List your Ninjatrader issues here

Discussion in 'Programming' started by EliteTraderNYC, Jan 24, 2013.

  1. I'm posting this thread to gather any known issues with this platform (and to help other developers with the issues I've had). Overall, I'm still giving it a neutral rating, but here are the major issues I've had so far:

    I had a problem with Ninjatrader erasing my profitable strategies. I found out it was reverting to an earlier saved version of the saved strategy upon restart. I think it may have been caused by a local problem with my .NET and Windows Vista installation so I reinstalled my operating system and everything on my computer. This seems to have worked - Now I am performing testing on multiple computers to make certain the results are correct.

    Splits and dividends. The system does not tell you that there was a split during the backtesting period so you have to go through and manually adjust each symbol that you are testing. This also skewed my backtesting results initially. I use kineticks as my data provider.

    Backup and restore. When you restore a backup, it only restores your strategies not your symbols lists. So these are lost.
  2. Its not on Linux. There's one...
  3. The settings on the backtest will inexplicably change on their own! Another gotcha, so hard code them in instead.
  4. downfall of figuring out shheeitt is that if you have an idea you wanna program in.. and you need help with it.. all of a sudden it isn't your idea anymore haha... jumping from the wisywig to coding indicators and the such seems relatively easy.. but then again.. all i've had is problems.. hard to blame that on ANY software haha..
  5. Here is a small list of issues. This is on top of NinjaFreeze crashing regularly.

    * Backtest has data issues. Gets data, if not gets bad results. EmaCross (sample strat), optimizing 3 times on same computer, same settings. 3 results. OUCH.
    * No tick data in backtest. This leads to concolusive code to make tick exact order handling (tick 1 on top o regular bars) which blows ninja memory limits.
    * Running totally outdated .NET version because "support of current systems" is down on their list right above "proper data handling".
    * Talk about errors. WHo was the idiot who writes errors into the log as "An error has occured". No message, no stack trace. Hey, if an error occurs in my strategy, you could log the standard info.
    * SLOW backtests. You have to see the speed of a event plaback that is optimized to believe it.
    * Standardbacktest borked. Bad fill calculation (not always applying splippage), bad analysis methods.

    The list really is long.
  6. yeah.. what would you expect... you quickly get the feeling its a platform to sell upon.. typical give away software hook deal going on... i've bogged the thing down many times .. but thats me putting the backtest algo to much to work with.. i actually like doing that to my computer.. haha make it hot.. Excel drives me insane as well and i don't evne know what i'm doing .. thing just crashes.. poof boom crack.. frankly i couldn't imagine writing a piece like this myself.. they get the visual appeal thing , and wisywig attraction thing going for sure...
  7. Well, we did our own non visual stuff and now an optimization runs in a cluster of computers that nicely split up the work. Database driven, most things are not kept in memory, so memory use is low, too. Even with just 6 computers (still test setup) we factor in a gain of about 40 to ninja at the moment.
  8. yeah... with SSD's and Sata 3 now.. the speeds on drives.. hyperthreading on the new i7's .. and memory is relatively cheap... haha i literally start raid zeroing SSD's and jamming memory in my built computer.. not that i don't care about the cost.. but custom building a computer or server is easy and cheap.. a little reading and asking around to my friends that are all in the field of storage arrays. clustering and every aspect of computing .. none of my friends are proficient programmers though.. i'm on linux now.. i visualize a windows box.. multiplex resources in a cluster sounds like the ticket though.. correctly configuring hardware is just the beginning of the deal.. picking a scalable and efficient storage engine. then picking a language to write it in, then a application framework.....usually guys just write things in the languages that are native to them.. no matter what the consequence.. we were sitting up at starbucks laughing about some software that prompted a "garbage removal" haha what a joke.. you mean you sell me software i have to take the trash out on? haha my best friend works at semantic on their clustering software.. my other friend is a Linux guy thats finally got me off windows.. Funny how this thread shows up right as i reinstall NT ... what else can i do.. to very quickly backtest some customer indicators in strategies.
  9. my list is long too....unfortunately I haven't been able to find a replacement....Multi-Charts doesn't really have what I "need" so I just keep my mouth closed and shut the fuck up
  10. i`ve only one.not enough money im pulling out of it.

    #10     Jan 25, 2013