How Could a Trading Simulator possibly be accurate? | Forex Factory

Options

Permalinks
Print Thread

Similar Threads

VHands Trading Simulator 428 replies

Trading simulator or practice tool 10 replies

Trading simulator 5 replies

Renko charts in the trading simulator 9 replies

Which Trading simulator? 2 replies

How Could a Trading Simulator possibly be accurate?

  • #1
  • First Post: Jul 27, 2009 12:13am Jul 27, 2009 12:13am
Hello

Just have a question about the trading simulator V Hands - or any other trading simulator for that matter...

Meta trader only collects tic data in the form of how many tics are in a certain bar (volume). The trading simulators then use this "volume" data from the M1 timeframe along with the open high low and close data to then "replay" the market from previous days...

What I don't get is how can the simulator be ticking up and down on each bar accurately when (as far as I know) meta trader doen't collect the acutual price value of each individual tic...

So would I be right in saying that when using a simulator that using "each tik" as oposed to using "open prices only" is pointless?

Or am I missing something here?

What I mean is - Is each tick on trading simulators the way actually happened tick to tick? I dn't see how this is possible

Hope this makes sense

Cheers
Time hides Nothing
  • #2
  • Jul 27, 2009 12:26am Jul 27, 2009 12:26am
I'm not familiar with the Simulator function, but certainly using the back-tester on 'every tick' mode is epic folly which has been well documented, basically you end up with a massive over estimation of the volatility within a given bar. In short you are correct, it is inaccurate.
The breaking of a wave cannot explain the whole sea.
  • #3
  • Jul 27, 2009 1:05am Jul 27, 2009 1:05am
Quoting CindyXXXX
Hello

Just have a question about the trading simulator V Hands - or any other trading simulator for that matter...
IMO it also depends on trading TF. If I'm trading H1 candle patterns, for example, and looking to ride moves of (say) 50-100 pips, then M1 OHLC (let alone tick data) is easily accurate enough for my purpose.

My view is that 'randomness' in price movement makes trading a game of approximations. Probabilities can not be precisely calculated (as they can with casino games, for example), and I believe that some traders spend way too much time trying to optimize their systems to the n-th degree. I realize that everybody's approach is different, hence I try to exercise some caution when generalizing, but I believe that a robust edge needs to somehow have depth beyond 'mere numbers'.

Having said that, back-testing can definitely be useful in giving the trader a reasonable feel as to how profitable a system might be, and a simulator gets one a little closer to the action than simply analyzing performance reports. That's especially relevant if one's trading style is more 'discretionary' in its nature.
  • #4
  • Jul 27, 2009 1:22am Jul 27, 2009 1:22am
Quoting hanover
IMO it also depends on trading TF. If I'm trading H1 candle patterns, for example, and looking to ride moves of (say) 50-100 pips, then M1 OHLC (let alone tick data) is easily accurate enough for my purpose.

My view is that 'randomness' in price movement makes trading a game of approximations. Probabilities can not be precisely calculated (as they can with casino games, for example), and I believe that some traders spend way too much time trying to optimize their systems to the n-th degree. I realize that everybody's approach is different, hence...
I agree with you on that -

the reason I ask is because regardless of Timeframe, when watching the market in fast forward on a simulator - its very easy to see when the market is about to make a push in one direction or the other...

If the tik data supplied on the simulator is acurate then a very profitable system could be developed out of this "momentum riding" however my guess is that the tick data on a simulator couldn't posibly be acurate for the reasons I mentioned in the first post and therefore it is pointless.

I'm guessing that the simulator when producing these very predictable momentous moves is simply filling its M1 volume (tic) "quota" so to speak

but I could be wrong.
Time hides Nothing
  • #5
  • Jul 27, 2009 3:51am Jul 27, 2009 3:51am
Quoting hanover
IMO it also depends on trading TF. If I'm trading H1 candle patterns, for example, and looking to ride moves of (say) 50-100 pips, then M1 OHLC (let alone tick data) is easily accurate enough for my purpose.
Indeed this is true, one has to be careful of the context as the accuracy depends on many things, in the case you mention I would feel better about testing this on open or close only data as your exits may be overstated if they are based on generated ticks, sometimes it is the smallest things (transaction costs for example) which can make the difference between winning and losing. One seeks robust edges, these usually don't depend on modeling ticks. I avoid all of this by only forward testing on real time market data, I concern myself with the modeling of fill statistics in order to get as realistic a simulation as possible. But in essence I agree, there is no one size fits all.
The breaking of a wave cannot explain the whole sea.
  • #6
  • Jul 27, 2009 4:35am Jul 27, 2009 4:35am
1. Create and EA
2. run it on live account/forward test from a defined start time
3. After you completed for the day, download the ticks data M1, M5,etc..
4. Back testing the EA.
5. Compare your entry and exits

I had done this and it shows 90% accurate time of entry and exits
(quite comparable between back test and forward test)

So I think it depends on how your EA coded and your system is.

Back testing is still much relevant to me.
  • #7
  • Jul 27, 2009 5:03am Jul 27, 2009 5:03am
Quoting akukaya
1. Create and EA
2. run it on live account/forward test from a defined start time
3. After you completed for the day, download the ticks data M1, M5,etc..
4. Back testing the EA.
5. Compare your entry and exits

I had done this and it shows 90% accurate time of entry and exits
(quite comparable between back test and forward test)
Makes Sense to me. Thanks for Advise ! Can check EA's this way !
Regards
100 Fold Challenge->Interested? ->https://www.forexfactory.com/thread/32152
  • #8
  • Jul 27, 2009 6:02am Jul 27, 2009 6:02am
Quoting hanover
IMO it also depends on trading TF. If I'm trading H1 candle patterns, for example, and looking to ride moves of (say) 50-100 pips, then M1 OHLC (let alone tick data) is easily accurate enough for my purpose.

My view is that 'randomness' in price movement makes trading a game of approximations. Probabilities can not be precisely calculated (as they can with casino games, for example), and I believe that some traders spend way too much time trying to optimize their systems to the n-th degree. I realize that everybody's approach is different,...
great words imo
  • #9
  • Jul 30, 2009 8:12pm Jul 30, 2009 8:12pm
Quoting Craig
I concern myself with the modeling of fill statistics in order to get as realistic a simulation as possible. But in essence I agree, there is no one size fits all.
I suppose you wrote your own simulation engine. My experience is that modelling fills is seriously difficult to get right as it depends on many factors and the specific exchange/ECN model.
  • #10
  • Jul 30, 2009 9:30pm Jul 30, 2009 9:30pm
Quoting BlackMage
I suppose you wrote your own simulation engine. My experience is that modelling fills is seriously difficult to get right as it depends on many factors and the specific exchange/ECN model.
I keep the model fairly simple & dynamic, I'd rather have simulated fills than simulated ticks.
The breaking of a wave cannot explain the whole sea.
  • #11
  • Last Post: Aug 8, 2009 5:21pm Aug 8, 2009 5:21pm
Quoting Craig
I keep the model fairly simple & dynamic, I'd rather have simulated fills than simulated ticks.
Watching the tick move the bar, especially in lower TF's, during live sessions is an invaluable indicator, even if no one has yet programmed it into a simulator or otherwise .
The timing of a tick with the price movement on the chart and, optionally, a volume indicator, shows when the price-time line reduces or increases speed (speed=distance*time).
Thus, a tick's timing may not only slow down the price into a near sideways line but can also indicate early on, with a significant exception of price(/volume) change with a tick, a later jump or fall in the price.
Going for relative or average price like the above also helps reduce the effect of price differences among brokers.
0 traders viewing now