Vast difference in Historical vs Live

[Burga1]

Greetings,

I developed a strategy using NT that I had been testing using the historical forex feed from Gain. When conditions permit trade entry, the strategy enters at 2 pips above/below the high/low (depending on long/short trades) and sets an initial StopLoss at 1 pip above/below the high/low...

Using the historical data everything seemed to be working as it should. So I decided to use that Gain feed to enable live "paper trading" using the demo account. The "live" trading results are dramatically different from the "historical"...among the differences:

1. Entry for trades many times "misses" the 2 pip requirement
2. StopLoss (INITIAL) are being set/activated upon at incorrect times
3. StopLoss RESETS (to simulate a "TrailStop") are not happening at all (as they do using historical info)
4. Trades and Stops are sometimes being activated at prices a bar doesn't even hit...how is that possible?

I'm attaching two screens to illustrate my concerns. The first screen shows Short trades (Condition 4 Entry) that were stopped out for some unknown reason (not when they should have) and never had the StopLoss changed as the price moved in the trades direction favor...
The second screen shows some points where either entry or StopLoss took effect even when the price never reached the point of execution (there are at least 3 little markers indicating this at points where the bar never reached...)

Again, none of these observed behaviors happen in the historical context...what could be happening here?

[Burga1]

Hi again,

Just to follow-up my original post, I went back on the charts and re-applied my strategies to them in what is now a "historical" context. And, not suprisingly...the results are dramatically different (much more positive). In addition, everything in the historical context is functioning properly...entry/initial stops in proper locations, proper usage of resetting (trailing) stops, etc...

Once again--why the "night vs. day" difference?

[NinjaTrader_Dierk]

There are multiple reasons for that. I suggest browsing the web to understand why backtest never matches realtime.

Some of them are listed here: http://www.ninjatrader-support.com/H...TimeVsBacktest

[Burga1]

Hello,

Thank you for posting. I understand there will be slight differences in trade execution when using "live" data vs. "historical". That does not explain, however, why StopLoss orders are submitted nowhere near where they should be, why StopLoss levels are not changed as the market moves in a favorable direction, or why entry/Stop prices are filled even when the bar in question never actually reaches those levels...

[NinjaTrader_Dierk]

You would need to debug your strategy to understand why it does not work as expected. Please see here:

http://www.ninjatrader-support.com/vb/showthread.php?t=3627

http://www.ninjatrader-support.com/vb/showthread.php?t=3418

[Burga1]

Yes, I know--I've spent weeks doing that already. Now that I had the strategy finally working as expected using historical information, you tell me I now need to go back and begin again from "square 1"...?

I would have thought that once a strategy is working using "historical" data it would/should perform at least very closely to "live feed" executions, if not quite exactly...that is why I made my original posting...it would seem there is something dramatically different going on between "live" and "historical"...things not explained by dynamic market execution.

[NinjaTrader_Dierk]

Order are placed at the prices you coded. NinjaTrader does not "add or subtract" anything. You would need to debug your strategy to understand why it triggers orders at particular prices.

[Burga1]

I have multiple strategies running on the same chart (example: EUR/USD)...is it possible they could be iterfering with one another?

Also, what does this line from the manual mean regarding StopLoss?:

simulated
If true, will simulate the stop order and submit as market once triggered

[NinjaTrader_Dierk]

>> I have multiple strategies running on the same chart (example: EUR/USD)...is it possible they could be iterfering with one another?
What do you mean by inferring but likely not.

>> Also, what does this line from the manual mean regarding StopLoss?:
Which help guide page are you referring to? Please paste a link.

[Burga1]

--I mean I have "strategy #1", "strategy #2"..."strategy # 6" all running at the same time on a chart...could they interfere in any way? I don't think so, but I'm trying to find some answers...

--From the online Help under the information for "SetStopLoss()"...no page number.

[NinjaTrader_Dierk]

- if you mean by "interfering" that one strategy changes the order price of an another then no, this will not happen
- this allows you to place simulated stop orders. Please browse the documentation to understand the concept of simulated stop orders in NT.

[Burga1]

Thank you. I got one more from the online Help, it doesn't seem to be explained extensively, it is from the section labeled "Running a NT Strategy from a chart":

Stop & target submission
Sets how stop and target orders are submitted

[NinjaTrader_Dierk]

http://www.ninjatrader-support.com/HelpGuideV6/helpguide.html?SubmittingOrders

[Burga1]

Thanks but I'm not sure that's it...please refer to screenshot.

[NinjaTrader_Ray]

Please use PerEntryExecution unless trading equities. This just determines if a unique stop order is submitted for each incoming fill of an entry order or one stop order for all fills.