MS Exploration Differ from Tradesim Results

[jacche]

Hi All,
I completed tradesim testing on my system and proceeded to turn it into a normal exploration for daily use by copying only the entry trigger code into the MS "Filter" column in Explorer. To test it I ran the exploration on a previous/specific date (29 and 30.12.03) and subsequently ran Tradesim (period: 1.1.01 thru to 31.12.03) expecting to see the same stocks being "picked up" in my simulation but it was not.

Reading other postings in this forum I have checked:
1. In Explorer Options: I load 2500 records (test period: 2001 to 2003)
2. Unchecked the "Ignore Open Trades" option in Tradesim to view any "missing" trades
3. Manually checked the code (for Tradesim run and normal exploration) for any differences
4. have also confirmed that my simulation was not currently "in-trade" for these "missing" stocks (i am not pyramiding). Of the 5 stocks triggered by the MS exploration, Tradesim Trade Database only picked up one stock.

I am very puzzled about the discrepancy even though the code is identical. Can anyone offer any suggestion perhaps if they have come across this problem or a hint of areas that might cause this issue ?

many thanks in advance
jack

[David Samborsky]

Results will change as new security data is appended to existing data. Even when only one bar of data is added to all securities the results can sometimes vary in a dramatic way. Why is this so?

Simply because as new data is added data scanning proceeds starting at the one bar after yesterdays starting bar and this can change the whole scenario as you may previously have had an entry trigger at the start of the data series (which would have been recorded in the Trade database) but now todays starting bar does not exhibit a trigger.

[jacche]

David, many thanks for the explanation and quick turnaround.

Also noticed that for one of the trades it was due to insufficient capital so that fixed one of the problems.

best regards
jack

[garin]

Hi All,

I am also tradesim testing on my system and proceeded to turn it into a normal exploration for daily use by copying only the entry trigger code into the MS "Filter" column in Explorer.
I ran Tradesim (period: 1.1.96 thru to present) and then checked the trade database.
I then found an trade entry and then ran the Metastock exploration for the day prior to the entry expecting to see the same stock being "picked up" in exploration but it was not. (I have the DelayAllEntriesByOneBar on)

Both simulations are being done on identical share data.

Am I missing something ? Would including the code be helpful ?
(This is my 1st system test with Tradesim)


Garin

[sectorbets]

I've got to say that other than getting comfortable with a new piece of software, it makes no sense to develop code in TradeSim and then use plain vanilla exploraions for daily use.

I have been using TradeSim explorations exclusively for over 18 months and rather than checking against MetaStock, I check every simulated trade against my actual trading results with real money, and they are extremely close--in fact my actual results are a little better than the simulated results since the actual slippage is a littel less than what I use in the simulation.

[garin]

sectorbets,
Are you talking about the Metastock exploration that generates the trade database ?
If so how do you use this to signal what share you should be buying on the day after the trigger day (assuming you are only trading end of day) ? as the result window only shows all the trades for the data you loaded upto the present.
The problem I have presently is that Tradesim is telling me that the system I have created is profitable.
But trying to simulate real trading manually ie I would get home update the days data, run the Metastock exploration, view the results and make my buys. But trying this on historical data led me to find that the triggers I got from the Metastock exploration (without the tradesim output) were different than the Tradesim trade database for the same day. So this means the simulation proved nothing since I can't see the buy signal on the day.

Oddly though I did try placing the trigger code into metastock's system tester and tried it on a stock that showed no trigger using the plain exploration - and presto the system tester showed the entry - odd

I think all this maybe how Metastock handles the TroughBars and PeakBars function I'm using, and somehow they are different when used in a filter executed for the most recent day - I did read somewhere that the zigzag function (that trough and peak are based on) does have a quirk when applied to the last bar in your data - but I'm not sure.

Thanks for the reply anyway

[sectorbets]

Hi garin--

I use real time signals to trade during the day using TradeStation. I develop and test my trading systems in MetaStock because it is easier, but when I go live, I convert everything to TradeStation, so my environment is different from yours. I haven't looked at a "pure" MetaStock exploration since a week or two after I got TradeSim--at best it is redundant, and at worst, just plain useless.

As to trading signals using your approach, it would seem that if you ran your system every night, TradeSim will tell you exactly what to trade the next day, including position sizing if you use those features--you never have to use MetaStock without TradeSim again.

If you are building a system around the zigzag funtion, it is REALLY dangerous since you can't tell whether you are at a peak or valley until AFTER the fact, so any simulations you make will likely be much more profitable than you could ever achieve in real life.

As to your specific problem (other than using zigzag), I'd suggest you post your code ovcer in the MetaStock section of the Forum, and you will get the proper attention.

Good luck.

[garin]

Thanks for that reply.
Point taken about the use of the zigzag function, I believe this is where the difference between the Tradesim database and the Metastock exploration occurs. And as you stated the simulation appeared much more profitable than expected.
Ii have modified the entry and now the two seem to match and the profitability seems much more realistic, and more importantly manual testing confirms the Metastock simulation - somethng that makes me more comfortable this being my first system test using Tradesim.

Thanks again