I'm not sure if you realise this so I'll say it first. When I talk about people adding their eyeball criteria to the selection criteria - I'm not just talking about picking the candidate you like the most. I'm saying that by applying an eyeball filter, there will also be times where you have capital to invest, but you keep the cash on the sideline because you don't like the look of any of the candidates your system suggested. How can you say that doing this makes no difference whatsoever to the outcome? In one case you're 100% invested, in the other you have one trade worth of cash in the bank.
Still disagree
I've had 15 candidates picked each time I run a scan
Everyday I run it.
At the time I run the search I may have one in my portfolio
Exited so I only have one spot.
This is common in any of the 38 systems I have so far designed.
Yep, with you so far.
Picking the first middle or last chart which has the correct buy criteria
Satisfied will place you in the range from the mean.
Yes you will be in the range of your backtested results. But you can manipulate where in the range you end up by adding additional criteria.
You can't possibly tell wether a random selection or a chosen selection will
Outperform each other.
The selections ALL satisfy the buy criteria.
But you can tell if a random selection will outperform a chosen selection. That's why we wrote the system rules in the first place. We identified criteria that narrow the field down to the ones most likely to give better performance. If you then narrow the field even further with additional criteria, you're changing the outcome once more.
At some point infact most points you'll have to select a stock from those
In a list when you only have one slot to fill.
Using your grey matter or a dart won't guarentee which out performs which.
Agreed. Just as there is no guarantee that our system generated buy signals will outperform the ones it filtered out. But we know that over time our criteria will cause the results to skew in our favor. In the same way, using your grey matter (assuming you always look for the same thing) over a random selection will also cause a skew in the results over time, one way or the other.
If your results are under at the low end of your mean.
DONT EVER TRADE DISCRETIONARY!
If what you say is true then my discretionary picking of candidates had no affect on the outcome. So it has no bearing on my skill in discretionary trading.
Look Tech, I'm not trying to undermine you. You probably have more knowledge and experience in the area right now than I will ever have in my life. And if the two of us were to disagree I'd advise people to follow your lead, not mine. One of us doesn't understand where the other one is coming from. It's usually me.
:
Lets take Tech Trader for example. You have previously stated that there are additional eyeball filters you apply to Tech Trader after it spews out a list of buy signals. Why do you apply these additional criteria if they don't make any difference to the outcome? Because you know that they do make a difference. Any time you add an additional criteria, you change the outcome in some way.
Lets use a pseudo code example:
Backtest code:
Buy signal =
Price has risen 20% in the last 10 days.
Sell signal =
Price has fallen 20% from its high.
Real portfolio trading decision:
Buy signal =
Price has risen 20% in the last 10 days AND (eyeball test) the last bar closed in the top half of the range for the day.
Sell signal =
Price has fallen 20% from its high.
You're telling me there will be no difference between the results of the above two examples? Whether you program the buy criteria in, or add it in later as an eyeball test is irrelevant. It still changes the criteria by which the stocks get chosen and therefore changes the results.