Are trader's skills really important?

Candle, I'd prefer a guy who can do both - pulling the trigger and applying sound planning / risk-management.

The pure "pulling the trigger" guys with no trade-management skills lead to the recent developements concerning daytrading in margin accounts and to a few people shot to death, as I recall correctly. Let alone the many who lost all of their money in trading. This certainly didn't happen because of their great skills in risk / money-management but by pulling the trigger without thinking first.

Just my humble opinion.

Adding the right mindset to skills is certainly the best trading approach.

However, if one has the skills, he should also have the right mindset already, otherwise, how could he have developped the skills ?

I'm not talking about knowledge here - knowledge and skills are 2 different things . You can teach knowledge, but skills ( or experience ) only come from applying knowledge.

regards
 
IF market prices are not random THEN right skills ( or trading systems ) are important in trading, and the only thing is to build them and follow them. IF market prices are random THEN - i'm afraid - traders don't have much chances of success in the long term. ( I'm talking about traders who make a lot of trades, if someone made less than about 200 trades - his result can be affected by luck ).

I've tried to check the randomness of Nasdaq100 index ( end-of-day data ). In 1324 sessions ( from 16 July 1996 to 16 Oct 2001 ) the average point difference between sessions was 41.20.
The range of NQ100 index in that time period was 4106 pts.

Now, if a coin flip will decide about price advance by 41.2 or decline by 41.2 , 1324 times in a row, what's the average price range can we have in that period? I made a simulation in excel, the result is: 2238 pts. The biggest range, that simulation produce was 4665 pts.

So, NQ100 index prices had more longer trends than random-walk. Or - the probability of move in the same direction as in the previous session is greater than 50%.

But! there was a time, when my simulation produce a price range of 4665 pts, which is greater than NQ100 range. Both, random prices, and real NQ100 prices have the same number and average value of single moves.

It's all about probabilities. There's a small chance, that in the last 5-year period NQ100 index prices were moving just just like random prices. I have to check this issue on other indicies.

DT-waw
 
Privateer: I think that luck and probability come together. If you make 5 trades, impact of luck on your final P/L is greater than with 500 trades. It's difficult to count an impact of luck. But it exsist. Each strategy works on some certain market behaviour. Let's say your strategy made good results in the past. It's not because your strategy is "good", it's becasue the behavior of market in that past period makes your system good. You can't predict market behavior in the future periods. Maybe it will be similar to the past one, maybe not.

I know that commissions and slippage can damage the result of positive expectancy systems. That's why trading is so difficult - even if you have great system, it doesn't mean that in real trading including commissions, slippage the system will make positive result.

Now: I've tested SP500 index prices in the same way as NQ100 ( described earlier in this thread ). Results are similar. 1302 random moves can look like S&P500 chart ( the same range, patterns, etc. ) for the last 1302 sessions.

To verify random-walk theory, you must test large amount of data. And of course, that will only be some probability, that this theory describes nature of financial markets or not.
 
DTWAW
I wonder, why you consider the numbers of 200 or 5000 trades as a qualifier for a system of being affected by luck or not ( or as being valid for that matter ).

A trader trading QQQ only on weekly signals of a simple MA crossover system ( completely mechanical ) + rigid stop-loss strategy might have made more money over the last 6 years with much less trades than a daytrader making 5000 trades in QQQ over this timeframe

Multiday swingtraders may do anything between 10 - 100 trades a month and might make more profit that any scalper doing 10 times the number of trades.
Of course, you'll find also scalpers who made quite some money ( see Hitmans reports about some of his colleagues ). But most probably, they'll also apply a sort of system ( i.e. simply tight stops.)

Sector-traders who trade strength and weakness of industry sectors and who are in a particular trade for several days / weeks do much less trades than any day/ swingtrader and still might come up with the same or more profit by applying their system.

The sheer number of executed trades says nothing about reliability / profitability of a mechanical system versus luck.

I'd even go so far to say, that no trader executes 5000 trades without having a system ( whether it's profitable or not is another question )

It has to be seen in the context of the whole trading - setup and style.

Just because someone made 5000 trades doesn't merely qualify him or his system in the first place for anything else other than that his broker loves him ( high probability for this, at least ).

As mentioned already, I doubt, that a trader makes 5000 trades without using any kind of a system, so the factor "luck" is diminishing again.

You can test that yourself fairly easy. Just sit down in front of your RT system and paper-scalp a liquid stock you haven't traded yet all day long with a random choice of entries and exits, random selection of positionsize per trade, relying on plain luck only ( Hard to do, since after a while you will automatically try to use a kind of system to determine entries and exits and you'll not anymore rely on plain luck ).

Do this with 10 or 20 stocks from different sectors for a couple of days each and you have a huge number of trades and a lot of experience to draw conclusions from.

You' ll probably endup with a 100 or 200 trades a day but nothing to show for in terms of profit after comissions + slippage unless you apply a kind of trading-system - which means that from a certain moment on, luck is not anymore the dominant part of your trading.


regards
 
I give an example. You have 2 systems:

1)
# of trades: 10
% of profitable trades: 50%
Av profit: 2.33
Av loss: 1.00
Total P/L : 6.65

2)
# of trades 100
% of profitable trades: 50%
Av profit: 2.33
Av loss: 1.00
Total P/L: 66.5

What's the probability that the first system will have P/L below zero in the next 10 trades ( opened and closed with the same system rules )?

Expected P/L in the next 10 trades is 6.65
So, expected % of profitable trades is 50% in the next 10 trades, with av profit 2.33 av loss 1.00.

To answer the question above, we should count what's the probability of making some % of profitable trades ( or less ) with which the total P/L is zero ( or less ), in the next 10 trades, with 50% chance of a single success.
With 30% profitable trades ( av profit 2.33 av loss 1 ) the total result is zero.

Number of shots: 10
Number of successes: 3
Probability of a single success: 50%

Result: probability of 3 or less successes in the next 10 shots with 50% prob. of a single success is 17.187%

Now, I want to know what's the probability of P/L below zero for the second system.

Number of shots: 100
Number of successes: 30
Probability of a single success: 50%

Result: probability of 30 or less successes in the next 100 shots with 50% prob. of a single success is ... almost zero.

Of course this method assumes, that if you have some result in x number of shots, then the expected result for the next x shots is the same as you achieved. If you disagree with this assumption, you have to assume, that expected P/L for every system in any future # of trades or periods is zero.

And again, if you assume, that expected P/L for any system is zero, then you can count what was the probability of your system's to achieve its results. For example 2nd system presented above. If expected P/L is 0, then it is absolutely impossible to make 50% profitable trades ( from total 100 trades ) with av profit/loss ratio 2.33

All this is EXCLUDING commission costs. Of course, commissions will depreciate profits or apreciate losses of any system.

My point is: to verify trader's skills, or system performance, you have to rely on a big number of trades.

DT-waw
 
Commission costs are completely different story. If your system has:

# trades: 20
# of profitable trades: 10
# of unprofitable trades: 10
av profit: $20
av loss: $10

Commission costs per round-trip: $8

then, net P/L is negative. $200-$100-$160 = -$60

System with fewer trades, but with higher stops will have for example:

# trades: 10
# of profitable trades: 5
# of unprofitable trades: 5
av profit: $40
av loss: $20

Commission costs per round-trip: $8
net P/L : 200-100-80 = +$20

Level of stops ( and your trade time period ) should include the impact of commissions. Tight stops and relatively low profits, big number of ultra-short-term trades can damage your result due to commissions costs.

But big number of trades is necessary to verify system's performance.

DT-waw
 
Originally posted by DT-waw
It's difficult to count an impact of luck. But it exsist. Each strategy works on some certain market behaviour.
More...
there! you said it. i think we need to distinguish between two "types" of luck:

1. being lucky to end up with trades that - on average - produce greater profits than the true expectancy of your system/approach.

2. being lucky to trade during market conditions that are favorable for your system/approach.

it is trivial to account for (1) -- after a sufficient number of trades, it is safe to assume that this is 0.

however, accounting for (2) is much more difficult if not impossible. hence it is always near impossible to be sure whether trader's performance can be attributed to skill or luck of the "2nd type".

Originally posted by Privateer
The sheer number of executed trades says nothing about reliability / profitability of a mechanical system versus luck.
More...
yes it does. "luck" is by definitition something that will cancel out in the long run. therefore - all other things being equal - one should always prefer the system/approach/fund manager with more trades under his belt.

- jaan
 
It's all about probabilities. There's a small chance, that in the last 5-year period NQ100 index prices were moving just just like random prices. I have to check this issue on other indices.

I'd say that's a very small chance, which might be more obvious to those who lived in the US during that time and saw all the hype and 'e-mania' up close. There was clearly an increase in demand for NASDAQ 100 stocks for some time.

Now, of course, the increased demand could have been the explanation for the increase in NDX value, while at the same time the short term price moves could have been random ( I don't believe that myself.)

As far as the 100 trades vs. 5000 trades, DT-WAW is essentially saying that if you flip a coin 10 times the probability that you'll get heads at least 70% of the time is much higher than if you flipped that same coin a 100 times, which is of course true. So, the smaller amount of trades you have done, the less likely it is that you have a consistent, winning system and are not just relying on luck.

voodoo
 
Originally posted by jaan

i think we need to distinguish between two "types" of luck:

1. being lucky to end up with trades that - on average - produce greater profits than the true expectancy of your system/approach.

2. being lucky to trade during market conditions that are favorable for your system/approach.

however, accounting for (2) is much more difficult if not impossible. hence it is always near impossible to be sure whether trader's performance can be attributed to skill or luck of the "2nd type".
More...

Uuuh... so nobody knows how high is the 2nd type of luck. everybody agree that 2nd type of luck exist?

We are getting close to say, that it's impossible to verify weather any trader's strategy or trading system results are made by luck or skills!!!

Voodoo - thanks for clarification. :)

I believe that there's some statistical method of verifying the impact of two types of luck.
We must assume that, expected result of any system is zero. The method I describe at the beginning of this thread is good, I hope so.

There's another interesting thing. Maybe it's possible to discover what type of market behavior is the most probable?
1. less "trendy" ( i don't know how to say that: market with trends ) than random
2. random
3. more "trendy" than random

It's a simple function:
1. probability of trend continuation <50%
2. probability of trend continuation =50%
3. probability of trend continuation >50%

I bet on the 3rd one. :p
Economy changes in trends, people are over-reacting, more and more things depend on each other in the economy. there're the reasons for trends.

DT-waw
 
Economy moves in trends. GNP, inflation, unemployment numbers move in clear trends. But financial markets... If majority of money is traded with a direction of economical ( political, financial, etc. ) trends, there will be no ( or not sufficient ) money on other side of a trade, the number of price gaps will increase, and we will have the efficient, random market.

That's the advantage of professional trading. Quick reaction to the news/events. Real time news, quotes, and ability to enter a trade with a blink of an eye.

Since there're always be traders with unefficient tools and/or with no positive expectancy system or strategy - other traders will have a chance to win.

But, you never know what happens. Maybe 21st century economy will be more stable? :confused:
 
You must log in or register to reply here.
Back
Top