when you are wrong."

 

Think about that. If you had 72% losers of average 10 ticks, 8% breakevens, and 20% winners of average 60 ticks ... wouldn't that be profitable too?

 

Now, its an entirely different discussion as to whether it suits you psychologically or not.

 

-----

 

Following your trading rules consistantly can be all the advantage that you need to become succesfull trader

 

This statement is actually entirely correct. But as usual, its only half the Lesson.

We can't jump 2 Big Lessons right now, but suffice to say that...

the statement does NOT explain how to judge if a concept is worthy of being a consistently applicable trading rule.

 

Good luck with your trading in 2010.

[Tams 20]

...

I am not denying the value of your post. It is a very important step.

But, I hate it when experienced traders hide their secrets from new traders.

...

 

 

There are always people who's willing to share...

 

you just have to push the right button.

[dla133 10]

Hi,

 

Maybe I didn't explain myself very well. By stating that you need to have 51 winners and 50 losers to be profitable. I was just trying to highlight the fact that no matter what system or approach you take it will ALWAYS give you losers as well as winners. What I should have said is that you only need a CONSISTANT 1% overall gain out of 100 trades to be profitable. That is why I have stopped looking for the HOLY GRAIL and I am instead just looking to increase my overall winning percetange from 1% to 2% to 3% etc. I believe if I had understood this at the beginning, I would have concentrated more on sticking to sound principles (even when I suffered losses over the short term) and learned discipline rather than skip from one method to another searching for something that worked all the time. From my poor explanation of what I consider a fundermental lesson, all new traders need not worry about me trying to push any kind of trading seminar in the future. I would probably confuse myself!

 

All the best for 2010

Lee

[dla133 10]

Ekshay has definately explained it better than I did.

 

Thanks Ekshay

[thalestrader 14]

I don't agree that you need to have more winners than losers. There are plenty of methods that have more losers than winners that are profitable. What is imporant is how much you win on your winning trades and how much you lose on your losing trades. I think a bigger and more accurate lesson is that you don't need to have more winning trades than losing trades to be successful.

 

If you have 6 losing trades each losing $50 and 4 winning trades each winning $100 then you are profitable with only 40% trades profitable.

 

On the opposite side, you can have 6 winning trades making $50 each and 4 losing trades losing $100 each and you will have a losing method even with 60% profitable trades.

 

If you want to really learn an important lesson, google expectancy and risk management.

 

Excellent post, Sevensa. Nothing helped me along more than when I came to appreciate the concepts you so clearly set out here.

 

Too many focus on winning. What they need to focus upon is making sure that when they win, they win at large enough multiple of their losses so that over the long haul, profits outstrip losses. If I had to lose $1000 50 times to win 100K once, I'd be quite happy with that winning percentage. Of course, I may not have the stomach to sit through that draw down. But th epoint is the point.

 

 

Best Wishes,

 

Thales

Edited January 16, 2010 by thalestrader

[jonbig04 10]

Excellent post, Sevensa. Nothing helped me along more than when I came to appreciate the concepts you so clearly set out here.

 

Too many focus on winning. What they need to focus upon is making sure that when they win, they win at large enough multiple of their losses so that over the long haul, profits outstrip losses. If I had to lose $1000 50 times to win 100K once, I'd be quite happy with that winning percentage. Of course, I may not have the stomach to sit through that draw down. But th epoint is the point.

 

 

Best Wishes,

 

Thales

 

Couldn't agree more. Accuracy should only be discussed in the context of risk/reward. My accuracy goal is 33%.

 

all just my opinion of course.

[Kiwi 16]

There is a lot to this stuff and it requires a bit of thought to see it. The OP expressed one view, and then improved upon it with the following posts. This is about one part of the game, money management, and I'd just like to hint/touch on the complexities so that the OP can do the research. I'll ignore costs (spread and commission) just to keep it a little simpler.

 

It is correct that what you want is a profit per trade. So ... 35% wr, 50% wr, 60% 80%

 

All can give a profit per trade. Profit factor = Won money divided by Lost money. At 35% you need an average win 3.7 times the average loss for a PF of 2.0. At 80% you need only 0.5 times the average loss for a PF of 2.0 (yes, that is right, win size half the loss size).

 

But PF is not a good enough measure. Expectancy is a better measure. It is (the Won Money minus the Lost Money) divided by the Risk per Trade. Why is it better? Because if you bet x% of your account on each trade and have yyy trades per year then you can use it to calculate your wins for the year. Its also better because you can have a decent profit factor and a low expectancy.

 

In the two examples above a 35% pf 2 system has an expectancy of 1.56 so for every dollar you risk you get the dollar back + 56c. The 80% pf 2 system has an expectancy of only 1.17 so for each dollar you risk you get the dollar back + 17c.

 

Hmmmm ... this makes the 80% system sound not so good.

 

 

But. What about bet sizing. Should you bet as big on the 35% system as on the 80% system.

 

As it happens the answer is a resounding NO!

 

Risk of Ruin (running out of money). With a 35% win system you are much much more likely to get a long string of losses than with an 80% system. Using the kelly ratio as a proxy for that risk. The kelly at 35% was 17% odd and the kelly at 80% was 40% odd so you could bet more than twice as hard on a 80% system as on a 35% system with the same risk of ruin.

 

When you do that your expectancy on the 80% system roughly doubles to a much more respectable 1.35 relative to the 35% system.

 

Confused? Well, you probably should be because money management and bet sizing is a little complex. But, if you're going to figure out how hard you should press an edge (entry, stop, capture process) then you will end up having to figure it out.

 

 

 

For the record: I like an accuracy goal near of at least 45% ... its very hard to take long strings of losses. Entry and exit design for 45% systems (usually trend capturers) is very very different from that for 80% systems. Thales, for example, like many discretionary traders errs to the high side and will try to take a portion of his profits in the very high win ratios to give smooth and easily traded equity curves. He may keep a portion for a longer run but with low probability of doing better than break even on that part.

Edited January 16, 2010 by Kiwi
to add "for the record"

[jonbig04 10]

I have to be honest kiwi, that's all a bit too confusing for me.

[Kiwi 16]

Thats what I mean. It takes reading then experimenting a bit with numbers to make sense of it.

 

Then of course, I forgot to mention the frequency with which the opportunity arises. Would you rather have a setup that gives you an expectancy of 10 once a month or a setup that gives you an expectancy of 1.3 three times a day (lets assume you have to watch the charts the same extent for both)?

 

If you give me an answer I'll show you how to do the calculation - lets assume that you can risk 1% with a very low risk of ruin (1 in 10,000).

[jonbig04 10]

haha well the 1.3 of course.

[BlowFish 10]

 

Risk of Ruin (running out of money). With a 35% win system you are much much more likely to get a long string of losses than with an 80% system. Using the kelly ratio as a proxy for that risk. The kelly at 35% was 17% odd and the kelly at 80% was 40% odd so you could bet more than twice as hard on a 80% system as on a 35% system with the same risk of ruin.

 

Nice to see someone else mention RoR, possibly the most important metric you might consider. I thought I was the only one that bangs on about it. The primary reason to look at stats for expectancy (assuming you are profitable) is to determine the %chance of going bust with different bet sizes. RoR.

 

The big advantage of systems that have high % winners (at the expense of reduced RR) is that they have far smoother equity curves and so a much lower chance of a bad streak wiping you it (lower RoR).

 

This is why what I call 'old fashioned trend following systems' (e.g. the turtles) which have maybe 30% -35% winners require lots of capital to trade.

 

Incidentally improving your % winners is (imho) possibly not an effective use of your energy (if making more money is your objective). Increasing size (obviously within your risk tolerance as calculated by RoR) will always yield far more. If you have a plan that you believe in but is not yet profitable you need to discover if it is due to being a slight flaw in the plan or flawed execution. If the former stop trading while you fix it.

 

If you want more stability, diversification (trade more instruments, different time frames and even different methods) will help that. That's why the old school trend followers would always trade a dozen or two instruments. All you need is one or two in good trends and you can be being chopped to death in others and still be OK.

 

having recently re-read market wizards it was interesting that a high percentage of these guys where 'old fashioned trend followers'.

[TradeRunner 10]

In the two examples above a 35% pf 2 system has an expectancy of 1.56 so for every dollar you risk you get the dollar back + 56c. The 80% pf 2 system has an expectancy of only 1.17 so for each dollar you risk you get the dollar back + 17c.

 

Hi Kiwi

 

How did you get these numbers for expectancy?

 

I can't find expectancy as defined in you post but did find this one...

 

E = Expectancy

P(w)= Probability of winners

S(w)= Average winner Size

P(l)= Probability of losers

S(l)= Average loser size

E = [P(w)*S(w)]-[P(l)S(l)]

 

So with 35% pf 2 doesn't that give

P(w)=0.35

S(w)=130

P(l)=0.65

S(l)=65

E = 3.25

 

TradeRunner

[DugDug 10]

"Incidentally improving your % winners is (imho) possibly not an effective use of your energy (if making more money is your objective). Increasing size (obviously within your risk tolerance as calculated by RoR) will always yield far more." - Blowfish

 

100% agree. I believe that this is an important but often forgotten element of understanding money management and position sizing.

- understanding the risk v reward ratios, positive expectation

- risk of ruin

- knowing how to manipulate the same risk reward parameters to incorporate the personality of the person based on their risk tolerance and risk of ruin.

 

the first is up to the strategy itself and the traders ability to follow it, the second will not change, it is about maths, ONLY the third is entirely up to the trader.

This is one of the reasons why when a lot of systematic traders know all the parameters of their system talk about "heat", and that money management and position sizing is the most important element of running a trading book - more so than the entries and exits.

[Kiwi 16]

Hi Kiwi

 

How did you get these numbers for expectancy?

 

I can't find expectancy as defined in you post but did find this one...

 

E = Expectancy

P(w)= Probability of winners

S(w)= Average winner Size

P(l)= Probability of losers

S(l)= Average loser size

E = [P(w)*S(w)]-[P(l)S(l)]

 

So with 35% pf 2 doesn't that give

P(w)=0.35

S(w)=130

P(l)=0.65

S(l)=65

E = 3.25

 

TradeRunner

 

That is almost the same as mine (but not as good (I'll explain why in a second)).

 

Same:

P(w)*S(w) = Won Money

P(l)*S(l) = Lost Money

and E = 3.25

 

That sounds good! A good strategy, right?

 

 

but, a better definition of expectancy is "How much do you expect to win if you risk $1"?

 

To find out "how much will I win if I risk $1.00" you have to divide ((Pw*Sw)-(Pl*Sl)) by the amount you risk each time.

 

So for your stats First Result = (0.35*130)-(0.65*65) = 3.25 ... but how much was risked to earn an average of 3.25 per trade?

 

Well, at the least the risk is the average loss (I use max loss in testing systems).

 

So in this case E = 1 + [ ((0.35*130)-(0.65*65))/65 ] = 1.05

 

So, at best, you only get $1.05 per $1.00 risked. I like strategies that return better than 1.35 per dollar risked so a strategy that sounded good at 3.25 is unclothed at 1.05.

 

 

I guess one is "What do you expect per trade?" but I prefer the other one "What do you expect per dollar you risk?" Horses for courses. I don't recall where I finally got that equation ... perhaps tharps money management book ... possibly on the net.

[Kiwi 16]

Thats what I mean. It takes reading then experimenting a bit with numbers to make sense of it.

 

Then of course, I forgot to mention the frequency with which the opportunity arises. Would you rather have a setup that gives you an expectancy of 10 once a month or a setup that gives you an expectancy of 1.3 three times a day (lets assume you have to watch the charts the same extent for both)?

 

If you give me an answer I'll show you how to do the calculation - lets assume that you can risk 1% with a very low risk of ruin (1 in 10,000).

 

On this one, jon got it right but its closer than one thinks.

 

So, annually, what is each worth if you could risk 1% on each trade and compound it?

 

The formula is:

 

(1 + (fraction risked * (expectancy-1) )) to the power of number of repetitions

 

Fraction risked = 0.01 (1% as a fraction)

Expectancy-1 = 9 for the first and 0.3 for the second

Number of Repetitions = 12 (1 per month) for the first and 750 (3 per day) for the second

 

So if we took no risk every day

Result = (1+(0.01*0))^250 = 1 .... no change

 

So if we took 1% risk every month for a 10 expectancy (get $10 back each time we risk $1)

Result = (1+(0.01*9))^12 = 2.81 so at the end of the year we'd have 2810 if we started with 1000

 

So if we took 1% risk 3x a day for a 1.3 expectancy (get $1.30 back each time we risk $1)

Result = (1+(0.01*0.3))^750 = 9.45 so at the end of the year we'd have 9450 if we started with 1000

 

Much better?

 

But what would have happened if we'd taken the risk only once per day.

 

So if we took 1% risk 1x a day for a 1.3 expectancy (get $1.30 back each time we risk $1)

Result = (1+(0.01*0.3))^250 = 2.11 so at the end of the year we'd have 2114 if we started with 1000

 

 

Frequency is important