Trade Performance Report is wrong...

[edstar]

Excellent work, look forward to seeing it,maybe post some screenshots. I've actually always thought that a major area which could be improved is the whole performance reporting with more metrics to aide in evaluating strategies. A starting point though is haveing a databse which correctly matches trades and shows the profit per trade.

[tgn55]

Thanks edstar...Another confession ;-) I have concentrated on functionality...at the expense of the GUI - which really needs a serious makeover. I'll try to clean it up and share some screenshots real soon. Although, having said that.. it was definitely some GUI aspects (skins) that cost me probably 3 months of fairly continuous effort... Anyway, more later... with pics!

Cheers,
T.

[howardtrade]

Any update or resolution to this?

[tgn55]

Kinda... but nothing good ;-(

A while back I discovered a flaw in my assumptions about how Ninja recorded things in their database, which I use to track trades. It's about what trades are recorded as entries. Bottom line is it threw a spanner in the works, and whereas I thought I was almost ready to release this thing... I can't in good faith do that whilst it sometimes creates a incorrect trade report. Most frustrating.

I've been trying to find time to revisit this, but, to be candid, I almost despair at sorting it out, unless Ninja provide some assistance understanding how they decide to mark a trade as an entry (or exit)

Sorry it has been so drawn out, but my original objective was always to provide a trade report that was ALWAYS reliable/correct... and now, I can only do it most of the time - not always. Which is not much better than Ninja's builtin report.

Finally, it is now clear that if your database file gets something out of sync... it stays that way forever (or at least, until you trash your database and start over) So one bad execution (or missing row) really screws things up.permanently. All subsequent executions will then be recorded with some inconsistency - at least that's what I see.

These two issues have resulted in this project going nowhere for a while. I expect to be talking to Ninja shortly, that should clarify if they can help.

Regards,
T.

[NinjaTrader_Jim]

There are some limitations that come up from pairing executions together to form trades for the performance report that have been discussed in this thread.

The post below summarizes some known limitations, what would be ideal (using an actual report from the broker, rather than relying on logging executions,) and what we would consider suspicious and what we would want to look into as issues with the current system.

https://ninjatrader.com/support/forum/forum/ninjatrader-8/platform-technical-support-aa/1095020-trade-performance-report-is-wrong?p=1107434#post1107434

If there are tests that we can set up that specifically break the performance report, we would like to look into that. These cases may be best illustrated by a custom NinjaScript strategy that takes just the trades necessary to break the report.

As for giving further direction for reading the data base and identifying entries and exits for your own Trade Performance AddOn, I don't have any more in depth information to share. I believe these to be fundamental difficulties with trade pairing.

[tgn55]

Thanks NinjaTrader_Jim , I always appreciate your input.

As you know, this has been quite a long-running saga, and truth is, I have done little on it in the last 6 months or whatever, when I sort of hit the wall. I am HOPING to return to it "soon", but I fear that without clarity regarding how/when Ninja records an execution as an entry or exit (by which I mean the IsEntry/IsExit fields in the db are set) I don't have much confidence in my cracking it.

I don't know if divulging that detail would be asking too much - I understand if it is considered internal IP. Last time I worked on this, I think I had a case where executions were recorded in the db as exit, then entry, (I think incorrectly) yet NT's Trade Performance tool managed to identify it correctly as a trade. That's about when I figuratively "gave up" !!

I understand that ultimately, in the absence of the broker's statement, there can never be a 100% guaranteed way to ID trades. This whole saga began, as you may recall, when I saw NT starting "trades" with an exit... purely due to the date which was used as the starting point of the report... and I maintained that that was dumb...and set out to prove I could do better. And it worked... albeit my initial reverse engineering of the db had some flaws.

So, here we are about 18 months later... and I just don't really know if I am on a fool's errand ;-)

To all other readers... I will try to bring this to a conclusion one way or the other (ie, succeed or admit defeat!) in the next month or 2. Apologies for the really long delays, after promising so much, but I will never release code that I know has inherent flaws.

Regards,
T.

[OldHatNoobCoder]

Found this thread while trying to figure out why my trade reporting is worthless. Sad to see it's an issue that has been going on for (at least) three years.

I skimmed through prior posts, and several of the problems discussed don't explain why I can't even get a single day to look right. For example...One day, one account, one instrument, one strategy. Started the day flat (always does).
Executions are normal. You can see these trades don't last long. None more than a minute.

Yet, the trade report randomly matches a few sets of unrelated executions to create four trades that didn't happen. This is just one day, controlling for all the discussed limitations, and it still gets it wrong. How?

[NinjaTrader_ChelseaB]

Hello OldHatNoobCoder,

May I confirm you are using 8.1.1.5 and reproducing with new orders?

To confirm, the Strategy Performance window is correct, and there have been absolutely no manual orders or orders from other strategy instances, is this correct?

[OldHatNoobCoder]

Hello OldHatNoobCoder,

May I confirm you are using 8.1.1.5 and reproducing with new orders?

To confirm, the Strategy Performance window is correct, and there have been absolutely no manual orders or orders from other strategy instances, is this correct?

In this case it is on an account that is only being traded in by an ATM strategy and nothing else. So the account performance is the strategy performance. Strategy performance can't be tracked for ATM strategy as you know.

[NinjaTrader_BrandonH]

Hello OldHatNoobCoder,

Thanks for your notes.

The current version of NinjaTrader 8 is 8.1.1.6 and contains fixes to the Trade Performance window.

Please update to the latest version of NinjaTrader by using the download link on the NinjaTraderarticle page below.

https://ninjatrader.com/support/requ...desktop-users/

After updating the platform, test placing an order with an ATM strategy attached and after the trade is entered and exited, open a Trade Performance window and generate a performance report for today's date, then view the 'Trades' Display on the Trade Performance window.

Does the order appear in the Trades Display of the Trade Performance window?

[OldHatNoobCoder]

Yes Brandon, this matching error seems to have been fixed, at least for my manual accounts. I haven't confirmed by running that strategy recently. I can test that out next week. Thank you!