I’m trying to develop a strategy on the Dax, TF15min. I have some encouraging results, but wanted to improve the position management thank to the MTF. My idea was simply to take position on a signal given on the TF15min, and exit the position on the UT5min.
When I backtest the strategy only on the TF15min , I get in two years several hundred of trades. Logically, adding a code snippet for managing existing position on the TF5min shouldn’t impact the number of trades, or marginally. However, I get down to only 5 trades on the same period. Can someone confirm they have the same behaviour, or do I do something wrong?
Here below a very simple strategy that aims to explain my problem. If you run this code on the TF15min, you get hundred of trades, If you run the same one on the TF5min but with the code in comment uncomment, you get far less trades…
First, try to GRAPH your buylevel and selllevel with dummy trading conditions that don’t trigger any order.
Do it with and without “timeframe (15 minutes, updateonclose)”, in order to see if the levels are calculated the same way and to incriminate the TIMEFRAME instruction or not.
Ok, I have been able to isolate the code line which is causing trouble. Nothing to do with Demo or Live account, sorry my mistake. The code I provided here above was an example, which was very different from most of the original code with which I have some trouble. So the code here above works fine, the code here below doesn’t work because in the conditions to take a position, I added “barindex-tradeindex(1)>=3”, which is a pretty standard code snippet. Any idea why this condition doesn’t work with the MTF?
BarIndex and TradeIndex are determined by the lowest TF (which is the MAIN or Default one). So in your case it’s the index to bars in the 5-minute TF, no matter where you place that line!
If you need to count bars elapsed in the 15-minute TF you should add your own variables to mimick BarIndex and TradeIndex or replace 3 with 9 (three 5-minute bars are required for each 15-minute bar).
Thx Robertogozzi for the info, didn’t know. However, multiplying by 3 the factor doesn’t change the issue. Storing the barindex of the trade seems to work better (via the onmarket not onmarket snippet).
So this code seems now to work fine. However, the fact that tradeindex function isn’t working smoothly with MTF remains a bug for me, or do I miss something?
Ok but if it works smoothly, can you rework a code using tradexindex and which would work under MTF ? Because I can’t, and the workaround with storing the barindex of the trade is actually not working exactly as the tradeindex function.
it would be 12 in a 1-hour TF or 60 in a 1-hour TF with a 1-minute DEFAULT TF, and so on…
MTF requires that you change the way you think of TF’s. The main one, the one setting the pace, is the LOWEST which guarantees that ANY TF used will at least be executed once when its own bar closes and that’s why all TF’s must be multiple of the default (or main) one.
If, in your case, the 15-minute TF would set the pace, how could ProOrder guarantee that the 1-minute code is executed every minute? Instead, letting the 1-minute TF set the pace, your 15-minute code will be executed every minute and it’s granted it will be executed when the 15-minute bar closes!
This has some drawbacks:
you have less history data
you have to do some math to count elapsed bars
But advantages are far more!
MTF is just a new feature, which you can take advantage of or not. You may still write strategies with a single TF because MTF may not be beneficial to them, while other strategies you write do improve when using MTF.
But if tou want to use MTF you must write your code in accordance with it.
.
.
