MultiCharts Easter Sale has jumped in! Up to 50% off Explore offers
+1 888 340 6572
MultiCharts Project Management
previous_open_issue.png
Go to the previous open issue
previous_issue.png
Go to the previous issue (open or closed)
star_faded.png
Please log in to bookmark issues
bug_report_small.png
Open Bug report MC-2544

Outlier being calculated incorrectly

action_vote_minus_faded.png
0
Votes
action_vote_plus_faded.png
next_issue.png
Go to the next issue (open or closed)
next_open_issue.png
Go to the next open issue
Description

In the Strategy Report, the Outlier definition currently uses the Average trade and the deviation from it.

I do not understand how this is still the case at V12 of MC when it is clearly wrong.

To discover outlier trades, trades that contribute a greater profit or loss than normal, one MUST use Average WIN and Average LOSS separately.

For those who hadn't given it much thought, here is a quick comparison to highlight how critical it is to get it right.

System 1 has only small wins ($100) and losses ($100)
System 2 has only large wins ($1000) and losses ($1000).
Let's assume both have exactly the same Average Trade of $100.

Currently MC determines an Outlier by reference to the $100 Avg trade. So clearly System 2 has 100% of its trades as outliers, and System 1 has NONE.
CLEARLY ABSURD. Both System 1 and System 2 has precisely ZERO OUTLIERS. Every single trade is precisely in line with the rules for taking a profit or loss.

Traders want to know what % and how many trades resulted in such large profits, that a single trade might have provided way too much of the total profit.
In other words we want to compare trade outcomes to the average trade outcome, not of all trades, but only WINS or only LOSSES. How does our winning trade compare to the average winning trade.

So in our example System 1 made $100 on every winning trade. ZERO outliers. System 2 made $1000 on every trade. ZERO outliers. Since the average win = $1000. But if we use Average TRADE of $100, then every win of $1000 is an outlier.

I hope my explanation isn't too confusing and in fact obvious. It is extraordinary that such a grave error has not been detected back in V1 or V2 of MC or whenever this functionality was added.

Please VOTE to have this corrected. Anyone who does backtesting MUST vote on this issue to be corrected.

You can find Outlier data under the Menu View/Strategy Performance Report/Trade Analysis/Outliers.
Note that it refers to 'Avg Trade'. This needs to be split into Avg WIN and Avg LOSS, along with separate Standard Deviation calculations based on each of those 2 numbers.

Steps to reproduce this issue
Nothing entered.
Comments (3)
#1
user-offline.png  Zheka (Zheka)
Jan 12, 2019 - 17:48

Man, calm down:-) ....and study statistics ....and MC Outlier report.

  1. Sys1 in your example cannot have an Ave.Trade of $100 - it should have no losing trades for this to be true.

  2. "So clearly System 2 has 100% of its trades as outliers" - No. What will 1 Std.Dev be for System2?

  3. MC Outlier page reports exactly what you are talking about, clearly separating Winning Trades (under Positives) and Losing Trades (under Negatives).
    Each has a different Avg.Trade, different Std.Dev and other calculations.

Where this logic has to also be applied is in calculation of Select GP/GL on the summary page.

It is not clear if outliers for these are calculated off Total Ave.Trade or off Ave.Win and Ave.Loss.

@MC,
how are these metrics really calculated?

#2
user-offline.png  AdrianP (AdrianP)
icon_reply.pngJan 13, 04:05, in reply to comment #1

Calm down?? Sounds like you are out of touch or out of sync with the simple message I expounded, and its meaning Zheka.

I have also studied statistics at university level. Have you?

PLEASE...can I ask that only professionals who understand what is being discussed, respond to this thread. Otherwise we waste time getting distracted and misdirected.

Zheka....Sys1 ..yes ZERO losers. Giving an avg trade of $100 and a StDev on Wins only of ZERO, and a StdDev on Losses of ZERO. Clearly not realistic, but that's not the purpose of the message.

And for Sys 2 under CURRENT coding, and assuming only $1000 wins, then obviously the Stv Dev will be above zero, when in fact the std Dev of the winners IS ZERO. Not rocket science, certainly as basic as statistics gets.

The MC Outlier page does NOT report what I suggest it should. You accuse me of not understanding statistics, and yet you show a 100% failure to comprehend my discussion. In fact you don't even read what it says on the Outlier page "Avg.Trade + 2 Std Deviation'... You do understand what avg trade is right? And you have confirmed it actually is avg trade, being an average of all winners AND losers. Not Avg winners nor avg losers. I don't think the MC programmers could make it clearer.

So I am 99.9% confident in what I have originally stated in my message.
Feel free to apologise at any time Zheka.

It would be great if someone from MC support can acknowledge this initial message, so that we may move forward.

'''Zheka wrote:'''

Man, calm down:-) ....and study statistics ....and MC Outlier report.

  1. Sys1 in your example cannot have an Ave.Trade of $100 - it should have
    no losing trades for this to be true.

  2. "So clearly System 2 has 100% of its trades as outliers" - No. What will
    1 Std.Dev be for System2?

  3. MC Outlier page reports exactly what you are talking about, clearly
    separating Winning Trades (under Positives) and Losing Trades (under
    Negatives).
    Each has a different Avg.Trade, different Std.Dev and other calculations.

Where this logic has to also be applied is in calculation of Select GP/GL
on the summary page.

It is not clear if outliers for these are calculated off Total Ave.Trade or
off Ave.Win and Ave.Loss.

@MC,
how are these metrics really calculated?

#3
user-offline.png  AdrianP (AdrianP)
Feb 09, 2019 - 15:05

Just following up on this issue as well.

How do we genuinely know when an issue is 'under review'?
As that status appears immediately upon posting an issue.
And based on many peoples posts, can stay like that for YEARS.

Guys @MC can we fix this please.

As for the Outlier issue, How are bug fixes prioritized? This is probably a 5 min coding issue. YEP JUST 5 minutes.
How many programmers does MC employ? Just trying to get some perspective, as clearly time costs money, and there are always issues to be fixed or new ideas to be added.

But I never see any feedback here?

History
Issue basics
  • Type of issue
    Bug report
  • Category
    Not determined
  • Targeted for
    Not determined
  • Status
    Under Review
  • Priority
    Not determined
User pain
  • Type of bug
    Not triaged
  • Likelihood
    Not triaged
  • Effect
    Not triaged
Affected by this issue (0)
There are no items
People involved
Times and dates
  • Posted at
  • Last updated
Issue details
  • Reproducability
    Not determined
Attachments (0)
There is nothing attached to this issue
Commits (0)
There are no code checkins for this issue
Duplicate issues (0)
This issue does not have any duplicates